Sea Surface Salinity from Space: A Canadian Perspective Sea Surface Salinity from Space: A Canadian Perspective A Report Prepared by B. J. Topliss1, J.F.R. Gower2, J.A. Helbig3, A. W. Isenor1, and I. Rubinstein4 for The Canadian Space Agency Earth and Environmental Applications Program March 2002 1 Bedford Institute of Oceanography, Fisheries and Oceans Canada Institute of Ocean Sciences, Fisheries and Oceans Canada 3 Northwest Atlantic Fisheries Centre, Fisheries and Oceans Canada 4 York University/CRESTech 2 ii _______________________ Four reports were commissioned in 2001 on various aspects of sensing sea surface salinity from space (Borstad and Horniak, 2001; Rubinstein 2001; Seibert, 2001; and Simms, 2001). A contract (DFO number F6093-1-JAH01) was issued to E.L. Simms and A. Simms to combine these reports into a draft of the present one. iii Executive Summary Under its Earth and Environment Applications Program, the Canadian Space Agency has funded the authors of this report to study developments in space-based remote sensing of sea surface salinity (SSS). The broad project goals are to define Canadian interests relative to various proposals for salinity satellites, to identify potential Canadian benefits, and to promote Canadian involvement in all aspects of SSS monitoring. Particular objectives are to provide the background necessary for Canadian scientists and industry to anticipate and thus plan participation in the remote sensing of salinity and to identify partnering and Canadian industry opportunities for the Canadian Space Program. The purpose of this first report is to review recent developments in the remote sensing of salinity and to place these developments in a Canadian context. In doing so, a number of issues are addressed, including: • What significance does the routine remote sensing of SSS hold for Canadian science and resource management? • What special issues must be addressed for SSS remote sensing in Canadian waters, which are generally cold, are fed by low salinity estuarine runoff, and are sometimes ice covered? • What can Canada contribute to the remote sensing/physical-biological oceanographic community? Ocean temperature and salinity have been routinely measured since the earliest oceanographic expeditions, and they comprise the two most basic physical variables that define the state of the ocean. For the past 30 years, sea surface temperature (SST) has been monitored operationally from space, but the concomitant capability to measure sea surface salinity has not existed. This is a critical shortcoming, as there are no salinity data for about 40% of the 1° latitude by 1° longitude areas that make up the earth's ocean surface. The capability to measure SSS from space now exists. In fact, proposals for a passive microwave satellite are currently before both European Space Agency (ESA) and National Aeronautics and Space Administration (NASA) with the goal of monitoring SSS and/or its terrestrial counterpart, soil moisture (SM). These proposals all aim for a 2006 launch date. From an oceanographic viewpoint, salinity is important because: • Together with temperature, it determines water density and thus is intimately linked to the global ocean circulation. • It determines the depth to which water, cooled at the surface in winter, can sink. That is, it is an important component of the thermohaline circulation and is thus directly linked to the dynamics of the earth’s climate. • Through its effect on density, it partially controls the stability of the upper mixed layer of the ocean. This in turn has important physical and ecological consequences. • It is one of the prime determinants of the environment in which fish and other marine life live. • It modulates air-sea interaction including gas and heat exchange. iv In terms of applications, sea surface salinity has significance to a large number of areas including: • climate dynamics and prediction, • global scale ocean-atmosphere modelling, • regional scale ocean-atmosphere modelling, • fisheries management, and • environmental monitoring. Large-area, long-term surface salinity anomalies, having significant effects on ocean circulation, climate and fisheries, have been observed in the past in the northern Atlantic. Tracking of the 1965 to 1980 "Great Salinity Anomaly (GSA)" relied extensively on observations from ocean weather ships, which have since been discontinued. Similar patterns may well be undetected today, both in the North Atlantic, and in other ocean basins, which have always been less well monitored. The proposed satellite missions provide a good match to the time and space scales of the GSA. Spaceborne SSS mapping should provide significant benefits to Canada. From a scientific vantage point, it should provide important information for climate studies, both in terms of data for assimilation into numerical models as well providing observations in regions where historical data coverage is extremely poor. Similar considerations also apply to regional modelling activities. As well, benefits will accrue in the areas of fisheries management and environmental monitoring as workers gain access to data on a longterm, year-round basis never before possible. This will permit, for example, better estimates of mixed layer depths that are important in ecological models. Similarly, workers will be able to better quantify the fresh water distribution due to estuarine discharge. Sea surface salinity signals in Canadian waters are large, both spatially in terms of horizontal gradients and temporally on seasonal and interannual time scales. Consequently, Canada should be able to usefully exploit remotely sensed SSS. The accuracy of the proposed satellite sensors are about 0.4 practical salinity units (psu) for a single pixel, single pass observation and about 0.1 psu for data averaged over space and time into standard products (200 km and 10 days). Salinity varies by at least 1 psu across Canada’s Pacific and Atlantic continental shelves, a value at least twice as large as the largest expected errors. Similarly, the amplitude of the seasonal cycle exceeds 1 psu in many areas. Year-to-year variations are generally smaller, but still often exceed 0.5 psu. The remote detection of salinity exploits the relatively weak dependence of the emissivity of seawater on salinity at microwave (L-band) frequencies. At these frequencies the atmosphere is essentially transparent, so that atmospheric contamination of the signal is minimal. However, significant error can be introduced by reflected galactic and solar radiation as well as by changes in apparent emissivity due to sea surface roughness and foam. The sea surface temperature must also be known, because the emissivity is temperature dependent. v There exist two problems in remote sensing of SSS that are of special significance to Canada, namely cold water and ice. The emissivity varies less with salinity in cold water as compared to warm water, thus reducing sensor sensitivity. Similarly, since ice has a different microwave signature from seawater, partially ice covered pixels will present a data processing challenge. On the other hand, the Canadian science community has considerable experience with these issues and is especially well suited to address them. vi Table of Contents EXECUTIVE SUMMARY ........................................................................................................................ IV LIST OF FIGURES.................................................................................................................................... IX LIST OF TABLES.....................................................................................................................................XII 1.0 INTRODUCTION............................................................................................................................... 1 1.1 2.0 WHY SEA SURFACE SALINITY?......................................................................................................... 2 REMOTE SENSING OF SEA SURFACE SALINITY ................................................................... 5 2.1 MICROWAVE EMISSION FROM THE OCEAN........................................................................................ 5 2.2 PASSIVE MICROWAVE RADIOMETRY ................................................................................................ 8 2.2.1 Environmental Factors Affecting TB ..................................................................................... 10 2.2.1.1 2.2.1.2 2.2.1.3 Contributions from Space and Sun.................................................................................................. 10 Atmosphere Interactions.................................................................................................................. 11 Effect of Sea State: Waves, Breaking Waves and Foam ................................................................. 13 2.3 REVIEW OF SPACEBORNE AND AIRBORNE EXPERIMENTS................................................................ 15 2.3.1 Remote Sensing of Sea Surface Salinity in Canada .............................................................. 15 2.3.2 Spaceborne Experiments ...................................................................................................... 16 2.3.3 Airborne Experiments........................................................................................................... 17 2.4 PROPOSED SEA SURFACE SALINITY SYSTEMS ................................................................................. 19 3.0 SALINITY DATABASES ................................................................................................................ 22 3.1 GLOBAL DATABASES ...................................................................................................................... 22 3.1.1 USA National Oceanographic Data Center ......................................................................... 22 3.1.1.1 3.1.1.2 3.1.1.3 3.1.1.4 3.1.1.5 Online Oceanographic Profile Database.......................................................................................... 23 World Ocean Database 1998 Version 2 (WOD98).......................................................................... 23 World Ocean Atlas 1998 (WOA98) ............................................................................................... 23 World Ocean Atlas 1998 (WOA98F) ............................................................................................. 25 World Ocean Atlas 1994 (LEVITUS94) ......................................................................................... 27 3.1.2 Marine Environmental Data Service (MEDS) ...................................................................... 31 3.2 CANADIAN REGIONAL DATABASES ................................................................................................ 32 3.2.1 Atlantic Region ..................................................................................................................... 32 3.2.2 Pacific Region ...................................................................................................................... 35 4.0 SEA SURFACE SALINITY IN CANADIAN WATERS............................................................... 37 4.1 SPATIAL SEA SURFACE SALINITY DISTRIBUTIONS .......................................................................... 37 4.1.1 Northwest Atlantic ................................................................................................................ 37 4.1.2 Northeast Pacific .................................................................................................................. 41 4.2 TEMPORAL DISTRIBUTION OF SEA SURFACE SALINITY ................................................................... 43 4.2.1 Northwest Atlantic ................................................................................................................ 43 4.2.2 Northeast Pacific .................................................................................................................. 49 4.2.2.1 4.2.2.2 Lighthouse Stations ......................................................................................................................... 49 Station Papa..................................................................................................................................... 52 5.0 SCIENCE REQUIREMENTS ......................................................................................................... 56 6.0 ONGOING APPLICATIONS OF SEA SURFACE SALINITY................................................... 57 6.1 FISHERIES RESOURCES IN CANADIAN WATERS ............................................................................... 57 6.1.1 West Coast ............................................................................................................................ 57 6.2 OPTICAL WATER PROPERTIES ......................................................................................................... 57 7.0 EFFECTS OF SEA ICE ON THE REMOTE SENSING OF SSS................................................ 61 vii 7.1 L-BAND RADIOMETRIC CHARACTERISTICS OF SEA ICE ................................................................ 62 7.1.1 Brine ..................................................................................................................................... 62 7.1.2 First Year and Multiyear Ice ................................................................................................ 62 7.1.3 Melt Ponds on Consolidated Ice........................................................................................... 65 7.2 EFFECTS OF ICE IN THE FIELD OF VIEW ........................................................................................... 65 8.0 CONCLUDING REMARKS ........................................................................................................... 67 9.0 REFERENCES.................................................................................................................................. 69 9.1 9.2 9.3 CITED REFERENCES ........................................................................................................................ 69 RELEVANT WEBSITES ..................................................................................................................... 72 CORRESPONDENCE .......................................................................................................................... 74 viii List of Figures Figure 1. The terms ∂ρ/∂T and ∂ρ/∂S as functions of T and S. The solid lines are lines of constant density. The dashed lines are contour lines of ∂ρ/∂T or ∂ρ/∂S. ................... 4 Figure 2. Horizontal profiles of surface temperature, surface salinity, surface density, and the changes in density due to changes in T and S between stations along the Bonavista transect in November 2001. The variable σt ≈(ρ - 1015)/1000. Figure 1 indicates that density is almost independent of temperature when the temperature is low............................................................................................................................... 4 Figure 3. Dielectric loss factor or sea water (S=32.54 psu) and pure water as a function of the frequency, for T=0°C and 20°C (Ulaby et al., 1986). ...................................... 7 Figure 4. Spectral variation of the salinity sensitivity at T=293K or 20°C (Ulaby et al., 1986). .......................................................................................................................... 7 Figure 5. The variation of brightness temperature with salinity for a number of temperatures computed with the Klein and Swift (1977) model. ............................... 8 Figure 6. Attenuation coefficients of water clouds as a function of temperature (Ulaby et al., 1986). .................................................................................................................. 12 Figure 7. Attenuation coefficient of ice clouds as a function of temperature (Ulaby et al., 1986). ........................................................................................................................ 12 Figure 8. Brightness temperature as a function of surface wind speed, for surface salinity of 36 psu and surface temperature of 5°C. x indicates HH polarisation, while o indicates VV. Parameters from Hollinger and Lo (1981). ..................................... 14 Figure 9. Top of atmosphere brightness temperature as a function of surface salinity for wind velocity of 10m s-1 and surface temperature of 5°C, at 1.4 GHz. x indicates HH polarisation, while o indicates VV. Goodberlet-Swift model . ......................... 14 Figure 10. Vertical salinity profile for a station off the Queen Charlotte Islands on July 1, 1997........................................................................................................................... 24 Figure 11. Global annual mean surface salinity (PSS), from the WOA98F database ..... 24 Figure 12. Winter mean surface salinity (PSS), from WOA98F. .................................... 26 Figure 13. Annual mean surface salinity (PSS) in the Pacific Ocean, from WOA98F. .. 26 Figure 14. Global long-term mean surface salinity (psu) from LEVITUS94.................. 29 Figure 15. January to March global mean surface salinity (psu), from LEVITUS94. .... 29 Figure 16. Annual mean surface salinity (psu) in the Pacific Ocean, from LEVITUS94. ................................................................................................................................... 30 Figure 17. Annual mean surface salinity (psu) in the Gulf of Alaska from LEVITUS94. ................................................................................................................................... 30 Figure 18. Location of temperature-salinity profiles acquired by MEDS over the global ocean, for 1998. The colour scale indicates the number of observations. ............... 32 Figure 19. Samples distribution and count of surface salinity data archived in the Computer Atlas of the NW Atlantic. ........................................................................ 33 Figure 20. Distribution and count of samples collected in May of all years between 1900 and 1997.................................................................................................................... 34 Figure 21. Map of the British Columbia coast showing the location of the lighthouse stations. ..................................................................................................................... 35 Figure 22. General geographic location of the stations along the P-Line........................ 36 ix Figure 23. Annual cycle of surface salinity from 1956 to the present ............................. 36 Figure 24. Average salinity based on all years from the Computer Atlas of the NW Atlantic...................................................................................................................... 38 Figure 25. The horizontal distribution of surface salinity over the Grand Banks of Newfoundland and the Northeast Newfoundland Shelf during the period of AugustNovember 2000. (Figure courtesy of E.B. Colbourne, Northwest Atlantic Fisheries Centre, DFO)............................................................................................................. 39 Figure 26. The horizontal distribution of surface salinity anomalies over the Grand Banks of Newfoundland and the Northeast Newfoundland Shelf during the period of August-November 2000. (Figure courtesy of E.B. Colbourne, Northwest Atlantic Fisheries Centre, DFO). ............................................................................................ 40 Figure 27. Strait of Georgia and Gulf Islands; Salinity and chlorophyll data collected from a ferry on March 23, 2001................................................................................ 42 Figure 28. Salinity horizontal profile from Research Vessel ‘Tully’, February 2001..... 42 Figure 29. Horizontal profiles on the North West coast of Vancouver Island, in 1979. . 43 Figure 30. Stations where the seasonal cycle in SSS was computed............................... 45 Figure 31. Upper panel: The seasonal cycle in surface salinity at Station 27 (see Figure 25 for location). The vertical bars denote the variability around the long term mean. Lower panel: August anomalies in surface salinity at Station 27. The anomalies are computed relative to the 1971 to 2000 mean. ........................................................... 46 Figure 32. Surface salinity at Ocean Weather Station Bravo in the North Atlantic. ....... 48 Figure 33. The path of the Great Salinity Anomaly (from Belkin, et al., 1998).............. 48 Figure 34. Strait of Georgia; Annual cycles of monthly mean salinity. .......................... 50 Figure 35. Amphitrite Point; Annual cycle of monthly mean salinity............................. 50 Figure 36. Langara Island; Annual cycle of monthly mean salinity................................. 50 Figure 37. Amphitrite Point; Monthly average surface salinity anomalies. .................... 51 Figure 38. Langara Island; Monthly average surface salinity anomalies. ....................... 51 Figure 39. Surface salinities in the vicinity of Station Papa from 1956 to 1996 ............. 53 Figure 40. Annual cycle of surface salinity at Station Papa from data in Figure 39. ...... 53 Figure 41. Residual anomalies in surface salinity data at Station Papa, from 1956 to 1996........................................................................................................................... 54 Figure 42. Monthly surface salinity anomalies at Station Papa, from 1980 to 2000....... 54 Figure 43. The Southern Oscillation Index plotted for 1979-1998.................................. 55 Figure 44. Detrended monthly surface salinity anomaly at Papa station plotted with the Southern Oscillation Index. ...................................................................................... 55 Figure 45. Relationship between dissolved organic material concentration (OD350) and surface salinity in Barkley Sound, April 1987.......................................................... 59 Figure 46. Correlation between Gelbstoff (OD400) and sea surface salinity in Atchafalaya River, March 2001................................................................................ 59 Figure 47. Sea surface salinity distribution maps estimated from SeaWiFS water colour acquired on February 24, 1999. ................................................................................ 60 Figure 48. Brine salinity in sea ice as a function of negative temperature (Ulaby et al., 1986) ......................................................................................................................... 63 Figure 49. Dielectric constant of liquid brine as a function of frequency (Ulaby et al., 1986). ........................................................................................................................ 63 x Figure 50. Calculated penetration depth in pure ice and in first year and multi-year sea ice (Ulaby et al., 1986). ............................................................................................ 64 Figure 51. Brightness Temperature evolution from open water to 15 cm thick young ice. Nadir angle=50°. Just before the 20th hour the high brightness values are reached at a time when the ice became optically thick; with the L-band the ice was then 4 to 5 cm thick (Grenfell et al., 1998)....................................................................................... 65 Figure 52. Brightness temperatures at L-band as a function of ice thickness (Ulaby et al., 1986, sea ice parameters used in simulation; CRESTech radiometer simulation model). ...................................................................................................................... 66 Figure 53. Simulation of L-band brightness temperature for an ice-covered ocean. The assumption is made that the ocean fraction could have variable salinity. ................ 66 xi List of Tables Table 1. Typical brightness temperature contribution from space and atmosphere, for nadir observations (Blume et al., 1978).................................................................... 11 Table 2. Airborne microwave radiometers and some of their characteristics.................. 17 Table 3. Airborne deployments that used passive L- and S-band sensors....................... 18 Table 4. Summary of activities and publications related to space borne passive microwave remote sensing missions for sea surface salinity mapping..................... 20 Table 5. Technical and performance specifications of passive radiometry satellite missions currently in progress. Sources: HYDROS - HYDROS web site; AQUARIUS - Koblinsky et al. (2001); SMOS - Kerr (1998). ................................. 21 Table 6. Search results for oceanographic profiles in the NE Pacific, 1980 to 2000. ..... 27 Table 7. Example of annual mean salinity data (psu) from LEVITUS94 ....................... 28 Table 8. Sea Surface Salinity Data Statistics ................................................................... 43 Table 9. Time series statistics for selected stations of the Canadian eastern seaboard. .. 44 Table 10. Time Series Statistics for three British Columbia lighthouse stations and Station Papa. ............................................................................................................. 49 Table 11. Summary of scientific requirement for the study of surface circulation patterns. ................................................................................................................................... 56 xii 1.0 Introduction Ocean temperature and salinity have been routinely measured since the earliest oceanographic expeditions of the 19th century, and they comprise the two most basic physical variables that define the state of the ocean. For the past 30 years, sea surface temperature (SST) has been monitored operationally from space, but the concomitant capability to measure sea surface salinity (SSS) has not existed. This is a critical shortcoming. Indeed, SSS (as well as subsurface salinity) has never been observed for 42% of the world ocean1 and has been observed fewer than four times over the past 125 years for 88% (Levitus et al.,1998). On the other hand, the potential for measuring SSS with passive L-band microwave radiometers is well known and was proved in the 1970s and 80s in airborne and spaceborne (Skylab) experiments. Now, proposals for a passive microwave satellite are before both ESA and NASA with the goal of monitoring SSS and/or its terrestrial counterpart, soil moisture (SM). These proposals all aim for a 2006 launch date. The Canadian Space Agency Earth and Environment Applications Program has funded the authors of this report to study developments in space-based remote sensing of sea surface salinity. The broad project goals are to define Canadian interests relative to various proposals for salinity satellites and to identify and promote the widest range of Canadian benefits and involvement in measuring ocean salinity from space. Particular objectives are to provide the necessary scientific background to enable Canadians to participate in the planning of salinity sensor systems and to identify partnering and Canadian industry opportunities for the Canadian Space Program. The purpose of this first report is to review recent developments in the remote sensing of salinity and to place these developments in a Canadian context. In doing so, a number of issues are addressed, including: • What significance does the routine remote sensing of SSS hold for Canadian science and resource management? • What special issues must be addressed for SSS remote sensing in Canadian waters, which are generally cold, are fed by low salinity estuarine runoff, and are sometimes ice covered? • What can Canada contribute to the remote sensing/physical-biological oceanographic community? For the purposes of this document, it is useful to partition the ocean into two zones, the coastal and the offshore. In the Canadian Atlantic zone, the coastal zone is conveniently defined to extend from the coast to the continental slope, a distance ranging from about 100-400 km. In the Pacific, where the continental shelf is narrow, we define the coastal zone to be that region within which the effects of large rivers (e.g. the Fraser and Columbia) runoff are significant, i.e. about 200 km from the coast. Almost all of the 1 Based on surface area and a 1° latitude by 1° longitude observational grid. 1 documentation supporting the salinity satellite missions have focused on the offshore regions and deep-sea oceanographic processes including climate. 1.1 Why Sea Surface Salinity? From an oceanographic viewpoint, salinity is important because: • Together with temperature, it determines water density and thus is intimately linked to the horizontal circulation of the ocean. • It determines the depth to which water, cooled at the surface in winter, can sink. That is, it is an important component of the thermohaline circulation and is thus directly linked to the dynamics of the earth’s climate. • Through its effect on density, it partially controls the stability of the upper mixed layer of the ocean. This in turn has important physical and ecological consequences. • It is one of the prime determinants of the environment in which fish and other marine life live. • It modulates air-sea interaction including gas and heat exchange. We now consider these in more detail. The density of seawater is important for at least three reasons. First, horizontal differences in density generate ocean currents. Second, the vertical distribution of density determines the vertical stability of the ocean and governs the degree to which forcing at the sea surface by winds and heating and cooling is felt by subsurface waters. Third, the vertical density distribution also determines the depth over which convection due to cooling can take place. The density of seawater depends on its temperature and salinity through the equation of state. Within the ocean’s offshore temperate zone (approximately equatorward of 50°), temperature is by far the most important variable. However, in coastal waters that experience large fresh water discharge or in waters at low temperatures, salinity is significant and can, in fact, dominate temperature effects. Such conditions are typical of Canadian waters and have received relatively little discussion in the documents supporting SSS satellite missions. To illustrate the influence of salinity on density, consider Figure 1 which displays the dependence of density (ρ) on temperature (T) and salinity (S) as a function of T and S. At low temperatures, ρ is almost T-independent; by contrast ∂ρ/∂S shows little temperature dependence. Figure 2 further illustrates the importance of salinity with an example from the Canadian Atlantic zone. The three upper panels show horizontal profiles of surface temperature, salinity, and density [through the variable σt ≈ (ρ 1015)/1000] measured in November 2001 along the Bonavista transect (see Figure 25 for location). The lowest panel shows the change in density between stations due to the respective changes in temperature and salinity. It is apparent that salinity plays an important role in determining the horizontal density gradient and thus in forcing the horizontal circulation over the shelf. 2 Through its effect on density, surface salinity also indirectly determines the depth of penetrative convection in high latitudes. In the autumn and winter when surface waters are cooled to near-freezing temperatures, the salinity governs the ultimate density and thus determines the depth to which it will sink. As such, SSS plays a critical role in the formation of the intermediate and deep ocean water masses that feed the thermohaline conveyer belt; SSS is thus directly linked to the climatic cycle. In the sub-polar North Atlantic, interannual variations in SSS modulate the thermohaline circulation, and intrusions of low salinity water can prevent the formation of deep water (see the discussion of the Great Salinity Anomaly in Section 4.2.1). Salinity is also one of the prime determinants of the ecosystem within which fish and other marine life live, and it is recognized as a significant factor in fisheries management and stock prediction. In addition, year-to-year and annual variations in sea salinity are important variables in ocean productivity forecast models, especially as they affect the depth and stability of the surface mixed layer. The annual cycle of primary productivity in most Canadian waters is linked directly to the pumping of nutrients from deeper waters into the generally nutrient-poor mixed layer. If salinity is relatively low, the mixed layer will be more stable, and the nutrient pump may be partially inhibited, possibly leading to reduced productivity or a delay in the onset of spring and autumn phytoplankton blooms. Finally, salinity plays a potentially important role in the air-sea exchange of gases. In the tropical ocean, heavy precipitation can create pools of relatively fresh water that locally increase the stability of the upper layer and lead to significantly reduced rates of gas transfer across the pycnocline (i.e. the base of the mixed layer). Similarly, current atmospheric circulation models used in global and regional climate studies ignore the influence of salinity on the saturation vapour pressure of seawater. However, SSS may have a measurable influence on the surface specific humidity gradient, thus affecting computed evaporation rates. 3 T (d e g C ) Figure 1. The terms ∂ρ/∂T and ∂ρ/∂S as functions of T and S. The solid lines are lines of constant density. The dashed lines are contour lines of ∂ρ/∂T or ∂ρ/∂S. 8 6 4 2 0 50 100 150 200 250 300 350 400 450 200 250 300 350 400 450 S (p s u ) 36 34 32 30 σ t 26 25 0 .4 | (∂ ρ / ∂ S ) δ S | | (∂ ρ / ∂ T) δ T | 0 .2 0 0 50 100 150 d is ta n c e o ff s h o r e ( k m ) Figure 2. Horizontal profiles of surface temperature, surface salinity, surface density, and the changes in density due to changes in T and S between stations along the Bonavista transect in November 2001. The variable σt ≈(ρ - 1015)/1000. Figure 1 indicates that density is almost independent of temperature when the temperature is low. 4 2.0 Remote Sensing of Sea Surface Salinity This section presents the principles of remote sensing of sea surface salinity. In particular we include the space and airborne experiments in sea surface salinity retrieval. Information currently made available on designs and status of proposed sea surface salinity missions is also presented. 2.1 Microwave Emission from the Ocean All matter emits electromagnetic radiation as a result of the thermally induced random motion of atoms. An object that is a perfect absorber and emitter of electromagnetic radiation at all frequencies is defined as a blackbody. According to Planck's radiation law, the spectral brightness Bνbb (Wm-2sr-1Hz-1) of a radiating blackbody is: Bνbb =2hν3c-2[1/(ehν/kT-1)] where: h is Planck's constant (6.63 x 10 -34 J s ), ν is the frequency (Hz), k is Boltzmann's constant (1.38 x 10-23 J K-1), T is the absolute temperature (K), and c is the speed of light (3 x 10 8 m s-1). (1) For most purposes, the microwave region is defined as the interval from about 1 to 300 GHz (corresponding to wavelengths of 20 - 0.1 cm). For the ocean at microwave frequencies we have T > 270 K and so hν << kT. In this regime, Planck's law can be approximated by the Rayleigh-Jeans formula (Kraus, 1966): Bνbb =2ν2kc-2T (2) This approximation is useful because it allows one to express the signal observed by a radiometer in terms of the physical temperature of the observed scene. However, most natural objects are not perfect blackbodies. The ratio of the spectral brightness of an object to that of a blackbody at the same temperature defines the emissivity ε of that body: ε = Bν/ Bνbb (3) The emissivity is dependent on the surface dielectric properties, composition, object's shape and surface roughness. For "grey bodies", we may define a brightness temperature TB through Bν= ε Bνbb = 2ν2kc-2TB (4) 5 Note that TB is always smaller or equal to the object's physical temperature. Detection of the variations in brightness temperatures of bodies at the same physical temperature enables one to differentiate between types of material, such as fresh and salt water, or land and ocean surfaces. When considering seawater as the target, we note that the emissivity of seawater varies with salinity, temperature, and the surface texture (i.e., roughness and foam coverage). For a smooth surface, the ocean can be represented by a flat, infinite, half-space. In this approximation the emissivity is given by: εH,V = 1 – RH,V (5) Here the subscripts H and V refer to horizontally or vertically polarised radiation while RH,V are the Fresnel reflection coefficients: RH = {[cosθ i-(ε-sin2θi)1/2 ] / [cosθ i +(ε-sin2θ i ) 1/2 ]}2, and (6) RH = {[εcosθ i-(ε-sin2θ i)1/2 ] / [εcosθ i +(ε-sin2θ i)1/2 ]}2 , (7) where: ε is the complex dielectric constant (in this case of seawater), and θi is the angle of incidence. ε is given to sufficient accuracy by the Debye (1929) equation, ε = ε ∞+(εs ε∞)/(1+iωτ)-iσ/ωε0 = [ε ∞+(εs ε∞)/(1+ω2τ2)] –i [σ/ωε0 + (εs ε∞ ωτ)/(1+ω2τ2)] where: (8) ω= 2πν is the radian frequency, ε∞ is the dielectric constant at infinite frequency, εs is the static dielectric constant, τ is the relaxation time, σ is the ionic conductivity, and ε0 is the permitivity of free space. It is seen that the imaginary part of the complex dielectric constant contains a term directly proportional to the conductivity and hence is strongly dependent on the salinity. However, the relationship is actually more complex as εs, σ, and τ are all functions of the temperature and salinity. For given values of salinity and temperature, the emissivity of seawater can be calculated using the model of Klein and Swift (1977). Figure 3 demonstrates that the dielectric loss of seawater and pure water are nearly identical at frequencies above 5 GHz for 0°C surface temperature, and above 11 GHz for 40°C surface temperature. Figure 4 shows that the brightness temperature for vertically 6 polarised radiation is significantly more sensitive to salinity variations than for horizontally polarisation. Figure 5 displays the dependence of brightness temperature on salinity for a number of different temperatures and for two frequencies (L band at 1.43 GHz and S band at 2.65 GHz). The dependence is essentially linear and is clearly much greater at the lower frequency. From these considerations, Ulaby et al. (1986) concluded that salinity remote sensing is best done with vertically polarised radiation with frequencies less than 5 GHz. Figure 3. Dielectric loss factor or sea water (S=32.54 psu) and pure water as a function of the frequency, for T=0°C and 20°C (Ulaby et al., 1986). Figure 4. Spectral variation of the salinity sensitivity at T=293K or 20°C (Ulaby et al., 1986). 7 Figure 5. The variation of brightness temperature with salinity for a number of temperatures computed with the Klein and Swift (1977) model. 2.2 Passive Microwave Radiometry Instruments designed to measure thermally induced electromagnetic energy are called radiometers. A typical microwave radiometer consists of three units: 1) an antenna subsystem for receiving radiation, 2) an electronic subsystem for detecting and amplifying the received signal within a specific frequency band, and 8 3) a control and data processing subsystem to process and record radiometric data (Ulaby et al., 1986; Skou, 1989). The performance of microwave radiometers is measured in terms of their sensitivity at different wavelengths, gain stability, absolute calibration accuracy, angular resolution, beam efficiency, and stability of the instrument spin rate. Properties of a passive microwave radiometer include the angular resolution (9) and the surface resolution (10). Θr ≈λ/L (9) r = zΘr = λz/L where: (10) Θr is the angular resolution, λ is the wavelength, L is the antenna length, r is the surface resolution, and z is the altitude above sea level. A spaceborne radiometer pointing towards the ocean measures a radiation field that is made up of microwave energy emitted by the ocean, energy emitted by the atmosphere, and external energy reflected from the sea surface. To good approximation for a smooth sea, the radiative transfer equation describing the radiation field is given by TB(θ) = ε t(θ) Ts + sec(θ) ∫0h TA(z) α(z) t(z,h,θ) dz + (1- ε) t(θ) sec(θ) ∫0h TA (z) α(z) t(θ,0,z) dz + (1 - ε) t2(θ) Text where (11) θ is the viewing angle relative to the vertical; t(a,b,θ) = exp{- sec(θ) ∫ab α(s) ds} is the atmospheric transmittance from height a to height b; t(θ) = t(0,h,θ) is the total atmospheric transmittance from the sea surface to the satellite; α(z) is the atmospheric attenuation due to absorption and scattering; (1 - ε) is the reflectance from the sea surface; TB is the brightness temperature observed at the radiometer, TA(z) is the height dependent temperature of the atmosphere; Ts is the sea surface temperature; and Text is the external radiation from space. The respective terms in (11) represent radiation emitted from the ocean and attenuated by atmospheric absorption and scattering, radiation upwelled from the atmosphere, atmospheric radiation reflected from the sea surface, and external radiation reflected from the sea surface and attenuated on its downward and upward paths. 9 The external brightness temperature contains contributions from the isotropic cosmic background, galactic radiation, and the sun, Text = Tcos + Tgal + Tsun (12) In order to extract the sea surface brightness temperature, and hence to estimate the sea surface salinity, it is necessary to account for all the other contributions to TB. This is clearly a complex task and requires models of the various components. The validation of these simulation models is crucial because spaceborne sensors provide observations for areas and weather conditions that cannot be ground-truthed by airborne surveys. Existing spaceborne data should be used to extract information on variables such as precipitation intensity and wind velocity, then to account for their effect in sea surface salinity estimations. 2.2.1 Environmental Factors Affecting TB 2.2.1.1 Contributions from Space and Sun The Cosmic radiation term in (12) is isotropic and is represented by a constant brightness temperature contribution of 2.7 K (Table 1). Since it is constant, it does not affect measurement accuracy. The Galactic radiation term in (12) originates in our galaxy and is strongest in the direction of the Milky Way. Its magnitude varies from 0.8 - 16°K between the galactic pole and centre. An average value of 0.98°K is generally used to represent the galactic noise for budget estimations in the atmosphere brightness temperature calculation. Galactic noise contributions may be minimised by selecting an appropriate satellite orbit or by mapping galactic noise contributions and correcting radiometer measurements. The Sun glint term in (12) can dominate the radiation upwelling from the ocean surface and therefore must be avoided through appropriate orbit selection (e.g. morning sunsynchronous). However, when the sea surface is rough, sunlight will be scattered in all directions. This indicates the requirement for a near-dawn or dusk orbit. 10 2.2.1.2 Atmosphere Interactions Microwave radiation is attenuated by absorption and scattering as it propagates through the atmosphere, and its plane of polarisation is rotated as it interacts with the ionosphere. Absorption and scattering occur mainly in the lower atmosphere (troposphere) and are primarily due to oxygen, water vapour, and water droplets. The oxygen concentration is approximately constant in the atmosphere and the oxygen absorption coefficient is known, so it is relatively straightforward to correct for it. On the other hand, water vapour content is highly variable. Fortunately, absorption by water vapour at L-band frequencies is very small and is usually neglected. Correcting for scattering by water droplets is more difficult. Limited data is available for L-band observations, largely because airborne surveys are usually carried out under fair weather conditions. For L-band frequencies, Rayleigh theory can be used to compute the scattering by clouds and hydrometeors. Figure 6 and Figure 7 show attenuation coefficients for several values of cloud temperatures associated with water and ice clouds, respectively. Table 1. Typical brightness temperature contribution from space and atmosphere, for nadir observations (Blume et al., 1978). Source of contribution to the brightness temperature T2.65GHz (K) T1.43GHz (K) Cosmic Background, TCOS 2.7 2.7 Galactic Radiation, TGAL 0.2 1.0 Downward atmospheric temperature, TDN 2.2 2.1 Beam efficiency corrections 0.4 0.14 Wind speed corrections (surface roughness) 1.1 0 0.0091 0.008 5.2 4.2 - - Total atmospheric opacity Typical total correction Sun (not yet determined) The plane of polarization of a signal emitted from a target within a radiometer’s field of view will undergo Faraday rotation before reaching the radiometer. The degree of rotation depends on the locations of the emitting object and the radiometer, as well as on the orientation of the Earth's magnetic field. The magnitude of the rotation is estimated to be 8.7° at a frequency of 1.4 GHz (Hollinger and Lo, 1981; Skou, 2001). 11 Figure 6. Attenuation coefficients of water clouds as a function of temperature (Ulaby et al., 1986). Figure 7. Attenuation coefficient of ice clouds as a function of temperature (Ulaby et al., 1986). 12 2.2.1.3 Effect of Sea State: Waves, Breaking Waves and Foam The emissivity of a smooth ocean surface can be computed using (5) – (8) together with additional equations for εs, τ, and σ. For a surface roughened by the wind, one must use empirically derived relationships. In addition, for wind speeds higher than 7 m s-1, the emissivity of surface foam becomes significant. Figure 8 presents the modelled top of atmosphere brightness temperature at L-band, as a function of the ocean surface wind speed, based on parameters from Hollinger and Lo (1981). Figure 9 presents the Goodberlet-Swift simulation of the top of atmosphere brightness temperature as a function of surface salinity, for wind velocity of 10 m s-1 and surface temperature of 5°C, at 1.4 GHz. Simultaneous observations with an optical sensor, sensitive to white caps, foam patches, and streaks are required to correct for wind effects. An alternative solution is to use a dual frequency radiometer system that allows for the uncoupling of the wind effects. Current spaceborne sensors such as SSM/I, RADARSAT, and ERS-SAR can also be used to correct for wind effects. Stogryn and Cardone (in Ulaby et al., 1986) modelled the emissivity of the sea surface foam. Aircraft observations were used to derive equations that relate the fraction coverage of the sea surface with foam and ocean surface wind speed. However, these equations were not validated. The Virtual Centre for Ocean Satellite Salinity (VCOSS; whose executive members are the authors of this report) are developing a simulation model for L-band brightness temperature based on the radiative transfer codes developed by Hollinger (1971) and modified by Rubinstein (1996). The effects of weather conditions such as wind induced sea surface roughness and foam, as well as precipitation will be accounted for by some of the modules in the simulation model. The simulation model will be used to establish the uncertainty in sea surface salinity induced by the errors in the sea surface temperature input. In addition, sun-glint and galactic contribution to the TOA radiance will be calculated using sea surface reflectance properties. The simulation model will contain the following four modules: • • • In the scenario initialisation the user may select meteorological conditions and surface state. Variables and their range of variability are specified. For example, the user will have a choice of selecting the air temperature, the surface temperature, the wind speed, and the wind direction. Using wind speed information, the foam fraction is calculated. The specular reflectance subroutine will compute the dielectric constant for a given salinity and temperature, based on the Klein and Swift (1977) model. The specular reflectance module also calculates the Fresnel vertical and horizontal reflectivity coefficients. The surface roughness and foam module will estimate the roughness effects using a two-scale model, and the contribution of emission from sea surface foam. 13 • The TSKY module will account for the space, sun, and atmospheric effects. It includes the estimation of galactic contributions, sun-glint effects, atmospheric attenuation, and atmospheric brightness temperature. Figure 8. Brightness temperature as a function of surface wind speed, for surface salinity of 36 psu and surface temperature of 5°C. x indicates HH polarisation, while o indicates VV. Parameters from Hollinger and Lo (1981). Figure 9. Top of atmosphere brightness temperature as a function of surface salinity for wind velocity of 10m s-1 and surface temperature of 5°C, at 1.4 GHz. x indicates HH polarisation, while o indicates VV. Goodberlet-Swift model . 14 2.3 Review of Spaceborne and Airborne Experiments Microwave remote sensing was first applied in the 1930s by radio astronomers to observe and measure extraterrestrial sources of electromagnetic energy. In the late 1950s, radiometers recording in the 4.3 cm wavelength band (6.98 GHz) were designed to measure solar temperature and atmospheric attenuation. These measurements were also used for the observation of terrestrial targets such as grass, asphalt, and water (Straiton et al., 1958). 2.3.1 Remote Sensing of Sea Surface Salinity in Canada Universities, government agencies, and private companies conduct Canadian research and development activities relevant to remote sensing of sea surface salinity. Currently, three Canadian universities host research groups on passive microwave remote sensing. Sea ice and snow properties are being studied at the University of Manitoba and the University of Sherbrooke. Applications to soil moisture mapping are being developed at the University of Guelph and the University of Sherbrooke. The University of Guelph formed, with the Centre for Research and Space Technology and the Department of Agriculture, a research group to support the development of a RADARSAT soil moisture algorithm. Data collected for these projects included in situ measurements at agricultural sites in Ontario using a track mounted L-band radiometer and L- and C-band radar systems. An airborne survey was also flown in order to test coarser scale capabilities of the microwave sensors. This involved the deployment of the step frequency radiometer, SLFMR, over agricultural fields in Ontario and Quebec. The Compact Airborne Scanning Imager, recording in the visible and near infrared range, was also mounted on the aircraft. This allowed the research group to test the synergy between optical and microwave sensors and to evaluate the effects of different land surface types on passive microwave observations. An analysis of the data indicated that soil moisture retrievals can be made by using L-band radiometry. Xu et al. (1998) published the results of these surveys. In addition, dry-down experiments were conducted at the University of Guelph during the 1999 summer season. Unfortunately, due to the funding limitations, research on L-band radiometry was discontinued at University of Guelph. Environment Canada owns and operates several very-low and high frequency radiometers for in situ and airborne campaigns. The 1.4 GHz and 5 GHz radiometers were acquired recently and are still undergoing calibration testing. Other sensors record in the 18 GHz to 90 GHz band and are used to study clouds, sea ice, and snow. Most of these sensors were acquired 6 to 10 years ago. However, because of calibration or other technical problems, very little information was published on the analysis of the observations. An evaluation of publications related to the status of radiometry applications for the monitoring of ocean salinity indicates that only prototype retrieval algorithms exist. As well, these prototypes are specific to their respective studies. A consistent method for 15 accurate derivations of the ocean surface salinity must be developed to test and evaluate L-band sensor capabilities. This is a research subject of particular interest for Canadian university networks such as the Geomatics for Informed Decisions Network (GEOIDE), as well as individual research centres. Numerous industrial groups are working on sensor development and microwave antenna design. For example, EMS is building a L-band synthetic aperture antenna and ComDev is developing up- and down-link spaceborne communication sensors. In addition, several Canadian government and private agencies are focusing on building and setting up receiving stations (e.g. MDA, CCRS), developing signal processing methods (e.g. MDA, AUG Signals, RSI) and planning efficient data delivery to the user (e.g. CCRS, RSI, MEDS). 2.3.2 Spaceborne Experiments In 1962, Mariner 2 carried the first spaceborne passive microwave instrument. During that mission, data were recorded with a two-channel radiometer sensitive to the wavelengths of 1.35 cm (2.22 GHz) and 1.9 cm (1.58 GHz). In 1968 and 1970, passive microwave observations of the Earth were made with four radiometers on the Russian Cosmos satellites. In 1973 and 1974, a 1.4 GHz radiometer was mounted on the US Skylab-4 (Lerner and Hollinger, 1977). The Cosmos data acquisition lasted only a few days and was not sufficient for evaluation of the radiometer sensitivity to sea-surface temperature and salinity variability. The Skylab radiometer measurements provided data that confirmed the theoretical dependence of brightness temperature on salinity and sea surface temperature at the 1.4 GHz frequency. In 1978, a team of US scientists prepared a report describing the potential applications of spaceborne passive microwave systems. This team recommended the development of an imaging multi-frequency passive radiometer sensitive to the 1.4 to 185 GHz range, with a spatial resolution of about 5 km, a full orbital cycle of 6 to 12 hours, and a 3 to 4 m antenna. At present, none of the spaceborne passive microwave radiometers has an antenna larger than 2 m. This size antenna limits applications to about 100 km resolution (21 cm wavelength). The sensors working in the 19 to 150 GHz frequency range permit the monitoring of oceanic and atmospheric variables such as ocean surface wind speed, sea surface temperature, ice cover type and concentration, atmospheric water vapour and droplets density, and atmospheric temperature profiles. One of the terrestrial applications gaining the acceptance of user groups is the monitoring of snow accumulation and depletion. The most recent research and development activities in the application of microwave radiometry involve the measurement of precipitation rates, ocean surface salinity, and soil moisture. 16 2.3.3 Airborne Experiments Airborne observations are the main data source used in evaluating passive microwave radiometers for estimating sea surface salinity. Experiments started in the early 1970’s (Thomann 1973, 1976; Armand et al., 1979). More recently, experiments are being conducted at research facilities located in the University of Massachusetts at Amherst, the NASA’S JPL and GSFC laboratories, as well as at the Naval Research Laboratories (Table 2). Currently, only a few airborne salinity sensors are being used or developed for salinity measurements. A majority of the applications have been made in the last 7 years (see Table 3), starting with Scanning Low Frequency Microwave Radiometer (SLFMR) commissioned over the Chesapeake Bay and Delaware Bay areas. The 1996 mission, part of the Chesapeake Outflow Plume Experiment sponsored by NOAA and NRL (Goodberlet et al., 1997), produced the first sea surface salinity map (Lagerloef, 1998; Miller et al., 1998). The SLFMR was also flown over Charleston Harbor, South Carolina and the nearby coastal zone. Miller (2000) produced a 2 km resolution salinity map of Charleston Harbor and salinity transects for the coastal waters North and South of Charleston. The salinity retrievals have yet to be verified with in situ data. Table 2. Airborne microwave radiometers and some of their characteristics. Instrument ESTAR SLFMR PALS Developer GSFC-NASA Univ. of Massachusetts Quadrant Engineering JPL-NASA System Configuration Electronically Scanning Thinned Array Radiometer Pushbroom - Coastal, Gulf Stream Various sites 3 V-polarised beam 3 H-polarised beam Step-frequency radiometer used by NOAA and NRL (Miller et al., 1998) Passive and active L- and S-band Area of Interest Gulf Stream Technical Specifications of Interest Imaging with synthetic aperture techniques; 2-channel correlation radiometer connected to two mobile antennas mounted on a positioner and operating as a two element interferometer 17 Table 3. Airborne deployments that used passive L- and S-band sensors. Sensor Over Flight Location ESTAR Gulf Stream PALS 1999 Southern Great Plains Experiment (SGP99) SLFMR Date 1999 July 1999 2001 Soil Moisture Experiments In Progress Chesapeake Bay and Delaware Bay. April 1994 Charleston Harbor November 1996 STAR-Light Planned for Arctic land-surface hydrology investigation. 2003-2004 L- and Cband radiometers Arctic Ocean from Thule, Greenland July 2000. Salinity data and simultaneous in situ data were obtained with the ESTAR airborne radiometer in 1999. Howden and LeVine report the preliminary results of these surveys in Lagerloef (2000). In 1999, the Passive/Active L/S-band Airborne Sensor (PALS) was used for sea surface salinity test flights. The PALS and ship track data were recorded on the same day (although recorded 10 km apart). Results of this experiment are presented in Lagerloef (1998). Publications reporting airborne data experiments for sea surface salinity mapping do not include the retrieval algorithms. Given the information provided, one can assume that a look-up table approach or sets of regression equations are used. Sea surface temperatures are always measured in those surveys and theoretical values were used to correct for the atmospheric contributions and wind speed effects. Therefore, the sea surface salinity can be derived as a function of the brightness temperature and the sea surface temperature. However, a dual frequency radiometer recording in the 1.4 GHz and 2.65 GHz frequencies will eliminate the need for simultaneous sea surface temperature data, whereby theoretical or empirical equations can be used to derive sea surface salinity values (Blume et al., 1978). Experiments in an open ocean environment have been limited to the Gulf Stream, where NASA deployed the ESTAR and the SLFMR in 1999. The Gulf Stream represents an area of choice for the testing of remote sensing of sea surface salinity. This is because over a distance of about 300 km, from near shore bays to the Sargasso Sea, the salinity variation may exceed 20 psu. For the 1999 experiment, a NASA P3 aircraft was flown over the continental shelf waters, the continental slope waters, and the Gulf Stream, measuring the brightness temperature at 1.4 GHz with ESTAR and SLFMR. In addition, the NASA C130 carried prototype L- and S-band radiometers built by JPL. The aircraft 18 measurements were validated using in situ measurements from the R/V Cape Henlopen, the M/V Oleander, and three surface drifters. The aircraft also carried a pyrometer to measure sea surface temperature and a scatterometer for wind speed. The results of these experiments have not yet been made public. 2.4 Proposed Sea Surface Salinity Systems Results from the most recent airborne experiments with ESTAR and SLFMR are encouraging for potential spaceborne measurements. Indeed, the airborne experiments indicate that space measurements could approach a SSS accuracy of 0.1 psu averaged over a 300 km resolution cell for a period of one month. This level of accuracy would be useful for climate studies. The encouraging airborne results have prompted both American and European plans for spaceborne salinity measurement. Presently, the spaceborne concept is being promoted in the USA through the HydroSphere and Aquarius programmes and in Europe with the Soil Moisture and Ocean Salinity Mission (SMOS) mission. SMOS is at the most advanced stage of development with a mission planned for a satellite launch in 2006. Instrumentation related to these spaceborne programmes is also under development. In the United States, two instrument designs have received special NASA funding. The first system, OSIRIS, was developed by the JPL. OSIRIS is a large mesh antenna system aimed at attaining the highest possible SSS measurement accuracy. The system operates with a conical scanner, uses L- and S-band radiometers, and reads H- and V-polarised signals. In addition, a L-band radar system is designed to concurrently record energy backscattered from the surface. This last component is meant to collect data for wind and sea state correction. Pre-phase A studies are proceeding with the NASA Instrument Incubator Program. The second instrument design plan (so far unnamed) will be implemented by the GSFC. This plan calls for a two-dimensional aperture system to measure soil moisture and sea surface salinity. A vast amount of information on the most recent advances in spaceborne sea surface salinity remote sensing is regularly assembled in workshop reports of the Salinity Sea Ice Working Group (SSIWG). The mandate of this research group is to evaluate both the scientific merit and technical feasibility of measuring sea surface salinity using a spaceborne remote sensing platform. In 1998, the first SSIWG workshop established guidelines for the development of research avenues (Lagerloef, 1998). The second SSIWG workshop focused on defining the scientific requirements for sea surface salinity mapping. As well, the workshop outlined the technical support available for the collection of in situ salinity data (Lagerloef, 1999). The third workshop report (Lagerloef, 2000) provides an insight on the various in situ salinity data collection experiments conducted in 1999 and concludes with a series of scientific requirements. 19 Currently, research activities are being directed to improving antenna design and scanning mechanisms. As well, planning for satellite missions continue in research centres in the U.S. and in Europe (see Table 4). Aquarius is the most recent NASA remote sensing program for sea surface salinity. Table 4. Summary of activities and publications related to space borne passive microwave remote sensing missions for sea surface salinity mapping. 2005 Sponsor (s) or Author ESA 2001 ESA Pilot project 2001, Feb. 28 NASA Oceanography Scientific Conference. 2000, Dec. JPL/NASA 2000, Nov. Njoku et al., 2000 2000, March SSIWG 1999, July USDA ARS Hydrology Lab 1999, June 1999, June SSIWG ESA and CNES 1999, May 15 ESA 1998, Dec. 1998, Nov. SSIWG ESA and CNES U. of Michigan, NASA-GSFC, LRC, and others. Year 1998 Activity Planned launch of SMOS mission. CASA plans to have a full demonstrator of MIRAS finished. The Aquarius Mission – Sea Surface Salinity from Space. Talk presented by C. Koblinsky, NASA-GSFC. HydroSphere proposal development in progress for submission to the NASA Earth System Science Pathfinder (ESSP) program. Published paper: A large-Antenna Microwave Radiometer-Scatterometer Concept for Ocean Salinity and Soil Moisture Sensing. Report of the 3rd Workshop; Lagerloef, 2000. Report by T.J. Jackson (1999): NASA Post-2002 Land Surface Hydrology Mission: Soil Moisture Research Mission (EX-4) Report of the 2nd Workshop; Lagerloef, 1999. MIRAS-SMOS Phase A study approved by ESA Report by NERSC: Study of Critical Requirements for Ocean Salinity Retrieval using a Low Frequency Microwave Radiometer. (ESA report 98-S30) Report of the 1st Workshop; Lagerloef, 1998. SMOS space borne mission proposed to ESA HYDROSTAR space borne mission rejected by NASA ESSP program 20 HYDROS Application Radiometer Frequency Soil Moisture University of Michigan and Goddard Space Flight Centre 2006 3 years Sun-synchronous at 670 km; ascending node at 0600 local time; 3 days revisit L-band radiometer; Lband radar 1.4 GHz Radiometer Polarization H, V Agency Planned Launch Planned lifetime Orbit Instruments Antenna Type Incidence angle Swath Width (km) Spatial Resolution (km) Single observation error (psu) Conically scanning parabolic constant 40° 900-1000 km 40 km AQUARIUS SSS SMOS 2006 2 years SSS and Soil Moisture European Space Agency and Centre National d’Études Spatiales 2006 3 years Sun-synchronous, ascending node at 0600 local time; 8 day revist Sun-synchronous at 755 km; ascending node at 0600 local time; 3 day revisit L-band radiometer; L-band scatterometer 1.4 GHz L-band 2D interferometer 23.3°, 33.7°, 41.7° 250 70-90 km 1.4 GHz H-H and V-V (cross-polarization mode optional) Thinned array, Two-dimensional synthesis 15 - 50° 620 to 1050 35 to 50 km 0.43 at mid-latitude 1.2 in warm seas H-H and V-V parabolic Table 5. Technical and performance specifications of passive radiometry satellite missions currently in progress. Sources: HYDROS - HYDROS web site; AQUARIUS - Koblinsky et al. (2001); SMOS - Kerr (1998). 21 3.0 Salinity Databases The importance of salinity in the oceanographic framework has been established in previous sections. However, available salinity data must also be considered to place in context the ocean salinity signals of importance to Canadian issues. As well, the potential contribution of satellite based salinity measurements to the global data archives should be established to indicate the importance of such measurements. 3.1 Global Databases Global databases containing information on sea surface salinity are maintained through the Intergovernmental Oceanographic Commission (IOC). The IOC, a component of UNESCO, co-ordinates a network of data centres operated through the Intergovernmental Oceanographic Data and Information Exchange system (IODE). The IODE, made up of co-operating national data centres from United Nations member countries, distributes regional or data type responsibility to participating national data centres. Also, IODE maintains the network of World Data Centres, designated deep archives for global ocean data. Within the IODE system and particularly important for salinity data, are the USA National Oceanographic Data Center and the Canadian Marine Environmental Data Service (MEDS) National data centres of the member countries collect national oceanographic datasets. These centres then transfer the dataset, through international agreements, to the IODE responsible data centres to become part of the IODE system. 3.1.1 USA National Oceanographic Data Center The USA National Oceanographic Data Center (NODC) is one of three NOAA environmental data centres. Co-located with World Data Centre-A, NODC serves as a national repository and dissemination facility for global ocean data. The Ocean Climate Laboratory, a division of NODC, is mandated by the NOAA Climate and Global Change program to maintain quality controlled oceanographic databases. Data include historical in situ measurements of variables including temperature, salinity, oxygen, phosphate, nitrate, silicate, and chlorophyll. Data within the IODE system is publicly available through various avenues. The NODC makes data available online and on CD at different levels of processing. In terms of salinity data, relevant NODC online products include the Oceanographic Profile Database. Relevant CD products include the World Ocean Database 1998, the World Ocean Atlas 1998, the World Ocean Atlas Figures 1998, and the World Ocean Atlas 1994 - LEVITUS94. 22 3.1.1.1 Online Oceanographic Profile Database The Oceanographic Profile Database is an interface for locating and retrieving oceanographic profile data from NODC. Queries to this database are made in two steps: • Searching and generating an inventory of stations, and • Plotting and retrieving the data from the inventory. The system allows the user to search for oceanographic stations, to plot station location, and to plot 2D graphs of a selection of 26 variables, geographic location, and time. All variables can be plotted as horizontal or vertical profiles. The search may also be filtered by sampling method. For example, filtering based on bottle samples or ConductivityTemperature-Depth (CTD) profiles. The collated datasets can be downloaded via FTP. A standard data query session in the Oceanographic Profile Database consists of: 1) accessing the database, 2) making lists of the number of data files containing salinity, 3) plotting a map of the station location, and 4) generating sample vertical profiles. Figure 10 illustrates a vertical plot extracted from the NODC online Global Oceanographic Profile Database for the Queen Charlotte Islands area. 3.1.1.2 World Ocean Database 1998 Version 2 (WOD98) The World Ocean Database 1998 Version 2 (WOD98) is the NODC database containing raw data. Available data files are queried using filter options based on data type, time period, or year of interest. 3.1.1.3 World Ocean Atlas 1998 (WOA98) The World Ocean Atlas 1998 (WOA98) is an analysed fields dataset. Data files matching selected criteria are provided as “gz compressed” files for downloading. This comprehensive site allows salinity data to be selected by: • time period: annual, seasonal, or monthly, • grid size: 1 or 5 degree, and • type of analysis: data distribution, objectively analysed climatological mean, observed mean minus the analysed mean, standard deviation of the observed mean, standard error of the observed mean, or analysed mean minus the annual mean. 23 Figure 10. Vertical salinity profile for a station off the Queen Charlotte Islands on July 1, 1997. Figure 11. Global annual mean surface salinity (PSS), from the WOA98F database 24 3.1.1.4 World Ocean Atlas 1998 (WOA98F) The World Ocean Atlas 1998 Figures (WOA98F) is a searchable database of images, where queries are based on the: • time period: annual, seasonal, or monthly, • area of interest: global or regional (Atlantic, Indian, or Pacific Ocean), and • type of analysis: data distribution, objectively analysed climatological mean, observed mean minus the analysed mean, standard deviation of the observed mean, standard error of the observed mean, or analysed mean minus the annual mean. Datasets are displayed in a variety of filled colour or contour formats. Contour intervals are automatically set as a function of the full salinity range. The WOA98F database generates global map images with a 0.2 psu resolution over a salinity range of 33.0 - 37.4 psu. Figure 11 and Figure 12 present examples of global images from this database. Figure 11 displays the objectively analysed climatological mean surface salinity for all annual data available, while Figure 12 shows the winter season (January – March) mean surface salinity. Regional annual surface salinity image maps can be obtained from the WOA98F for three predefined geographic areas: the Pacific, Atlantic, and Indian Oceans. Figure 13 presents an example of the 1998 annual average surface salinity in the Pacific Ocean. The NODC maintains a global oceanographic profile database that is queried through the “Station Search Form”, whereby user-defined latitude and longitude boundaries are used to define the geographical extent for data extraction and representation. Time filtered data are available for monthly or annual periods in MM/DD/YYYY format. A search of the area 140° to 124°W, 55° to 46°N for the continuous period of 01/01/1980 to 01/01/2000 returned the summary in Table 6. 25 Figure 12. Winter mean surface salinity (PSS), from WOA98F. Figure 13. Annual mean surface salinity (PSS) in the Pacific Ocean, from WOA98F. 26 Table 6. Search results for oceanographic profiles in the NE Pacific, 1980 to 2000. Variable Profiles TEMP (Temperature) 17600 SAL (Salinity) 17688 DOXY (Dissolved Oxygen) 1293 PHOS (Inorganic Phosphate) 101 SLCA (Silicate) 95 NTRA (Nitrate) 44 CHPL (Chlorophyll) 1722 NTRI (Nitrite) 44 TPHS (Total Phosphorus) 2 AMON (Ammonium) 5 LGT$ (Light transmissivity) 1671 NTRZ (Nitrite + Nitrate Content) BEAC () 1672 LBSC () 1780 File Alias C100 (OCEAN STATION DATA (NANSEN)) F022 (CTD/STD, HIGH RESOLUTION) L303 (CTD) Total 3.1.1.5 57 Profiles Stations 5011 1729 24854 12427 13909 3532 43774 17688 World Ocean Atlas 1994 (LEVITUS94) The World Ocean Atlas 1994 (LEVITUS94) is an atlas of objectively analysed major ocean variables and is an excellent resource that appears to provide the most detailed salinity data. It comprises multi-source archival data from 1900 to 1992. Global or regional scale data are presented in tabular or graphic form. Table 7 gives an example of annual mean salinity. Graphic options include colour, contour, and line drawings. These forms are useful for illustrative purposes, but they show a very crude contour resolution. Time filtered summaries can be obtained by accessing the website from three separate points: ANNUAL - LEVITUS94 ANNUAL 27 MONTHLY - LEVITUS94 MONTHLY SEASONAL - LEVITUS94 SEASONAL Three unique display features permit the user to customise the data plot: • geographic windows, which may be used as filters thus allowing one to zoom into an area of interest, • selectable salinity ranges, which allow one to display only the data from within the area of interest, and • line drawing option, which is unique and very useful for the display of a salinity profile along broad latitude or longitude lines. Figure 14 and Figure 15 are examples of annual and seasonal salinity maps. The colour plots are informative in that regions with high salinity variability are easily highlighted. The salinity contour resolution depends on the salinity range selected, as only 10 contour values are displayed. For example, if the user selects a salinity range of 32 - 34 psu, then each contour line will represent a 0.2 psu step. Regional annual SSS map reports from LEVITUS94 are available for variable geographic regions, whose extents are defined by user-defined minimum and maximum latitudes and longitudes. Examples for the Pacific Ocean and the Gulf of Alaska are presented in Figure 16 and Figure 17. Table 7. Example of annual mean salinity data (psu) from LEVITUS94 28 Figure 14. Global long-term mean surface salinity (psu) from LEVITUS94. Figure 15. January to March global mean surface salinity (psu), from LEVITUS94. 29 Figure 16. Annual mean surface salinity (psu) in the Pacific Ocean, from LEVITUS94. Figure 17. Annual mean surface salinity (psu) in the Gulf of Alaska from LEVITUS94. 30 3.1.2 Marine Environmental Data Service (MEDS) MEDS is the Canadian national oceanographic data centre and is a branch of the Canadian Department of Fisheries and Oceans (DFO). Its mandate is: 1) to archive ocean data collected by DFO as well as data acquired through national and international programmes conducted in ocean areas adjacent to Canada, and 2) to disseminate data, data products, and services in accordance with DFO policies. Within the IODE system, MEDS is the responsible data centre for upper ocean thermal data. In this capacity, MEDS collects and manages global upper ocean temperature and salinity data under the international co-operative Global Temperature and Salinity Profile Program (GTSPP) (see www.nodc.noaa.gov/GTSPP/gtspp-home.html). The main purpose of this project is to collect all relevant upper ocean data and maintain the necessary databases to support this process. MEDS acquires the data via two distinct data streams. The first, termed the real-time data stream, uses the Global Telecommunication System (GTS) to deliver the reduced (or lower resolution) data profiles to numerous clients including MEDS. MEDS quality controls the real-time data and archives these data within their end-to-end management framework. On an annual basis, MEDS acquires approximately 60,000 oceanographic profiles over the world’s oceans via the GTS. As contributor to the IODE system, MEDS forwards the acquired profile data to World Data Centre-A three times per week. The real-time database (located at www.nodc.noaa.goc/GTSPP/gtspp-rt.html) contains global ocean temperature and salinity measurements that are regularly made by shipboard observers or automated instruments such as buoys. The database allows the operator to select a month and to 1) view the data file summary, 2) view the location plot (typically ship routes), and 3) download the dataset in MEDS ASCII file format. The GTSPP real-time datasets provide monthly data from January 2000 to present. All real-time data are acquired by MEDS in the form of reduced profiles or “messages” (e.g. inflection points of the profile). The real-time message may relate to only a temperature profile, or to a temperature-salinity (TS) profile. Figure 18 shows the number of TS profiles collected in real-time by MEDS for 1998, over the global ocean. The general lack of data over the global ocean is evident. MEDS also receives the full-resolution data profiles at a 1 to 4 year time frame after the real-time stream. This constitutes the delayed mode (higher resolution) data stream. Upon receipt, MEDS quality controls the dataset and updates the end-to-end database with the most current and complete dataset. 31 Within the GTSPP system, the updated dataset becomes part of the “Best Copy” database. This database is assembled to provide the most complete dataset without duplication (see www.nodc.noaa.gov/GTSPP/gtspp-bc.html). Real-time profiles are contained in the file if the higher resolution delayed-mode profiles have not been replaced by any of the contributing data centres. Available files are arranged in quarterly segments. GTSPP has incorporated the World Ocean Circulation Experiment (WOCE) Upper Ocean Thermal (UOT) science centre flags, and thus continues the co-operation between science programs and the data management community. Data are available for download, but the only graphical output available is that of the ship track routes. The records are typically provided by the “Ships of Opportunity” and thus cover the major routes of the Pacific, Atlantic, and Indian Ocean. The Best Copy database is complete and is not filterable on-line. Thus, the client must extract specific variables such as the surface salinity. Also, data files are available only at quarterly periods. Figure 18. Location of temperature-salinity profiles acquired by MEDS over the global ocean, for 1998. The colour scale indicates the number of observations. 3.2 Canadian Regional Databases 3.2.1 Atlantic Region The Bedford Institute of Oceanography (BIO) Climate Database consists of assembled historical bottle and CTD data sets that were collected using standard quality control procedures (Petrie et al., 1996). Data compiled for winter and summer seasons are used 32 to calculate mean temperature, salinity, and density fields. These data are interpolated onto a grid in a format useful for numerical modelling and process studies. BIO also maintains the Computer Atlas of the Northwest Atlantic (Yashayaev 1999a and b). This atlas contains all available temperature-salinity data from approximately 1900 to 1997, and includes data from the MEDS archive, the Russian Sections Program, BIO archives including non-released WOCE cruises, as well as data contributed by collaborators, and contained in WOA98F. The majority (90%) of the data used in the Atlas originates from bottle sampling, whereas the remaining data was collected in recent years with modern equipment such as CTD sensors and current meters. The existing version of the Atlas has undergone a preliminary data quality control. Some erroneous profiles and noisy data exist, but these only affect the statistics in areas where there is sparse sampling. Simple queries allow the user to visualise the data distribution, display salinity contour maps, and produce time series for a point or for a region. Figure 19. Samples distribution and count of surface salinity data archived in the Computer Atlas of the NW Atlantic. 33 Figure 20. Distribution and count of samples collected in May of all years between 1900 and 1997. The Atlas covers the Northwest Atlantic including the Gulf of St. Lawrence, the Grand Banks, the Scotian Shelf, and the Bay of Fundy. The data distribution images have a resolution of 0.5° latitude by 0.5° longitude. Figure 19 illustrates the spatial distribution and number of samples for the entire data collection period. The database comprises over 750,000 bottle samples and 82,000 CTD records. The highest data counts are found in the Bay of Fundy with more than 600 observations per unit area. The Scotian Shelf has counts greater than 500 per unit area, while the Grand Banks show counts of 300 to 400 observations per unit area in a few places. Figure 20 shows the data distribution at the surface for May and demonstrates that much of the Atlantic zone is only sparsely sampled. In fact, a large percentage of the area shown has no surface observations in May throughout the 97 year period. 34 3.2.2 Pacific Region Three databases maintained by the Institute of Ocean Sciences (IOS) for the Pacific region are important for salinity related studies: the British Columbia (BC) Shore Station Database, the Line P Database, and the Station Papa Database. The BC Shore Station Database contains sea surface salinity data from BC lighthouse locations. The data, some available since the 1910’s, have a reported precision of 0.1 psu. Figure 21 shows the location of the lighthouse stations along the British Columbia coastline. Unfortunately, the very nature of the dataset places it close to land and of limited use for offshore studies at scales comparable to proposed satellite datasets. Line P is a series of oceanographic stations managed by IOS. It extends about 1500 km from the mouth of Juan de Fuca Strait, south of Vancouver Island, to Ocean Station Papa at 50°N 145°W in the Pacific Ocean. Figure 22 shows the original 13 stations. In 1981 the number of stations was increased to 26. The location and history of Line-P are also available on-line. IOS has archived the data recorded at Station Papa since 1956. These data include CTD profiles of the water column. Figure 23 displays the annual cycle of surface salinity averaged over all data gathered from 1956 to the present time along Line P. The maximum variation in salinity occurs within 100 km of the coast and gradients are stronger during the rainy winter months. Figure 21. Map of the British Columbia coast showing the location of the lighthouse stations. 35 Figure 22. General geographic location of the stations along the P-Line Figure 23. Annual cycle of surface salinity from 1956 to the present 36 4.0 Sea Surface Salinity in Canadian Waters 4.1 Spatial Sea Surface Salinity Distributions 4.1.1 Northwest Atlantic Figure 24 was generated using data obtained from the Computer Atlas of the NW Atlantic and displays the mean salinity in the North Atlantic averaged over all years. The average SSS is 32.48 psu with a standard deviation of 1.78 psu. Figure 25 presents a more local view and was generated using data available from MEDS. It shows the horizontal distribution of surface salinity on the Grand Banks of Newfoundland. Several features are evident including the general increase in salinity away from the land towards the offshore. Over the continental shelf, salinity spans a range of 31-33 psu. If the plot extended into the deeper waters of the Labrador Sea, we would see salinity increase by another 2 units to about 35 psu. However, the figure does not extend further because sufficient data do not exist for the area off the shelf. The data used for this figure were collected by DFO during a series of annual fishery assessment and directed oceanographic cruises spanning the fall season (August to November) and such cruises do not extend beyond the 2000 m isobath. The oceanography of this region is dominated by the Labrador Current, which is fed by arctic water draining through Baffin Bay and Hudson Strait as well as by the West Greenland Current. The main branch of the Labrador Current is trapped along the continental slope (600 to 2000 m) and separates shelf waters from the warmer, more saline waters of the Labrador Sea. A weaker inner branch of the Labrador Current is trapped along the coastline. The surface salinity jump across the roughly 80 km wide outer branch is about 0.5 psu, similar to that across the approximately 50 km wide inner branch. Figure 26 shows the horizontal distribution of surface salinity anomalies in fall 2000. This field was computed by subtracting the spatially varying 1971-2000 mean from the field displayed in Figure 25. While there are features that appear suspicious and that are probably related to the fact that the data were collected over a four-month period, a number of spatially coherent features with amplitudes as large as 0.75 psu are apparent. 37 Figure 24. Average salinity based on all years from the Computer Atlas of the NW Atlantic. 38 Figure 25. The horizontal distribution of surface salinity over the Grand Banks of Newfoundland and the Northeast Newfoundland Shelf during the period of AugustNovember 2000. (Figure courtesy of E.B. Colbourne, Northwest Atlantic Fisheries Centre, DFO). 39 3LNO : Fall 2000 - Surface Salinity Anomalies 48 49 3L 0.50 46 0.25 0.00 45 -0.25 -0.50 44 -0.75 43 LATITUDE 47 0.75 3N 3O 54 53 52 51 50 49 48 47 LONGITUDE Figure 26. The horizontal distribution of surface salinity anomalies over the Grand Banks of Newfoundland and the Northeast Newfoundland Shelf during the period of AugustNovember 2000. (Figure courtesy of E.B. Colbourne, Northwest Atlantic Fisheries Centre, DFO). 40 4.1.2 Northeast Pacific DFO occasionally utilises the British Columbia ferries to collect salinity and in vivo chlorophyll fluorescence measurements. Figure 27 presents the salinity (heavy line, left axis) and chlorophyll fluorescence (light line, right axis) measurements recorded by the “Spirit of Vancouver Island” ferry, on its return trip between Swartz Bay and Tsawwassen on March 23, 2001. The total distance of the profile represents the return trip, whereby the beginning and end correspond to Swartz Bay and the middle part matches the vicinity of Tsawwassen. The two salinity dips, visible on the plot, correspond to the passage through the brackish plume of the Fraser River, where high levels of chlorophyll were also recorded. A small patch of high chlorophyll water was measured just after leaving Swartz Bay, but was not encountered on the return trip. Figure 28 presents the salinity values recorded by the Research Vessel Tully, in February 2001, as it travelled between Victoria and the start of Line P. The record begins South of Victoria at 123.3° W, continues along Juan de Fuca Strait to its mouth at 124.8° W, and crosses the continental shelf to deep water at 126° W. The seawater input on the Tully is at a depth of about 3 m. Fresher surface water at the mouth of Juan de Fuca Strait shows a salinity reduction of about 1 psu. Fresher water at the edge of the continental shelf causes a reduction of about 0.7 psu. Another, much older, example of continuous surface salinity horizontal profiles comes from a “Ship of Opportunity” operated in 1979 by Seakem Oceanography (Borstad et al., 1980). Figure 29 shows the horizontal profiles of salinity, temperature, in vivo fluorescence and zooplankton, measured from an intake at 1 m depth. The platform for the data collection was a commercial tug-boat travelling between Howe Sound, North of Vancouver, and Nootka Sound on the North West coast of Vancouver Island. From Howe Sound to about East Point, the records indicate relatively fresh water, which correspond to the southern Strait of Georgia. Between East Point past Sheringham Point, the vessel was traversing the cold well mixed, high salinity waters of Juan de Fuca Strait. Along the southwest coast of Vancouver Island, passing Carmanah Point , the vessel crossed a region of only slightly lowered salinity. Salinity increased to the North as it passed Cape Beale, Amphitrite Island, and Lennard Island. The salinity is slightly lower and stable beyond Estevan Point, until a sharp decrease occurred as the vessel entered the Nootka Sound, at the very end of the record. The salinity values tended to change concurrently with the measurements of the three other biophysical variables - temperature, fluorescence, and zooplankton - as the vessel transited different water masses with varying characteristics. Notably, the fresh waters of the southern Strait of Georgia are warmer, have higher phytoplankton fluorescence, and greater abundance of zooplankton, than Juan de Fuca waters. 41 Vancouver/Victoria Ferry Salinity (ppt 31 12 30 29 28 8 27 26 4 25 -3 Chlorophyll (mg.m ) 16 32 24 0 23.78 23.8 23.82 23.84 23.86 23.88 23.9 23.92 23.94 23.96 Day of March 2001 Figure 27. Strait of Georgia and Gulf Islands; Salinity and chlorophyll data collected from a ferry on March 23, 2001. Figure 28. Salinity horizontal profile from Research Vessel ‘Tully’, February 2001. 42 Figure 29. Horizontal profiles on the North West coast of Vancouver Island, in 1979. 4.2 Temporal Distribution of Sea Surface Salinity Time series for the Canadian eastern seaboard were extracted from the Computer Atlas of the NW Atlantic dataset and are presented in this section, along with time series for the Pacific region (British Columbia lighthouse stations and Station Papa). 4.2.1 Northwest Atlantic Seibert (2001) computed a number of SSS statistics using data in the Computer Atlas of the NW Atlantic. Monthly averages were calculated based on all the profiles available in the geographic area from 42°-70°N and 40°- 68°W. The monthly averaged sea surface salinities, standard deviations, and number of stations are presented in Table 8. Table 8. Sea Surface Salinity Data Statistics Jan Mean (psu) Std. Dev. Sample Count Feb Mar Apr May June July Aug Sept Oct Nov Dec Year 33.1 33.4 33.0 33.3 32.3 32.4 32.1 32.0 31.6 32.1 32.6 33.4 32.6 1.5 1.8 1.5 1.3 1.7 1.8 1.7 2.0 2.1 1.7 1.5 1.6 1.9 2972 3836 6962 9666 11319 12227 13720 11361 7422 7313 7699 2231 96728 43 The data were analysed to estimate the seasonal SSS cycle (annual harmonic) at seven different locations on the Canadian eastern seaboard (Figure 30). Table 9 presents the results of these analyses. Phase indicates the number of days after January 1 on which SSS reaches its minimum. Total root-mean-square (rms) indicates the overall standard deviation at each station while the residual rms refers to the variance unexplained by the annual harmonic and includes both high frequency and interannual variability. The table indicates that the high frequency and interannual variability can be between 58 and 95% of the total signal. However, a more detailed investigation to separate water masses (including the removal of eddies) using cluster analysis (Yashayeav, 2000) indicates that the seasonal - interannual split of variance is about 50% for near-surface salinity. Figure 31 displays similar information for Station 27, which is just outside St. John’s in Newfoundland (see Figure 25 for location). This station has been routinely sampled since 1946. The mean annual signal has an amplitude of almost 1.3 psu (upper panel), although the variations in a given year can be much greater. Interannual variations are considerably smaller (lower panel) with a standard deviation of 0.42 psu. Table 9. Time series statistics for selected stations of the Canadian eastern seaboard. Location Scotian Shelf Western Gulf St. Lawrence Grand Banks Labrador Shelf Labrador Sea Southern Baffin Bay Northern Baffin Bay Latitude (°N) 44 47 47 57 56 67 70 Longitude (°W) 62 62 49 59 51 57 58 4843 1477 3107 521 2304 1120 276 Phase (days) 265 237 237 - 258 304 254 Amplitude (psu) 0.30 1.33 0.49 0.42 0.20 0.74 0.88 Total rms (psu) 0.64 1.05 0.46 0.86 0.24 0.82 0.63 Residual (psu) 0.59 0.85 0.35 0.84 0.20 0.66 0.61 A-to-R ratio 0.5 1.6 1.4 0.5 1.0 1.2 1.4 Variance unexplained by Annual Harmonic (%) 85 66 58 95 69 65 94 Sample Count 44 Greenland 30 00 M Labrador 100 2000M Figure 30. Stations where the seasonal cycle in SSS was computed. 45 Station 27 Surface Salinity Seasonal Cycle 33 32.5 S (psu) 32 31.5 31 30.5 30 Jan Mar May Jul S ep Nov Station 27 August Surface Salinity Anomalies 1 S anom (psu) 0.5 0 -0.5 -1 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 Figure 31. Upper panel: The seasonal cycle in surface salinity at Station 27 (see Figure 25 for location). The vertical bars denote the variability around the long term mean. Lower panel: August anomalies in surface salinity at Station 27. The anomalies are computed relative to the 1971 to 2000 mean. Probably the most striking example of interannual variability is afforded by the so-called Great Salinity Anomaly (GSA; Dickson et al., 1988) that was detectable in the North Atlantic over the period from of about 1969 until 1982. It is useful to discuss the GSA from the viewpoint of available data. Starting in 1964, an international effort was made to collect improved weather data from ocean stations. Weather station Bravo in the Labrador Sea was occupied from 1964 to 1974, and Figure 32 shows surface salinity collected there for the period of 1964-74. The years 1964 to 1967 display a regular annual cycle of about 0.3 psu peak to peak (freshest water in late summer due to ice melt). Between 1969 and 1972 a significant drop of about 0.5 psu occurred, after which the annual cycle increased to about 0.5 psu. The drop corresponds to the passage of the GSA, and represents a volume of surface water fresher than normal by up to 1 psu, about 500800 m thick, and covering an area of about 100,000 square kilometres. The GSA was also 46 detected over the Canadian continental shelf; its effect at Station 27 (Figure 31) is apparent in the early 1970s. This anomaly could be traced around the Atlantic subpolar gyre for over 14 years from its origin north of Iceland in the mid-to-late 1960s until its return to the Greenland Sea in 1981-82. Figure 33 shows the trajectory of the GSA based on data from Belkin et al. (1988). It is anticipated that the Great Salinity Anomaly would have had a profound impact on deep-water formation and associated climate variability, but our ability to detect such anomalies has decreased since the demise of the ocean weather stations. 47 Figure 32. Surface salinity at Ocean Weather Station Bravo in the North Atlantic. Figure 33. The path of the Great Salinity Anomaly (from Belkin, et al., 1998). 48 4.2.2 Northeast Pacific 4.2.2.1 Lighthouse Stations Data from six stations located in the Strait of Georgia is displayed in Figure 34. The summer minimum at all stations is due to melt water from the summer freshet of the Fraser and other Rivers. A lower minimum in winter is due to local rainfall. The annual cycle of salinity at the Amphitrite Point light station (Figure 35), on the west coast of Vancouver Island, includes the standard deviations of individual monthly means about this cycle. Again, surface water near the coast is freshened by terrestrial run-off. The summer salinity maximum is attributed to coastal upwelling of deeper saline water. The Langara Island station is located at the northern tip of the Queen Charlotte Islands and is the most distant from fresh water effluents. The annual cycle of salinity at the station is shown with the standard deviations of individual monthly means (Figure 36). The time series of Departure Bay (not shown), Amphrite Point (Figure 37), and Langara Island (Figure 38) were analysed to extract the monthly average surface salinity anomalies, which are expressed as the rms variation that remains after the annual cycle is removed. Table 10 summarises the results of these analyses. The figures also show the anomaly values for the period of observation. The location of the three stations relative to freshwater runoff and currents is reflected in the anomalies. The anomalies are the highest at Departure Bay station, which is in the Strait of Georgia, followed by Amphitrite Point station, and Langara station. Table 10. Time Series Statistics for three British Columbia lighthouse stations and Station Papa. Station Latitude (°N) Departure Amphitrite Langara Papa 49 49 54 50 124 125 133 145 23 to 27 28.5 to 31 31.9 to 32.1 32.2 to 33.0 Period (years) 100 60 60 40 Anomalies (psu) 1.8 1.0 0.2 0.1 Trend (psu per year) 0 -0.007 -0.01 -0.005 Total variation (psu/period) 0 0.4 0.6 0.2 Longitude (° W) Salinity Range (psu) 49 Figure 34. Strait of Georgia; Annual cycles of monthly mean salinity. Figure 35. Amphitrite Point; Annual cycle of monthly mean salinity. Figure 36. Langara Island; Annual cycle of monthly mean salinity. 50 Y=-0.0066X+13.03; R2=0.021 Figure 37. Amphitrite Point; Monthly average surface salinity anomalies. 2 Y==-0.0099X+19.553; R =0.3177 Figure 38. Langara Island; Monthly average surface salinity anomalies. 51 4.2.2.2 Station Papa Salinity is slightly higher (0.2 psu) at Station Papa than at Langara station, but the annual cycle is similar. There is an annual cycle and a downward trend of about 0.005 psu per year, amounting to 0.2 psu over the period (Figure 39). The annual cycle was calculated as a combination of two harmonics (annual and semi-annual) giving a total peak-to-peak variation of 0.15 psu (Figure 40). The data in Figure 39 were re-plotted in Figure 41 with the annual cycle removed to show residual anomalies. The anomalies at Station Papa are lower that those observed at the three coastal stations and the individual year salinity values are about 0.1 psu above or below the long-term mean. Short term variations occurring in the time series may be correlated with the passage of mesoscale eddies. The surface salinity reaches higher levels after winter storms have transported deep, more saline water to the surface. Lower salinity values may be associated with rain. A more probable cause for decreased salinity at Station Papa in the summer is the advection of less saline water from the coast of North America. The data appear to show a statistically significant freshening trend which in this case has been linked to an increase in stability of surface waters with implications for large scale primary productivity of the region (Freeland et al., 1997). The Station Papa time series was analysed to evaluate the incidence of El Niño events in 1982-1983 and 1997-1998. Figure 42 presents the 1980 to 2000 monthly surface salinity anomalies as well as the long-term trend line based on the 1956 to 1999 measurements. This representation of salinity anomalies does not reveal the last two El Niño events. However, it is noted that the salinity measurements recorded at Station Papa have been 0.1 to 0.2 psu above the long-term trend line. A comparison of the time series from Station Papa (Figure 42) with the Southern Oscillation Index variation for the same period (Figure 43) indicates a possible correlation. There are three negative anomalies in both time series, but the first negative salinity anomaly at Papa was about a year prior to the 1982 SOI minimum, while those of 1987 and 1992 more or less coincided with the Southern Oscillation Index minima. For the 1983 to 1996 period, the Papa salinity anomaly and the Southern Oscillation Index agree very well, but in 1997, the Papa salinity anomaly rose sharply showing the greatest positive anomaly for the period, while the Southern Oscillation Index declined strongly. In Figure 44, the Station Papa surface salinity anomaly with trend removed is compared with the Southern Oscillation Index. The two time series show similar time scales of variation, leading to occasional periods of close correlation, but no correlation over longer intervals. 52 All surface salinities (ppt Station Papa 33.4 33.2 33 32.8 32.6 32.4 32.2 1950 y = -0.0048x + 42.11 R2 = 0.0945 1960 1970 1980 1990 2000 Year Figure 39. Surface salinities in the vicinity of Station Papa from 1956 to 1996 Figure 40. Annual cycle of surface salinity at Station Papa from data in Figure 39. 53 Station Papa All surface salinity anomalies (ppt) 0.8 0.6 0.4 0.2 0 -0.2 -0.4 1950 y = -0.0051x + 10.11 R2 = 0.1359 1960 1970 1980 1990 2000 Year Figure 41. Residual anomalies in surface salinity data at Station Papa, from 1956 to 1996. Y=-0.0027X+5.3932; Figure 42. Monthly surface salinity anomalies at Station Papa, from 1980 to 2000. 54 Figure 43. The Southern Oscillation Index plotted for 1979-1998. Figure 44. Detrended monthly surface salinity anomaly at Papa station plotted with the Southern Oscillation Index. 55 5.0 Science Requirements Table 11 lists a number of science requirements for sea surface salinity that have been derived by a number of groups (Lagerloef, 2000; Schmitt, 1995; CESBIO) as well as some proposed by the authors of this report. These requirements are based either on resolving some ocean process (e.g. deep-water formation) or monitoring a particular ocean area. For most purposes, a product accuracy of 0.1 psu is indicated. The required spatial and temporal scales differ significantly, however, and would seem to prohibit use of any proposed satellite sensor for coastal ocean purposes. This issue is further discussed in Section 8.0. On the other hand, higher resolution sensors are to be expected as technology improves. Table 11. Summary of scientific requirement for the study of surface circulation patterns. Process/Objective Accuracy (psu) Spatial Resolution (km) Temporal Resolution (days) North Atlantic thermohaline circulation 0.1 100 30 Halosteric Adjustments 0.2 200 7 Tropical Pacific heat flux barrier layer effects 0.2 100 30 Surface freshwater flux balance 0.1 300 30 Global scale climate modelling (GODAE) 0.1 200 10 Fisheries Management 0.1 ≤ 50 45 Environmental Monitoring (Coastal) 0.1 10 5 56 6.0 Ongoing Applications of Sea Surface Salinity 6.1 Fisheries Resources in Canadian Waters 6.1.1 West Coast Surface salinity data is used in fishery forecasts (Dr. Blackbourn, pers. comm.). It has been demonstrated that salinity levels are related to the survival of juvenile salmon in very nearshore waters. It has been speculated that surface salinity away from the coast may be related to migration, but this kind of data has never been available. At present, the only available time series are those taken at the coastal lighthouses. DFO stock assessment scientists state that surface salinity is the most common physical variable used in fishery forecasts off the coast of British Columbia. Usually, a simple regression model is used to predict returning salmonids, particularly in the spring. These studies tend to use monthly mean salinity at lighthouses (very close to the coast). Such studies may be enhanced through the use of offshore surface salinity data. Studies indicate that there is a strong correlation between salinity and Barkley Sound Sockeye salmon (Oncorhynchus nerka) juvenile survival rate. The salinity signal is very small, and probably indicates either the onset of the spring transition, increased amount of rain, or an adjustment of the coastal ecosystem. Low salinity may also be the result of a mild winter with lots of rain, or an ENSO event, while higher coastal salinity might be caused by cooler drier winters. Hyatt et al, (1990) and DFO (1999) report that shifts in ocean climate result in changes in the community structure and productivity of both the coastal and offshore ecosystems where the West Coast Vancouver Island sockeye salmon spend 1 to 3 years of their life. Marine conditions are especially unfavourable for juvenile survival during years when coastal ocean temperatures are high and salinities are low, because migratory predators, including Pacific hake (Merluccius productus) and mackerel (Scomberomorus sierra), arrive earlier and in greater abundance. 6.2 Optical Water Properties Measurements of water quality variables such as chlorophyll a, coloured dissolved organic matter, or sea surface salinity from satellite sensors would be of great interest to resource managers monitoring coastal regions. The contribution to blue light absorption by coloured dissolved organic matter is recognised, but the standard ocean colour algorithms to estimate such variables generally fare poorly in these waters due to the complex nature of the light field. Independent salinity measurements via passive 57 microwave might help improve chlorophyll a retrievals in coastal Case 2 waters. Freshwater runoff, on many parts of the British Columbia coast, exhibits a strong absorption at short wavelengths by dissolved organic materials (tannins, lignin and other products of plant litter breakdown). This is particularly evident when the waters of local creeks and rivers are often visibly brown. Figure 45 illustrates the inverse relationship between the surface salinity and absorption and concentration of dissolved organics as measured by the optical density at 350 nm of filtered surface (OD350) water samples (Borstad, 1987). These were collected in April 1987 from Barkley Sound on the West coast of Vancouver Island. This suggests that fresh water from local rivers, and possibly effluent from the pulp mill on the Alberni Canal at the head of Trevor Channel, are important sources of coloured dissolved organic matter. The geographic variations in OD350 should be apparent in blue to green ratio images. Higher OD350 values recorded at stations 6, 8, 9 and 21 have not been satisfactorily explained, and they may be related to a lab measurement error. An inverse relationship between coloured dissolved organic matter and salinity is also documented for the coast of Florida (D’Sa, et al., In prep.) and in the Atchafalaya delta in the Mississippi River system (Figure 46). Several sets of field measurements collected for Florida coastal waters indicate a conservative mixing or dilution behaviour for the optical properties of the dissolved material (Carder et al., 1993; Blough et al., 1993; D’Sa et al., 2000a,b; Hu et al., 2000). The intercept for OD350=0 occurs at a salinity of about 34 psu for the British Columbia coast and for OD400=0 at 37 psu for the Atchafalaya delta. This possibly represents the characteristic of the salinity in Northeast Pacific and North Atlantic Oceans, respectively. The slopes of the mean trend lines, 0.0066 and 0.119 (m-1 psu-1), describe the absorption coefficient of light where the latter value indicating the much higher concentration of dissolved organic matter in the fresh waters of the Mississippi. Similarly, a high correlation (r2=-0.85, n>7500) between coloured dissolved organic matter and sea surface salinity in the continental margin of northeastern Gulf of Mexico suggests that sea surface salinity distribution maps, as presented in Figure 47, may be derived from satellite watercolour data (Hu et al., 2000). Algorithms based on SeaWiFS data are being developed with the objective to retrieve realistic and accurate estimates of chlorophyll a and coloured dissolved organic matter distributions in Florida Bay and Shelf waters. One of the algorithms is for atmospheric correction while the other is for bio-optical calibration. The often-conservative mixing or dilution behaviour in coastal waters permits the calculation of salinity from coloured dissolved organic matter absorption. 58 Figure 45. Relationship between dissolved organic material concentration (OD350) and surface salinity in Barkley Sound, April 1987. Figure 46. Correlation between Gelbstoff (OD400) and sea surface salinity in Atchafalaya River, March 2001. 59 Chlorophyll-a from NASA atmospheric correction and OC2 algorithms CDOM from MODIS Chlorophyll-a from turbid water correction and MODIS semi-analytical algorithms Salinity estimated from DO400 Figure 47. Sea surface salinity distribution maps estimated from SeaWiFS water colour acquired on February 24, 1999. 60 7.0 Effects of Sea ice on the Remote Sensing of SSS The presence of sea ice in various forms and concentrations is one of the difficulties associated with SSS remote sensing in the polar regions. More generally than the effects of sea ice, the remote sensing of SSS in high latitudes raises four main concerns: 1) the relatively low brightness temperature/SSS ratio at low water temperatures, 2) the effects of ice on the retrieval of SSS, 3) salinity anomalies due to ice melt, and 4) natural sea surface salinity variability in space and time. This latter point is of great importance to the understanding and the monitoring of largescale events such as: • Break-up of Antarctic ice shelves; • Interannual sea surface salinity variations in the Nordic Seas for long time scale climate predictions and modelling; • Monitoring of events such as the Great Salinity Anomaly in the Northern North Atlantic (Dickson et al., 1988); • North Atlantic Oscillation (Hurrel, 1995; AWI website); • Observation of the North Atlantic thermohaline circulation (Schmitt, 1998). The interest in SSS measurements in areas with sea ice coverage is twofold. First, low frequency radiometry is attractive, as it would contribute to the knowledge of the ice characteristics. The SSIWG (Lagerloef, 1998) suggested ice mapping with L-band data would: 1) differentiate melt ponds on consolidated sea ice from open water in combination with C-band radiometric data and 2) allow estimates of multiyear sea ice concentrations, which is not currently feasible with the high frequency microwave data. Second, sea ice information is needed to evaluate its effect on the 1) accuracy of sea surface salinity retrievals and 2) surface salinity anomalies resulting from ice melt. Various organisations have studied the characteristics of sea ice related salinity anomalies from buoy data. However, the only literature sources on data from L-band passive remote sensing of sea ice are the Arctic Ice Dynamics Joint Experiment (AIDJEX) reports. Theoretical L-band models have been compiled by Ulaby et al. (1986). The AIDJEX study dealt with the 0.1 to 40 GHz microwave radiometric characteristics of snow and ice (Vant et al., 1978; England, pers. com.). AIDJEX project reports were complied by Vant et al. (1978), Campbell (1972), Edgerton (1972), Ramseier (1972), and Weller (1972). Vant et al. (1978) published a comprehensive report on the modelled as well as the measured dielectric permitivity and dielectric loss values. The data were, in part, collected using the 1.4 GHz band over three sites of first year ice with different salinities (i.e. 5.1, 5.5 and 10.5 psu). 61 A second data source involves current research from data collected in the Greenland area. As reported at the second SSWIG workshop, the plan called for a fly-over of the Arctic Ocean, in July 2000, from Thule, Greenland using the Wallops P-3. Gloersen (pers. com.) reported that flights were carried out in July 2000 over Baffin Bay. During this mission, C- and L- band radiometric data were recorded over thin and water-covered sea ice. The principal investigator for this project, Dr. D. Cavalieri of NASA Goddard Space Flight Center, has just begun analysis of these data. 7.1 L-BAND Radiometric Characteristics of Sea Ice 7.1.1 Brine In the –10 to 0°C temperature range, the salinity of liquid brine (Sb) is less than 100 psu but, may reach 250 psu at temperatures lower than –10°C (Figure 48). At a frequency of 1.4 GHz, the dielectric loss of brine of salinity 85.6 psu (Figure 49) is three times higher than that of seawater at 32.5 psu (Figure 3) at the same temperature (–5°C for brine and 0° for seawater) (Strogryn and Desargeant, 1985). Brine is therefore expected to be a radiometrically significant constituent of sea ice. 7.1.2 First Year and Multiyear Ice The brightness temperature of sea ice depends on its thickness and growth state. The salinity of first year sea ice may vary from 5 to 15 psu, depending on the depth, its age and temperature. Thin first year ice is transmissive to L-band energy upwelling from the ice-water interface. Thicker first year ice has a penetration depth of 20 to 80 cm, and multi-year ice of 80 cm to 2 meters (Figure 50). The penetration depth of newly formed ice (20 to 100 hours old) is less, in the order of 5 cm (Figure 51), reflecting the higher dielectric losses (i.e. for higher ice salinities) of newly formed ice. During the growth process, from open water to “opaque” ice, the measured brightness temperature gradually increases until the ice becomes "optically thick" to L-band energy (Figure 51). 62 Figure 48. Brine salinity in sea ice as a function of negative temperature (Ulaby et al., 1986) Figure 49. Dielectric constant of liquid brine as a function of frequency (Ulaby et al., 1986). 63 Figure 50. Calculated penetration depth in pure ice and in first year and multi-year sea ice (Ulaby et al., 1986). In order to determine the dielectric properties, one must consider the heterogeneous structure of ice. Debye-like dielectric behaviour pertains for salinities (of water or ice) below 40 psu. However, given the brine inclusion in sea ice, a combination of three dielectric constants, Εi of ice, Εsi sea ice, and Εb of brine, must be used. The shape and orientation of the brine inclusions in sea ice may vary, but are usually columnar and vertical (Tucker et al., 1992). The orientation of brine inclusions is usually random in newly formed ice and gradually takes a vertical orientation as the age of the ice increase. However, after a period of time the brine drains to the lower layers and the ice becomes more homogeneous and thicker. The complex dielectric constant of ice is difficult to predict because it depends on the ionic conductivity of the ice. This parameter is, in turn, determined by many factors including the ice temperature, the brine volume, and the shape, orientation, size and length of the brine pockets (Hallikainen and Winenbrenner, 1992; Vant et al. 1978). Mixing models such as the Polder-Van Santen described by Ulaby et al. (1986), have 64 been used by Hallikainen (1977) to model the dielectric loss of sea ice (for sea ice salinity of 8 psu) at frequencies below 10 GHz. Figure 51 shows how the brightness temperature of newly formed ice varies with time as a result of these combined effects. 7.1.3 Melt Ponds on Consolidated Ice The salinity of melt pond water is less than 1 psu and if refrozen it may display a crystal structure similar to frozen fresh water ponds (Tucker et al., 1992). The spectral characteristics of low salinity seawater indicate that the TB of melt pond water would, if detectable, be higher than the TB of the slightly more saline multiyear ice, where the salinity is about 1 psu (Ulaby et al., 1986). 7.2 Effects of Ice in the Field of View Numerical simulations (Figure 52 and Figure 53) indicate that the uncertainty in SSS retrievals will greatly increase if sea ice is present within the sensors field of view. Using the slope of about 0.2 K/psu at SST=0 (shown in Figure 5) implies that a change of 0.5 psu in SSS corresponds to about 0.1K in Tb. Figure 51 show that first year ice gives a Tb value about 125 K higher than sea water. An equivalent number for multi-year ice is about 110 K. A radiometer will therefore measure an increase of 0.1K in Tb when the sea water in its field of view contains 0.08% of first year ice, or 0.09% of multiyear ice. This corresponds to extremely low concentrations of ice, though for a typical field of view of order 50 km across (Table 5), a single floe would need to cover about two km2 to cause this error. If uncorrected, this increase in Tb corresponds to an SSS measurement 0.5 psu too low (Figure 5). High resolution ice charts generated using Radarsat or ERS SAR sensors or even coarser resolution sea ice cover information derived from SSM/I might be used to separate sea ice and ice-free surface signal contributions at L-band. Figure 51. Brightness Temperature evolution from open water to 15 cm thick young ice. Nadir angle=50°. Just before the 20th hour the high brightness values are reached at a time when the ice became optically thick; with the L-band the ice was then 4 to 5 cm thick (Grenfell et al., 1998). 65 Figure 52. Brightness temperatures at L-band as a function of ice thickness (Ulaby et al., 1986, sea ice parameters used in simulation; CRESTech radiometer simulation model). Figure 53. Simulation of L-band brightness temperature for an ice-covered ocean. The assumption is made that the ocean fraction could have variable salinity. 66 8.0 Concluding Remarks Sea surface salinity is an important oceanic variable with significance for a large number of ocean and atmosphere applications including: • climate dynamics and prediction; • global scale ocean-atmosphere modelling; • regional scale ocean-atmosphere modelling; • fisheries management; and • environmental monitoring. The capability to measure SSS remotely from space now exists, and at least one satellite carrying an SSS should be launched by 2006. Sea surface salinity signals in Canadian waters are large, both in terms of spatial gradients and temporally on seasonal and interannual time scales. Canada thus would stand to gain significantly from remotely sensed SSS. Proponents of SSS satellites have justified their missions primarily in terms of offshore applications, in particular those related to climate change. These include thermohaline forcing in the North Atlantic and equatorial dynamics as they relate to El Niño in the Pacific basin. In fact, the SMOS team (CESBIO) claims that application to the coastal zone, especially for operational modelling where data is required on a time scale of 10 days, is probably beyond the technological capabilities currently proposed. This would seem to limit application of remotely sensed SSS to the offshore region, as important as they are to Canada. However, it is far from clear if this pessimistic view is justified for the many coastal applications identified earlier in this report (e.g., fisheries management). Nor does it consider possible synergies with existing in situ observational programs like DFO’s Atlantic Zonal Monitoring Program or Project Argo. Given the large single pass, single pixel expected error, achievement of a specified SSS accuracy will require the averaging of images over space and/or time. For example, to achieve an accuracy of 0.1 psu, SMOS will average images over 10 day periods and 200 km by 200 km squares (Table 5). The reduction in uncertainty gained by averaging depends on the spatial and temporal structure of the individual errors that contaminate an SSS estimate. If all the sources of noise are white (i.e. if they are uncorrelated between pixels and between successive passes), the error will decrease as the inverse square root of the number of pixels averaged. However, it appears that significant additional effort will be required to determine the error structures. Until then, it will be difficult to predict the success of applying remotely sensed SSS. There exist two special problems in remote sensing of SSS that are of special significance to Canada, namely cold water and ice. The emissivity of cold water is smaller than that of warmer waters, and partially ice filled pixels may prevent useful measurements from being made during winter and spring in the Atlantic zone. On the other hand, the Canadian science community is especially well suited to address such issues. 67 In terms of potential for Canadian benefit, sea surface salinity mapping from spaceborne platforms will provide important information for climate studies. In particular, progress will be significant if it translates to observing salinities in northern regions where historical data coverage is extremely poor. Possibly with the involvement of the Canadian Ice Service, a Canadian contribution to the application of L-band radiometry to salinity monitoring could realistically include a study on the effects of sea ice. As well, benefits will accrue in the areas of fisheries management and environmental monitoring as workers gain access to data on a year-round basis never before possible. This will permit, for example, better estimates of mixed layer depths that are important in ecological models. Similarly, workers will be able to better quantify the fresh water distribution due to estuarine discharge. Canadian expertise in high-latitude ocean studies is recognised in the global science community. As well, Canadian contributions to existing space-based experiments are well known. Canadian businesses, universities and government can make a significant contribution to and benefit from involvement in the satellite based salinity measurements over the global ocean. 68 9.0 References This section includes references cited in the report, a set of relevant websites, and correspondence on the status of various research activities. 9.1 Cited References Armand, N.A.,A.E.,Bashrinov, A.M. Shutko, 1979, Recent Microwave Radiation Studies of Continental Covers, Acta Astronomica,6, pp.647-655 Belkin, I.M., S. Levitus, J. Antonov, S.A. Malmberg. 1998. Great salinity anomalies in the North Atlantic, Prog. Oceanog. 41, 1-68. Blough, N. V., Zafiriou, O. C., and J. Bonilla. 1993. Optical absorption spectra of waters from the Orinoco river outflow: Terrestrial input of colored organic matter to the Caribbean. Journal of Geophysical Research, 98: 2271-2278. Blume, H., J. C. Kendall, B.M., and Fedors, J. C, 1978, Measurements of ocean temperature and salinity via microwave radiometry, Boundary Layer Meteorology, 13: 295-308. Borstad, G. A. 1987. Unpublished Data, Borstad Associates. Borstad, G. A., G. C. Loutitt, R. D. Gale and J. R. Buckley. 1980. Ships of Opportunity Feasibility Study Part 3: Oceanographic Observations. Report by Seakem Oceanography Ltd. for Department of Fisheries and Oceans Canada, DSS file no. 08SB.KF833-8-1066, 172 pp. Borstad, G. and W. Horniak, 2001. Study on the Magnitude of the Surface Salinity Signal in the North East Pacific and its Relation to Climate, Fisheries and Optical Water Properties, Sydney: DFO Contract report, 40 p. Campbell, W.J. 1972. NASA-AIDJEX Remote-Sensing Overfflights/1972 AIDJEX Pilot Study. Kuiper Airborne Observatory Publications, Bulletin No.14:15-17. Carder, K. L., Steward, R. G., Chen, R. F., Hawes, S. K., and Lee, Z. 1993: AVIRIS calibration and application in coastal oceanic environments: tracers of soluble and particulate constituents of the Tampa Bay coastal plume. Photogrammetric Engineering and Remote Sensing, 59: 339-344. CESBIO. Mission objectives and Scientific requirements of the Soil Moisture and Ocean Salinity(SMOS) Mission, www.cesbio.ups-tlse.fr/SMOS2WS/MRD_V5.pdf Debye, P., 1929, Polar molecules: The Chemical Catalog Co., NY, 172p. DFO, 1999. West coast Vancouver Island Sockeye. DFO Science Stock Status Report D605. Dickson, R.R., J.S. Meincke, A. Malmberg, and A.J. Lee. 1988. The "Great Salinity Anomaly" in the northern North Atlantic 1968-1982, Progress in Oceanography, 20: 103-151. D'Sa,E.J., Zaitzeff,J.B., Stewart,R.G. 2000a. Monitoring water quality in Florida Bay with remotely suspended salinity and in-situ bio-optical observations. International Journal of Remote Sensing, 21:(4) 811-816. D’Sa E.J., C. Hu, F. Muller-Karger, and K. Carder. 2000b. Estimation of Colored Dissolved Organic Matter and Salinity Fields in Case 2 Waters Using SeaWiFS: 69 Examples from Florida Bay and Florida Shelf. Pacific Ocean Remote Sensing Conference'2000, Goa, India, December 5-8, PORSEC Proceedings, I: 34-38. Edgerton, A. 1972. Aerojet General Microwave Investigations/1972 AIDJEX Pilot Study, Kuiper Airborne Observatory Publications, Bulletin No.14: 18-19. Freeland, H.J., K.L. Denman, C.S. Wong, F. Whitney and R. Jacques, Evidence of change in the N.E. Pacific Ocean, Deep-Sea Research, 44 (12), 2117-2129, 1997. Goodberlet, M.A.. C.T. Swift, K.P. Kiley, J.L. Miller, and J.B. Zaiteff, 1997. Microwave Remote Sensing of Coastal Zone Salinity, Journal of Coastal Research, 13(2): 363372. Grenfell, T.C., D.G. Barber, A.K. Fung, A.J. Gower, K.C. Jezek, E.J. Knapp, Son V. Nghiem, R.G. Onstott, D.K. Perovich, C.S. Roesler, C.T Swift, and F. Tanis. 1998. Evolution of Electromagnetic Signatures of Sea Ice from Initial Formation to the Establishment of Thick First-Year Ice. IEEE Transaction on Geosciences and Remote Sensing, 36(5): 1642-1654. Hallikainen¸ M. 1977. Dielectric Properties of Sea Ice at Microwave Frequencies, Report S 94, Helsinki University of Technology, Radio Laboratory, Espoo, Finland. Hallikainen, M., and D.P. Winebrenner. 1992. The Physical Basis for Sea Ice Remote Sensing, In Carsey, F.D. (Ed.) Microwave Remote Sensing of Sea Ice. Geophysical Monograph 68, Washington: American Geophysical Union, p. 29-46. Hollinger, J.P. 1971. Passive Microwave Measurements of Sea Surface Roughness, IEEE Transactions in Geosciences and Electronics, GE-9: 165-169. Hollinger J.P., and R.C. Lo, 1981, Low Frequency Microwave Radiometer for N-ROSS, NRL report 481-27. Hu, C., D. Biggs, F. E. Muller-Karger, B. Nababan, J. Vanderbloemen, D.Nadeau, K. L. Carder. 2000. Comparison of ship and satellite bio-optical measurements on the continental margin of the NE Gulf of Mexico. International Journal of Remote Sensing. Hurrell, J.W. 1995. Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation, Science, 269: 676-679. Hyatt, K. D. M. Wright, P. Rankin, I. Miki and R. Traber. 1990. Sockeye salmon recruitment variations. In The marine Survival of Salmon Program. Program outline and investigators summaries for 1989-90: 13-25. Jackson, T.J. 1999. Soil Moisture Research Mission (EX-4). Report of NASA Post-2002 Land Surface Hydrology Mission Planning Workshop held in Irvine, California, April 12-14. URL: http://lshp.gsfc.nasa.gov/Post2002/smm3.html Kerr, Yann H. 1998. SMOS Proposal, CESBIO(CNES/CNRS/UPS), Reference COP 16. Klein, L.A. and C.T. Swift, 1977. An improved model for the dielectric constant of sea Water at microwave frequencies, IEEE Transaction on Antennas and Propagation, AP-25(1):104-111. Koblinsky, C.J., P. Hildebrand, Y. Chao, A. deCharon, W. Edelstein, G. Lagerloef, D. LeVine, F. Pellerano, Y. Rahmat-Samii, C. Ruf, F. Wentz, W. Wilson and S. Yueh. 2001. Sea Surface Salinity from Space: Science Goals and Measurement Approach, Extended Abstract, 11th Conference on Satellite Meteorology and Oceanography. Kraus, J.D. 1966. Radio Astronomy, McGraw-Hill Book Co., New York. Lagerloef, G.S.E., 1998. Preliminary Assessment of the Scientific and Technical Merits for Salinity Remote Sensing from Satellite. Report of the 1st SSIWG Workshop held 70 in La Jolla, California, 7-8 February. URL: http://www.esr.org/lagerloef/ssiwg/ssiwgrep1.v2.html Lagerloef, G.S.E., 1999. Progress Toward Salinity Remote Sensing Satellite Missions. Report of the 2nd SSIWG Workshop held in Greenbelt, Maryland, 19-21, April 1999, URL: http://www.esr.org/ssiwg-2/ssiwg_2.html Lagerloef, G.S.E., 2000. Field Programs and Algorithms, Satellite and Science. Report of the 3rd SSIWG Workshop held in San Antonio, Texas, 22-23 January, URL: http://www.esr.org/ssiwg3/SSIWG_3.html Levitus, S., T.P. Boyer, M.E. Conkright, T. O’Brien, J. Antonov, C. Stephens, L. Statoplos, D. Johnson, and R. Gelfeld, 1998. NAA/NESDIS 18, World Ocean Database, 1, 346pp. Lerner, R.M. and J.P. Hollinger, 1977. Analysis of 1.4 GHz Radiometric Measurements from Skylab, Remote Sensing of the Environment, 6: 251-269. Miller, J. L. 2000. Salinity Mapping with the Scanning Low-Frequency Microwave Radiometer. NOAA URL: http://www.csc.noaa.gov/crs/AECS/SLFMR/ Miller, J. L., M. A. Goodberlet, and J. B. Zaitzeff. 1998. Airborne Salinity Mapper Makes Debut in Coastal Zone. EOS Transactions, American geophysical Union, 79(14): 173, 176-177. Petrie, B., K. Drinkwater, D. Gregory, R. Pettipas and A. Sandstrom. 1996. Temperature and salinity atlas for the Scotian Shelf and the Gulf of Maine. Can. Tech. Rep. Hydrogr. Ocean Sciences, 171: v + 398 pp. Ramseier, R.O. 1972. NASA Ground Truth-Physics of Sea Ice/1972 AIDJEX Pilot Study, Kuiper Airborne Observatory Publications, Bulletin No.14: 20p. Rubinstein, I.G., 1996, Passive microwave observations of the temporal and spatial variability of the snow cover, In Proceedings of IGS International Symposium on Representation of the Cryosphere in Climate and Hydrological Models, August 1215, Victoria, B.C. Rubinstein, I. G. 2001a. Status of science and technology for remote sensing of the ocean surface salinity: Chapter 5 of the VCOS report. Dartmouth: DFO Contract report, 24p+Figures Schmitt, R.W. 1995. The ocean component of the global water cycle, Rev. Geophys., Vol 33, Suppl. (AGU) (found at http://earth.agu.org/revgeophys/schmit01/schmit01.html). Schmitt, R. 1998. GOSAMOR. A program for Global Ocean Salinity Monitoring, A proposed contribution to CLIVAR. Draft proposal, Woods Hole Oceanographic Institution, Woods Hole, URL: (AGU). http://www.bom.gov.au/bmrc/ocean/GODAE/gosamor.htm Seibert, G. H. 2001. Review of Salinity Signals and Data Availability for the Northwest Atlantic, Dartmouth: DFO Contract No.: F5957-000249, 26 p. Simms, É. L. 2001. Literature review on the effects of ice on the retrieval of sea surface salinity from space. St. John’s: DFO Contract no: F6093-001JAH01, 21 p. Skou, N. 1989. Microwave Radiometer Systems: Design & Analysis, Artech House, 1989, 162p. Skou, Niels. 2001. Faraday Rotation and L-band Oceanographic Measurements, unpublished manuscript. 71 Straiton A.W., C.W. Tolbert, and C.O. Britt, 1958, Apparent Temperature Distributions of Some Terrestrial Materials and the Sun at 4.3 mm, J. Appl. Phys., 29, pp 776-782 Strogryn A. and G.J. Desargeant, 1985. The Dielectric Properties of Brine in Sea Ice at Microwave Frequencies, IEEE Transactions on Antennas and Propagation, AP33(5): 523-532. Thomann, G.C., 1973, Remote Sensing of Salinity in an Estuary Environment, Remote Sensing of Environment, 2, pp.249-259. Thomann, G.C.,1976, Experimental Results of the Remote Sensing of Sea Surface Salinity at 21 cm Wavelength, IEEE Trans. Geosci. Electron.,GE-14, pp. 198-214 Tucker III, W.B. D.K. Perovich, A.J. Gow, W.F. Weeks, and M.R. Drinkwater. 1992. Physical Properties of Sea Ice Relevant to Remote Sensing, In Carsey, F.D. (Ed.) Microwave Remote Sensing of Sea Ice. Geophysical Monograph 68, Washington: American Geophysical Union, p. 13-28. Ulaby, F.T., R.K. Moore, and A.K. Fung. 1986. Microwave Remote Sensing: Active and Passive. Dedham: Artech House, Inc., 2162 p. Vant, M.R., R.O. Ramseier, and V. Makios. 1978. The complex-dielectric constant of sea ice at frequencies in the range 0.1-40 GHz. Journal of Applied Physics, 49(3): 12641280. Weller, G. 1972. Radiation Flux Investigation/1972 AIDJEX Pilot Study, Kuiper Airborne Observatory Publications, Bulletin No. 14:28-30. Xu, Q.P., J.B. Boisvert, I. Rubinstein, C. Hersom, N.T. Tremblay, R. Protz, and F. Bonn, 1998. Microwave, Visible, and Near Infrared Spectral Properties as a function of Water Content in Organic Soils, Canadian Journal of Remote Sensing. Yashayaev I.M. 1999a. Oceanographic Computer Atlas of the Northwest Atlantic, 1999, CMOS Bulletin, 26, no.6, P.163-164. Yashayaev I. 1999b. Computer Atlas of the Northwest Atlantic, Atlas of Ocean Sections CD-ROM, 1999. Yashayaev, I. M. 2000. 12-Year Hydrographic Survey of the Newfoundland Basin: Seasonal Cycle and Interannual Variability of Water Masses, ICES CM 2000/L:17. 9.2 Relevant Websites Alfred Wegener Institute for Polar and Marine Research (AWI): http://www.awi-bremernaven.de/index.html Databases http://www.nodc.noaa.gov/General/salinity.html http://www.nodc.noaa.gov/OC5/inv_all.html http://www.nodc.noaa.gov/OC5/data_woa.html http://www.nodc.noaa.gov/OC5/WOA98F/woaf_cd/search.html http://www.nodc.noaa.gov/cgi-bin/JOPI/jopi http://ingrid.ldeo.columbia.edu/SOURCES/.LEVITUS94 http://www.meds-sdmm.dfo-mpo.gc.ca/ http://www.nodc.noaa.gov/GTSPP/gtspp-home.html http://www.ios.bc.ca/ios/osap/data/ ESTAR 72 http://www.esr.org/ssiwg3/SSIWG_3.html GODAE http://www.bom.gov.au/bmrc/ocean/GODAE/ HYDROS http://hydros.gsfc.nasa.gov/ Kuiper Airborne Observatory Publications: http://sofia.arc.nasa.gov/project/library/kao/pub72.html Microwave Geophysics Group, University of Michigan: http://www.eecs.umich.edu/grs/intro.html MIRAS http://www.estec.esa.nl/ap2000/abstracts/a00565.html http://www.esa.int/est/prod/prod0157.htm http://www.esa.int/est/prod/prod0662.htm http://esapub.esrin.esa.it/bulletin/bullet92/b92marti.htm PALS (Passive/Active L/S-band Airborne Sensor) http://eis.jpl.nasa.gov/msh/mission+exp/pals.html Project ARGO http://argo.jcommops.org/index.html Salinity Sea Ice Working Group: http://www.esr.org/ SLFMR (Scanning Low Frequency Microwave Radiometer) http://www.csc.noaa.gov/crs/AECS/SLFMR/ SMOS and MIRAS http://www.esa.int/export/esaLP/smos.html http://www.cesbio.ups-tlse.fr/indexsmos.html www.esa.int/export/esaLP/ESAQGA2VMOC_smos_0.html SMOS and the US missions AQUARIUS and HYDROS www.knmi.nl/~meulenvd/esa/DOSTAG/ESA_PB-EO_2002_2.pdf STAR-Light (Scanning Thinned Array Radiometer-Light) http://www.eecs.umich.edu/grs/projects/starlight/index.html 73 9.3 Correspondence Researchers involved in sea ice and sea surface salinity mapping were contacted through electronic mail. The information obtained is primarily about the status of recent airborne deployments, and the availability of very low frequency radiometric data on sea ice. From Dr. E.G. Njoku, JPL/NASA Dear Ms. Simms: I don't know of much work that has been carried out at 1.4 GHz in looking at sea ice. There have not been many (if any) L-band airborne radiometer flights over ice. Seaice researchers who may be able to provide more information are Don Cavalieri or Joey Comiso, both at Goddard Space Flight Center, or Frank Carsey at JPL. Regards, Eni Njoku Eni G. Njoku M/S 300-233, Jet Propulsion Laboratory 4800 Oak Grove Drive, Pasadena, CA 91109 Tel:(818) 354-3693; Fax:(818) 354-9476 E-mail: eni.g.njoku@jpl.nasa.gov From T. England, University of Michigan Dear Dr. Simms, STAR-Light is an airborne radiometer we are currently building for use in the Arctic. It won't be operational for 2-3 years. Our work in the arctic so far has been from a tower over tundra and at the SSM/I frequencies of 19, 37, and 85 GHz. I wish I could help with L-band data but I can't yet. Please ask again in a couple of years. The cold regions folks at Goddard might have some L-band ice data. Also, I vaguely remember that Edgerton at Aerojet looked at sea ice at L-band back in the late '60's or early 70's. Their reports were all in the gray literature but well done. I would look in my files for you but we recently moved them and they are not in a configuration that I can reasonably access. Sincerely, Tony England Microwave Geophysics Group c/o Prof. A.W. England University of Michigan EECS Building Room 3120 1301 Beal Ave 74 Ann Arbor, MI 48109-2122 From Dr. Per Gloersen, NASA/GSFC Dear Dr. Simms, I have researched our compatriots at Wallops Flight Center on your behalf and have uncovered the following: 1) The WFC P-3 was out of commission, but a Navy P-3 was obtained as a substitute, and flew some missions out of Thule in July 2000, carrying, alternately, C- and L-band radiometers. 2) Dr. Don Cavalieri (cc'd on this message) was the mission scientist. 3) Flights dedicated to studying thin and water-covered sea ice were carried out in Baffin Bay only. There were no flights over the Arctic Ocean, to the best of my knowledge. 4) Dr. Cavalieri has just received copies of the raw data, and is a long way from making them available for distribution. I hope this helps. Sincerely, Dr. Per Gloersen/Code 971 NASA/Goddard Space Flight Center Greenbelt, MD 20771-0001 per.gloersen@gsfc.nasa.gov 301-614-5710; FAX: 301-614-5644 75