New Estimation and Study of Freshwater Forcing at the air-sea interface Balla Maggero Marine & Oceanography Services. Ngong road 30259 Nairobi. e-mail: nambuye@yahoo.com Abstract: The retrieval and application of sea-surface primary diagnostic variables with rainfall from space-borne passive microwave and infrared sensors has grown upstream. These products are necessary for the calculation of turbulent heat fluxes (evaporation) and the water-budget at the air-sea interface. The newly launched AMSR-E and MODIS instruments on AQUA platiform gives good complementary observations for the derivation of these basic state variables and rainfall. This article reviews concisely, the latest algorithm development for retrieval of these variables from AMSR-E/MODIS combination for the purpose of studying freshwater forcing at the sea surface. Keywords: OCGM, thermohaline circulation, data assimilation, sea-surface forcing 1 Introduction The sources and sinks of freshwater at the sea surface are crucial component of the global water balance. The ocean losses about 10% more freshwater through evaporation than it gains through precipitation (Baumgartner and Reichel, 1975). The remaining 10% is contributed by river run-off, with the change in storage of freshwater in the oceans being a much smaller residual. The surface freshwater flux determines upper ocean mixing and the thermohaline component of ocean circulation. The stabilizing effect of a local net freshwater surface input limits the efficiency of ocean-atmosphere communication: it impedes heat exchange with subsurface layers, nutrient transport to the upper ocean, and trace gas exchange with the atmosphere. In the atmosphere these fluxes determine atmospheric moisture content and latent heat that drive tropospheric convection, horizontal advection and precipitation patterns. Surface freshwater forcing is believed to play a primary role in setting the background state and timing of the large interranual phenomena of ENSO. Theory suggests that equatorial upper-ocean heat content and mixed-layer depth are crucial in setting ENSO time scales (Zebiak and Cane, 1987). Over the western Pacific warm pool region, the upper ocean mixed layer is observed to very shallow, on the order of 30m, apparently due to a combination of stabilization by precipitation and highly intermittent wind forcing (Lukas and Lindstrom, 1991). Long-term observations that permit the accurate estimation of precipitation, evaporation and wind stress are therefore likely to be important for an understanding of ENSO, and for the development of a quantitative ENSO prediction capability. Satellite observations, as it will be shown later, offer the unique opportunity to sample large oceanic areas with high temporal resolution. The accuracy of satellite estimates, however, have to be compared against in situ observations so that reliable estimates are assessed. The theme of this review is the retrieval algorithm for evaporation basic state variables together with precipitation from the Advanced Microwave Scanning Radiometer (AMSR) and the Moderate Resolution Spectroradiometer (MODIS), (currently on-board AQUA mission). The problem of modeling thermohaline circulation and ocean data assimilation are briefly introduced. 2. The AMSR and the MODIS The AMSR is a twelve-channel, six frequency, total power passive-microwave radiometer system. It measures TB at 6.9, 10.7, 18.7, 23.8, 36.5 and 89.0 GHz. Vertically and horizontal polarized measurements are taken at all channels. The earth-emitted microwave radiation is collected by an off-set parabolic reflector 1.6m in diameter that scans across the Earth along an imaginary conical surface, maintaining a constant Earth incidence angle of 550 and proving a swath width of 1445km. The reflector focuses radiation into an array of six feedhorns which then carry radiation to radiometers for measurement. Calibration is accomplished with observations of the cosmic background radiation and an on-board warm target. Spatial resolution of the individual measurements varies from 5.4km at 89GHz to 56km at 6.9GHz. The MODIS instrument employs a conventional imagingradiometer concept, consisting of a cross-track scan mirror and collecting optics, and a set of linear detector arrays with spectral interference filters located in four focal planes. The optical arrangement provides imagery in 36 discrete bands from 0.4 to 15.5m, selected for diagnostic significance in Earth science. The spectral bands have spatial resolutions of 250m, 500m, or 1km at nadir; signal-to-noise rations of greater than 500 at 1km resolution (at a solar zenith angle of 700; and absolute irradiance accuracies of 5 percent from 0.4 to 3m (2 percent relative to the sun) and 1 percent or better in the thermal infrared (3 to 14.5 m). MODIS provides daylight reflection and day/night emission spectral imaging of any point on the Earth at least every two days, operating continuously. 3.0 Data Set Evolution and Retrieval Schemes The retrieval algorithms for the freshwater elements are based on experience gained from previous sensors; SMMI, SSM/I, TMI and AVHRR. 3.1 Precipitation The AMSR rainfall algorithm is a slight modification of the TMI and SSM/I algorithm (Wilheit at al., 1991)., The information is obtained by matching the observed brightness temperatures to database of a priori cloud profiles derived from a variety of sources including cloud dynamical models as well as ground-based radars. The a priori is then used in Bayesian inversion scheme to derive a linear combination of profiles that most resembles the observed brightness temperatures. Despite differences, the land and ocean algorithms have been combined into a single framework to insure communication between the algorithm components. The level 3 algorithm sacrifices some accuracy of the instantaneous retrieval in order to minimise assumptions that might contaminate a climate record of rainfall. The frequency of occurrence of rain intensities in different rate categories can be plotted as a histogram or a smoothed curve that fits the histogram. 3.2 Evaporation Sea Surface Temperature determination is based on MODIS-calibrated mid- and far-IR radiances, using an algorithm that exploits the differences in atmospheric transimisivity in the different IR bands to enable highly accurate estimation of the atmospheric effects, thereby enabling ancillary input to the algorithm along with a land mask, which is used to mark non water pixels while an ice extent mask limits polar sea coverage. A sequence of spatial and temporal homogeneity tests is applied to validate the quality of cloud-free observations. Sea-surface wind speeds: wind speed is retrieved using non-linear approach from AMSR brightness temperatures. The neural network provides the most accurate estimates because of the unique relationship with TB. This relationship is given by the radiative transfer equation for atmosphere bounded at the bottom by a rough sea surface. The radiative transfer equation can be approximated by a relatively simple closed-form expression, which is called the TB model function ( Wentz, 1983). It is accomplished by varying the wind speed until the TB model function matches the observations. In order to obtain the highest possible accuracy, the model must include the effects of all other relevant parameters in the radiative transfer equation, and the TB model must be precisely calibrated. Air Specific Humidity: Over the ocean, near surface specific humidity is a direct function of the water vapor column of the planetary boundary layer (Schulz et al., 1993). Therefore, air specific humidity is retrievable. A linear scheme employing 10.65, 18.7, 23.8, and 36.5 GHz TB at both polarisation is used. 4.0 Thermohaline Modelling The development of a generalised vertical-coordinate ocean model include the same physics, such as the turbulance-closure model (which predicts the vertical diffusion of momentum, temperature and salinity), as in the Princeton Ocean Model (Blumberg and Mellor, 1987). The latter model uses sigma-coordinates in the vertical where the model layers follow the terrain and require considerable smoothing of the topography with decreasing resolution. Thus, modifications are necessary to accommodate some detail in topography and a course-resolution grid but retaining some boundary-layer physics, which is important for deep flows; thermohaline circulation. Inclusion of flexible vertical-layer structure could also permit the use of density surfaces to define layers because the layers have to be updated every time step. The freshwater cycling in the ocean is studied first in a diagnostic framework and later on in quasi-equilibrium runs for a specified freshwater forcing. Given the equilibration time scale of thousands of years for the oceanic thermohaline circulation, even the 2X2 degree model can be too expensive to run. However, tendencies in the model behaviour can be equally illuminating of the physics and adequate for the study of the sensitivity of the interbasin export from the Pacific to the Atlantic via the Arctic Ocean. 5.0 Data Assimilation The OCGM are among the most demanding tasks performed by computers today. Even when the fastest computers are used, these models and the solutions to the equations are only approximations. One way to correct for errors in the model solution is to apply a procedure known as “data assimilation”, which blends the approximations with observations of the real ocean in a least-square error sense. The blending takes account of errors in the model dynamics and the thermodynamics, as well as the measurement errors in the observations. The outcome is better than the analysis using either a model or observation only. Weather forecasting utilise mathematical models that encapsulates our understanding of the evolution of the various components of the system and the interactions between each other. The essential pre-requisite for prediction is best estimate of the initial state of the system, encoded as numbers on a spatial grid which may have little resemblance to the spatial pattern of available observations for a system whose evolution exhibits sensitive dependence on initial conditions (or chaos), such as the ocean. Several methods exist for assimilating observations into a numerical models. Two main approaches are usually seen in the techniques following either the optimal control theory or the statistical estimation theory (Ghil and Manalotte-Rizzoli, 1991). Seasonal climate forecasting is a revolutionary technology in meteorology and oceanography aimed at predicting long-range fluctuation in climate (as opposed to day-to-day weather events) several months in advance. This endeavour draws heavily on a wide range of satellite observations which are synthesised via coupled atmosphere-ocean models. The scientific basis for longterm climate prediction lies in the ocean, which acts as a ‘ pacemaker’ by slowly nudging the atmosphere into a pontentially predictable mean state. In 1997, the severe ElNino was detected six months earlier by various satellites. The anomalous warming of the ocean surface in the equatorial Eastern Pacific was predicted. A data assimilation system fed this results downwards into the ocean interior component of a coupled model. The model in turn successively predicted the onset of the event. Since satellites observe only the surface, this capability is vital to probing the interior processes such as the thermohaline circulation. Above images were captured by SSM/I and AVHRR flying on DMSP and the NOAA satellites (courtesy of NASA's Earth Science Enterprise and its Earth Observing System) The methods discussed are the brightness temperature model for wind speed, the split-window approach for the SST, the Bayessian inversion scheme for precipitation, and the retrieval of air specific humidity using a function that correlates with the water vapor column of the planetary boundary layer. Although the spatial resolution along the tracks is good there are large gaps between them, and only tantalizing glimpses of the split are provided by the basic observations on each day. Coupled atmosphere-ocean models have been used for forecasting warm events. When numerical prediction models can enhance the value of observations, data assimilation fuses disparate data streams into a coherent picture of the evolving ocean. By routinely confronting these models with observations, data assimilation facilitates rapid advance to simulate the profoundly complex, highly non-linear system, a capability yet to be exploited in climate modelling. It provides the essential scientific basis for using models to predict the system evolution. Future improvements are expected from a new generation of instrument-suite to be in space in the near future. The utilization of the planned Dualfrequency Precipitation Radar (DPR) together with the GPM Microwave Imager (GMI) both on Global Precipitation Mission (GPM) and expected dense sampling from constellation satellites will be a great stride in the measurements and study of freshwater fluxes. Acknowledgment: The author is grateful to two reviewers for their fruitful and constructive comments. The Information and data used in this work include those produced through the funding of NASA’s Earth Science Enterprise and the Earth Observing Systems Data and Information System Archive at GSFC, Greenbelt Maryland. References: 6.0 Summery The distribution of freshwater from sea-ice, river runoff and precipitation minus evaporation sources is of importance to the thermohaline circulation by modifying the surface-water masses and mixing processes. This variability affects deep waters as well as mode water, which can have a climate impact on decadal scale because they will come into contact with the atmosphere much faster than the deep waters. It is clear that observations of freshwater state variables from most recent advanced remote sensors pose special advantages. Though still under research missions, the AMSR and MODIS are superior in their respective class. This article has presented some retrieval techniques for the basic inputs for the ocean freshwater forcing. Baumgartner, A and E. Reichel, The world Water Balance: Mean Annual Global, Continental and Maritime Precipitation, Evaporation and Runoff. Elsevier, New York, 179, 1975. Blumberg, A.L., and G.L. Mellor: A description of a three-dimensional coastal circulation model, Three Dimensional Coastal Models (ed. N.S. Heaps), 1 – 16, Am. Geophys. Union. Wash. DC. 1987. Ghil , M. and P. Manalotte-Rizzoli: Data Assimilation in Meteorology and Oceanography. Ad. Gephys. 23, 141 –265, 1991. Lukas and Lindstrom, The mixed layer of the Western Equatorial Pacific Ocean, J. Geophys. Res. 96, 3343-3357,1991 Schulz, J., P. Schluessel and H. Grassii : Water vapor in the Atmospheric Boundary Layer over the Oceans from SSM/I Measurements. Int. J. Remote Sens. 14, 2773-2789. 1993. Wentz F.J. A Well Calibrated Ocean Algorithm for SSM/I. J. Geophys. Res. 102,8703-8718, 1997. Wilheit, T.T., A.T.C. Chang and L.S. Chui : Retrieval of Monthly Rainfall indices from Microwave Radiometric Measurements using Probability Distribution Function, J. Atmos. Oceanic Technology, 8, 118-136, 1991. Zebiak, S. A. and M. A. Cane, A model El Nino-Southern Oscillation. Mon. Wea. Rev., 115, 2262-2278. 1987.