Journal of Oceanography, Vol. 58, pp. 213 to 225, 2002
Japanese Ocean Flux Data Sets with Use of Remote
Sensing Observations (J-OFURO)
MASAHISA KUBOTA1*, N AOTO IWASAKA2 , SHOICHI KIZU 3, M ASANORI KONDA 4
and KUNIO K UTSUWADA 1
1
School of Marine Science and Technology, Tokai University,
Orido, Shimizu, Shizuoka 424-8610, Japan
2
Tokyo University of Mercantile Marine, Etchujima, Koto-ku, Tokyo 135-8533, Japan
3
Department of Geophysics, Graduate School of Science, Tohoku University,
Sendai 980-8578, Japan
4
Department of Geophysics, Graduate School of Science, Kyoto University,
Kitashirakawa-Oiwake, Sakyo-ku, Kyoto 606-8502, Japan
(Received 1 May 2001; in revised form 29 October 2001; accepted 29 October 2001)
We have constructed ocean surface data sets using mainly satellite data and called
them Japanese Ocean Flux data sets with Use of Remote sensing Observations (JOFURO). The data sets include shortwave radiation, longwave radiation, latent heat
flux, sensible heat flux, and momentum flux etc. This article introduces J-OFURO
and compares it with other global flux data sets such as European Centre for Medium
Range Weather Forecasting (ECMWF) and National Center for Environmental Prediction (NCEP) reanalysis data and da Silva et al. (1994). The usual ECMWF data are
used for comparison of zonal wind. The comparison is carried out for a meridional
profile along the dateline for January and July 1993. Although the overall spatial
variation is common for all the products, there is a large difference between them in
places. J-OFURO shortwave radiation in July shows larger meridional contrast than
other data sets. On the other hand, J-OFURO underestimates longwave radiation
flux at low- and mid-latitudes in the Southern Hemisphere. J-OFURO latent heat
flux in January overestimates at 10 °N–20 °N and underestimates at 25 °N–40 °N. Finally, J-OFURO shows a larger oceanic net heat loss at 10 °N–20 °N and a smaller loss
north of 20 °N in January. The data of da Silva et al. in July show small net heat loss
around 20 °S and large gain around 20 °N, while the NCEP reanalysis (NRA) data
show the opposite. The da Silva et al. zonal wind speed overestimates at low-latitudes
in January, while ECMWF wind data seem to underestimate the easterlies.
1. Introduction
The ocean actively exchanges heat, water and momentum with the atmosphere through the ocean surface.
The exchanged heat, water and momentum are transported
by the general circulation in the ocean and atmosphere to
redress the heat imbalance. Since the exchange and transporting processes are essential components of the global
climate, it is quite important to estimate these fluxes between the atmosphere and the ocean and of transports by
ocean and atmospheric circulation to understand the
mechanism of the global climate. However, it is difficult
Keywords:
⋅ Remote sensing,
⋅ sea surface flux,
⋅ radiation,
⋅ turbulent flux.
to estimate those fluxes and transports globally using insitu observation data from ships and buoys, since such
data are extremely sparse in time and space. By contrast,
we can obtain extremely homogeneous data with high
resolution using analysis and satellite data.
The heat flux between the ocean and the atmosphere
consists of four components, viz., shortwave radiation,
longwave radiation, latent heat and sensible heat fluxes.
Generally only shortwave radiation transfers heat from
the atmosphere to the ocean, while the remaining three
components transfer heat from the ocean to the atmosphere. There are broadly three methods to estimate the
surface fluxes at large scales. The first is a traditional
method using empirical formulae to estimate the flux from
surface meteorological observations such as air temperature, humidity and cloudiness. Iwasaka and Hanawa
* Corresponding author. E-mail: kubota@mercury.oi.utokai.ac.jp
Copyright © The Oceanographic Society of Japan.
213
(1990), Esbensen and Kushnir (1981) and Oberhuber
(1988) are examples of the estimation of heat fluxes using the empirical formulae and ocean observation data
from ships and buoys. The heavily biased distribution of
the observation data is a serious problem when constructing a global heat flux data set, even if each observed variable is accurate. Moreover, the empirical method is too
simple to precisely evaluate the flux. Kizu (1998) pointed
out that the empirical estimation of shortwave radiation
suffers from systematic error in the empirical formula
used. In-situ data will thus be used to validate satellite
data but not to construct a global data set without combining them with satellite data, although they are still
useful to construct a heat flux field for a long period of
about 50 years in the Northern Hemisphere and more than
100 years in certain area of the main shipping routes.
The second method for estimating global surface
fluxes is to use satellite data. Estimating shortwave radiation by satellite data has become popular in radiation
budget research. For example, the global radiation budget
was estimated in the International Satellite Cloud Climatology Project (ISCCP) and the Earth Radiation Budget
Experiment (ERBE). The ERBE data have been collected
since 1984 from three satellites, the Earth Radiation
Budget Satellite (ERBS), NOAA-9, and NOAA-10, carrying scanned and non-scanned ERBE instruments. The
goal was to measure global albedo, fluxes, and solar incidence. NOAA-9 and NOAA-10 provide global coverage, and ERBS provides coverage between 67.5 degrees
north and south latitude. On the other hand, ISCCP was
established in 1982 as part of the World Climate Research
Program (WCRP) to collect radiance data measured by
weather satellites and to analyze them to infer global distribution, properties, and diurnal, seasonal and interannual
variations of clouds. The resulting data sets and analysis
products are being used to study the role of clouds in climate, radiative energy exchanges, and the global water
cycle. Moreover, the utility of satellite observation for
tempo-spatial monitoring of insolation in the western
Pacific has been shown using data from the Japanese satellite, the Geostationary Meteorological Satellite (GMS)
(Nunez, 1988; Weymouth and Marshall, 1994; Chou et
al., 2001).
The longwave radiation flux at the sea surface has
received less attention than the other components. One
reason may be the difficulty in observing the longwave
radiation flux due to the complicated formulae used in
the estimation; furthermore, there are few reliable observation stations for the longwave radiation flux in the
ocean. One way of estimating the longwave radiation flux
at the sea surface is using satellite remote sensing data
and applying a radiation-transfer model to estimate the
flux. In most of the studies of this kind, empirical formulae that relate satellite observations to the outputs of so-
214
M. Kubota et al.
phisticated radiation-transfer models are used to obtain
longwave radiation flux from satellite data; for example,
Liu et al. (1997) used Special Sensor Microwave/Imager
(SSM/I) data, and Gupta (1989) used TIROS Operational
Vertical Sounder (TOVS) data. Other investigators, such
as Dr. W. B. Rossow and his group (e.g., Rossow and
Zhang, 1995) utilized the ISCCP data. When using data
from either ship or satellite, another problem that there is
almost no reliable sea truth except for some islands and
tentative ship measurements. This makes it difficult to
evaluate the accuracy of the estimated flux by each
method.
Latent and sensible heat fluxes are globally estimated
using a bulk formula for satellite data, which includes
several kinds of physical variable, and the estimation of
turbulent heat fluxes is complex. For this estimate, wind
speed, specific humidity, and saturated specific humidity
are necessary for latent heat, and wind speed, air temperature, and sea surface temperature (SST) are needed
for sensible heat. In these necessary variables, specific
humidity and air temperature are not easily obtained from
satellite data compared with other data. Liu (1986), Schulz
et al. (1993), and Schlüssel et al. (1995) proposed an algorithm for estimating specific humidity from satellite
data. Some studies proposed a method for estimating the
surface level air temperature by satellite sensors (Kubota
and Shikauchi, 1995; Konda et al., 1996; Prihodko and
Groward, 1997; Gautier et al., 1998; Jones et al., 1999).
However, satellite-derived specific humidity and air temperature are not yet so accurate as to be satisfactory for
the estimation of latent and sensible heat fluxes. In particular, air temperature is one of the most important elements of the climate parameters. Monitoring the global
distribution of surface-level air temperature is important
not only for evaluating the strength of thermal coupling
between ocean and atmosphere but also for monitoring
the long-term global warming and impact of the change
of SST on it. The effort to improve the accuracy of satellite-derived air temperature should be encouraged.
The momentum flux through the ocean surface, i.e.
wind stress, is one of the essential external forces driving
the ocean’s motions. Wind measurements by ships and
buoys have been used to calculate surface wind stress for
the last two decades, but their reliability is questionable
due to such measurement errors as uncertain anemometer height (e.g., Pierson, 1990). As an alternative, a satellite microwave scatterometer supplies wind speed and
direction data with much better time and space resolution over the world ocean. The first on-board satellite
scatterometer, SEASAT, was launched in 1978. SEASAT
supplied the wind data only for about three months. After
about a decade the European Space Agency (ESA)
launched the first European Remote-sensing Satellite
(ERS-1) in 1991 and subsequently ERS-2 in 1996. The
Table 1. Summary of data sets in J-OFURO.
Active Microwave Instrument (AMI) on ERS-1, 2 has
provided global wind data about nine years since August
1991, allowing us to examine variability on multiple time
scales. Another attempt with the satellite called the Advanced Earth Observation Satellite (ADEOS) was undertaken cooperatively by the National Aeronautics and
Space Administration (NASA) in USA and the National
Space Development Agency (NASDA) in Japan. Earth
observation by ADEOS unfortunately stopped 11 months
after launch due to an accident with the solar panel. As
compensation for this accident, NASA launched the
Quikscat (QSCAT) in June 1999 on board the new
scatterometer sensor SeaWinds. NASA scatterometer
(NSCAT) on ADEOS and QSCAT/SeaWinds have better
time and space resolution than the ERS-1, 2. Yet ERS-1,
2 provided us with global wind data for a long time, and
the global data set from ERS-1, 2 are used in our data set
introduced below.
The third method is to use Numerical Weather Prediction (NWP) data, which are outputs from an atmospheric general circulation model (AGCM). However,
NWP products are not pure outputs from AGCM because
in-situ data are assimilated into the AGCM. NWP products are therefore often called analysis data. NWP products provide us global heat flux data as well as satellite
data. Products of the European Centre for Medium Range
Weather Forecasting (ECMWF) and the National Center
for Environmental Prediction (NCEP) are typical examples of NWP products. It is an advantage that NWP products provide us homogeneous grid data, but the characteristics of the products are not continuous, because the
model configuration is frequently renewed. Moreover,
NWP products suffer from a heavily biased distribution
of in-situ observation, since a lot of in-situ data are assimilated into NWP models. This has led to the reanalysis
of the NWP products in many recent studies. The NWP
products are not independent of satellite data in some
cases because some satellite data have already been assimilated into the NWP models. For example, ECMWF
has been assimilating surface winds observed by ERS
scatterometer since January 1996 and SSM/I since Octo-
ber 1999.
Recently we constructed ocean surface flux data sets
using mainly satellite data. The data set, called Japanese
Ocean Flux data sets with Use of Remote sensing Observations (J-OFURO), includes radiation flux, turbulent heat
flux, momentum flux, sea surface dynamic topography,
geostrophic currents, and air temperature. This is the
world’s first global data set of heat fluxes (total and each
component) and momentum flux. The data can be obtained
from our website (http://dtsv.scc.u-tokai.ac.jp). Basic information for each data set is shown in Table 1. In the
present study we introduce each data set of J-OFURO,
except for sea surface dynamic topography and
geostrophic currents, which are not directly related to
surface flux. Data and methods are described in Section
2. Section 3 shows average fields for each heat flux and
total heat flux. The results of comparison between JOFURO and the ECMWF and NCEP/NCAR reanalysis
data are also given in order to clarify the characteristics
of J-OFURO data. A summary and discussion are provided in Section 4.
2. Data and Methods
2.1 Radiation data
Kizu (1999) produced a monthly data set of surfacelevel downward shortwave radiation fluxes (insolation)
from October 1992 to September 1993 with 1° × 1° latitude-longitude resolution. This product was extended to
more than twelve years from March 1987 to September
1999 by Kizu (2001b), and was included in J-OFURO. A
monthly data set of longwave radiation fluxes from October 1992 to September 1993 was produced and will be
extended in the near future. The data sets of the shortwave
and longwave radiation cover the region between 60°N
and 60°S, 80°E and 160°W in the eastern Indian Ocean
and the western and central Pacific Ocean, because we
use only the GMS data for this estimation. Although the
basic spatial and temporal resolutions of the radiation data
are 0.25° by 0.25° and 1 day, respectively, J-OFURO currently provides monthly radiation data with 1° × 1°.
Introduction of J-OFURO
215
2.1.1 Shortwave radiation
A spectrally-integrated model by Iqbal (1983; his
Model A) was used to formulate fundamental radiative
transfer processes in the atmosphere. Scattering by air
molecules and aerosol, absorption by ozone and water
vapor were modeled using simple formulas and
parameterization published in the literature. Primary data
used in the estimation are 3-hourly visible and infrared
histogram data from the Visible and Infrared Spin Scan
Radiometer (VISSR) on the GMS with 0.25° spatial resolution, monthly precipitable water from SSM/I (Wentz,
1994), monthly total ozone amount from the Total Ozone
Mapping Spectrometer (TOMS) on the Nimbus 7 satellite, and seasonal climatology of tropospheric aerosol
from Advanced Very High Resolution Radiometer
(AVHRR) on the NOAA satellite (Husar et al., 1997).
Long-term degradation of the VISSR visible channel was
corrected by Kizu (2001a). Based on comparison with
routine ground measurements of insolation mostly in Japan and Australia, the statistical error of daily and monthly
mean insolation values was estimated to be less than 20
% and 10 %, respectively. Mainly because of the simplification of the highly complex relation between cloud and
radiation, the estimation error partly depends on the illuminating and viewing geometry and cloud type. The estimation error over the ocean has not been thoroughly investigated due to a severe lack of sea truth. However, similar precision is expected over the ocean as well, since the
primary radiative transfer processes in the atmosphere are
common over land and ocean. The methodology and its
validation results are presented in detail by Kizu (2001b).
2.1.2 Longwave radiation
In order to estimate the longwave radiation flux at
ocean surface from GMS Infrared (IR) data, we first detected cloud area in the GMS image as follows: a pseudoSST was computed from the infrared data in each pixel
and compared with the MCSST. A pixel having a pseudoSST lower than Multi-Channel Sea Surface Temperature
(MCSST) by more than 3.0K was assumed to be a cloud
pixel. Unrealistic values of the MCSSTs were corrected,
and grid points with no data were filled by interpolating
Iwasaka and Tanimoto’s climatology (Iwasaka and
Hanawa, 1990; Tanimoto, 1993). For the cloud pixel, the
temperature in the pixel was assumed to be that at the top
of the cloud. The cloud top height was computed from
the cloud top temperature and the temperature profile in
the pixel inferred from the NCEP reanalysis data. The
cloud bottom height was determined by assuming cloud
thickness. Here, we do not deal with multi-layer cloud
because there are currently no practical ways to detect
multi-layer clouds from satellite observation so far. Cloud
thickness in the model was assumed to be 500 m and does
not vary in space and time.
Downward longwave radiation was then calculated
216
M. Kubota et al.
by applying the black body radiation theory assuming the
emissivity of the cloud bottom to be 0.9. Atmospheric
absorption and emission of the longwave radiation flux
between the cloud bottom and the ocean surface were
assessed by a narrow band model of radiation and based
on NCEP data. The atmospheric layer was divided into
seven sub-layers between the ocean surface and 300 hPa.
The narrow band model proposed by Goody (1964) was
used to compute the emission and absorption of the atmosphere. The H2 O rotation band, the H2O continuum,
the CO 2 15 µ m band and the H2O 6.3 µm band were included in the radiation model. The parameters in the model
were obtained from Rogers and Walshaw (1966),
Goldman and Kyle (1968) and Roberts et al. (1976). The
spectral range is 0–2200 cm –1, and was divided into 23
spectral bands for the radiation computation. Atmospheric
radiation below 300 hPa in a cloud-free area was computed using the narrow band model and NCEP data. Upward longwave radiation was computed following the
black body theory with emissivity at the ocean surface of
0.984 given by Konda et al. (1994).
Radiation measurement data at Kwajalein Island located at 9.25°N, 167.5°E was used as a reference to evaluate the error in the estimation of longwave radiation computed in this study. The data are recorded every three
minutes at an altitude of 10 m without high surrounding
terrain. This observation station is operated by the Climate Monitoring and Diagnostics Laboratory (CMDL) of
the National Oceanic and Atmospheric Administration
(NOAA). The data were obtained through the Baseline
Surface Radiation Network (BSRN, http://
bsrn.ethz.ch/). The downward longwave radiation tends
to be overestimated under cloudy conditions and to be
underestimated under clear sky. Bias of the estimated
downward radiation is about 10–20 Wm–2 for monthly
mean values and its random errors are comparable to the
magnitude of the bias. The annual mean value of downward radiation is very close to the observation value at
Kwajalein. However, no quantitative assessment of the
estimation error of the flux has been done for the entire
region, because there is almost no reliable in-situ observation in the objective area except for Kwajalein Island.
2.2 Turbulent heat flux
The data set of turbulent heat flux covers the global
ocean. The turbulent heat flux was computed for the period from 1991 through 1995. Original spatial and temporal resolutions are 1° × 1° and three days, respectively.
It takes three days for the SSM/I to cover the whole globe.
At present, only monthly mean data (monthly averages
of the three-day mean) are provided on our web site.
A bulk formula provided by Kondo (1975) was used
in the present study to estimate latent heat flux, although
the stability condition was not considered in the calcula-
tion. Physical variables included in the formula are wind
speed, specific humidity, saturated specific humidity, atmospheric pressure, and air density. Since atmospheric
pressure and air density are less effective than the other
three variables, climatological monthly means in the Comprehensive Ocean-Atmosphere Data Set (COADS) were
used for these two variables in this calculation. Saturated
specific humidity was simply derived from SST. Data
coverage of SST observed by an infrared radiometer such
as NOAA/AVHRR is generally incomplete due to clouds.
We therefore used gridded SST values provided by NCEP,
which are made by blending satellite-based and in-situ
SST data on a 1° × 1° grid at weekly intervals (Reynolds
and Smith, 1994, 1995).
Wind speed can be observed by three kinds of satellite sensor: microwave scatterometer, microwave radiometer, and microwave altimeter. We constructed grid data
using wind speed values included in SSM/I geophysical
data (Wentz, 1994), because specific humidity was also
estimated from SSM/I data. On the other hand, there are
several methods to derive specific humidity from satellite data (Liu, 1986; Schulz et al., 1993; Schlüssel et al.,
1995). Kano and Kubota (2000) compared the results
derived from these three algorithms with in-situ data and
concluded that Schlüssel et al.’s method provides the best
accuracy. The algorithm proposed by Schlüssel et al.
(1995) was therefore used to compute specific humidity
data using brightness temperature data observed by
SSM/I. Kano and Kubota (2000) compared J-OFURO latent heat flux data with in-situ data observed by the Japan Meteorological Agency (JMA) and with the Tropical
Ocean Atmosphere (TAO) array in the Tropical Ocean
and Global Atmosphere (TOGA). They showed that the
root-mean-square (RMS) error is about 40 Wm–2 for threeday means and 20 Wm–2 for monthly means.
We also used a bulk formula by Kondo (1975) to
estimate sensible heat flux, which requires data of wind
speed, SST and air temperature. Several authors have proposed some methods for calculating air temperature using satellite data (e.g., Kubota and Shikauchi, 1995;
Konda et al., 1996). However, the accuracy of air temperature derived from satellite data is inadequate for calculating sensible heat flux, as mentioned above. Kubota
and Mitsumori (1997) proposed a method to calculate
sensible heat flux by multiplying latent heat flux by the
Bowen ratio. They compared results by abovementioned
methods with in-situ data in the western North Pacific
and proved the effectiveness using the Bowen ratio. We
therefore adopted this method in this study. Using satellite data, we can estimate the latent heat flux but not the
Bowen ratio, and therefore we used climatological means
of the Bowen ratio derived from ECMWF data. The quality of the Bowen ratio in the polar regions is not very
good, and data lying outside the range of average ± stand-
ard deviation are not used to estimate the Bowen ratio.
The variability of the Bowen ratio is not very large, except in the western boundary current regions and the Polar Regions. The RMS errors are about 5 Wm–2 for threeday means in most regions.
2.3 Momentum flux
We constructed a data set of gridded surface wind/
wind-stress vectors using a data set of wind speed and
direction called Level 2.0, which were obtained from the
spaceborne scatterometer of ERS-1, 2. We used a CDROM database of the off-line products which has been
constructed by the Institut Français de Recherche pour
l’Exploitation de la Mer (IFREMER). The extraction of
surface wind vector from measured backscatter values
produces 2 to 4 ambiguity in wind direction at each measuring point. A procedure for selecting more reliable wind
direction values, called ambiguity removal, gives ranked
solutions in wind speed and direction. This process was
carried out for the IFREMER product of wind speed vectors greater than 3 m s–1. In calculations of our product,
we used the first ranked values in wind speed as well as
wind direction. This statistical reliability was described
in detail by Ebuchi et al. (1996) and Graver et al. (1996)
who examined the reliability of the IFREMER and the
Jet Propulsion Laboratory (JPL) products by comparison
with numerous oceanic buoy measurements.
The ERS/AMI data have an inhomogeneous distribution in space depending on the orbital motion of the
satellite, since the swath is only 500 km. In our construction of daily surface wind/wind-stress vectors with high
spatial resolution of 1° × 1° grid over the Pacific and Indian Oceans, 30°E–70°W, 60°S–60°N, we adopted an
averaging method using a weighting function varying with
time and space. This method was originally proposed by
Levy and Brown (1986) and was described in detail by
Kutsuwada (1998). We used radii of influence of 300 km
and 150 km in the zonal and meridional directions, respectively, for the NSCAT product (Kutsuwada, 1998) and
600 km and 300 km for our ERS-1, 2 products. The radius of influence is related to the time resolution of the
products and spatial coverage of the original wind data
obtained from the scatterometer. We calculated windstress vectors using the conventional bulk formula. We
adopted the drag coefficient depending only on wind
speed based on Large and Pond (1981). Wind/wind-stress
vectors on each grid point were averaged over a month
and three 10-day periods. Note that the last 10-day product in each month is not always the average over 10 days,
namely, over 10, 11 or 8–9 days.
We compared our products with in-situ measurements
at numerous oceanic buoys in the Pacific. Many of the
buoys are in the TAO array located in the equatorial Pacific. Results for NSCAT and ERS-1, 2 products were
Introduction of J-OFURO
217
documented in Kutsuwada (1998) and Kutsuwada and
Kazama (2000), respectively. They revealed that the RMS
differences do not exceed 2.0 m s–1 and 20° for wind speed
and wind direction, respectively, corresponding to the
accuracy of normal wind sensors. We can consider that
our products have sufficiently high reliability in the tropical region. On the other hand, the reliability in the areas
at mid and high latitudes may not be significantly high,
although verification is difficult due to the scarcity of insitu data in those areas. This problem will hopefully be
improved by future studies.
2.4 Air temperature
The method used to obtain the near-surface air temperature was based on Konda et al. (1996), who established the relationship between boundary layer parameters through the implicit equation of the Bowen ratio. In
this method we need specific humidity data, which can
be derived from microwave radiometer data. The empirical method to obtain air temperature established by
Schlüssel et al. (1995) was used in J-OFURO, rather than
the method of Liu (1986) used in Konda et al. (1996).
Wind speed and SST data were obtained from the SSM/I
geophysical data and the NOAA Pathfinder SST adjusted
to the skin temperature, respectively.
The 1991–92 means of the difference of monthly
mean air temperature in J-OFURO from the in-situ observation are 1.5 ± 1.4°C, 1.0 ± 3.0°C, 0.5 ± 2.1°C for
comparison with TOGA-TAO arrays, JMA buoys, and
NDBC buoys, respectively. We find that the estimated air
temperature tends to be higher than that observed, independent of the geographical condition. The difference
from observation is attributable to the estimation errors
of the surface air humidity as described by Konda et al.
(1996). The accuracy of the satellite-derived air temperature is very sensitive to that of the elementary physical
parameters such as SST, humidity, and wind speed.
3. Results and Intercomparison
Figure 1 shows average fields from October 1992 to
September 1993 for (a) latent heat flux, (b) sensible heat
flux, (c) shortwave radiation, (d) net longwave radiation,
and (e) total heat flux. Positive values mean heat transfer
from atmosphere to ocean. We can see a large latent heat
loss from the ocean, more than 150 Wm –2 in the subtropical regions, while latent heat loss is small in the equatorial regions and at high latitudes. Large latent heat loss
can also be found over the western boundary current, such
as the Kuroshio, even in the subtropics. Sensible heat flux
is generally quite small and negligible compared with
latent heat flux. However, sensible heat loss over the
western boundary current and at high latitudes in the
Northern Hemisphere is not negligible. The insolation
pattern is nearly zonal and the maximum insolation is
218
M. Kubota et al.
located along the zonal subtropical high centered approximately at 20°N. In the equatorial region, insolation
minima are discernable along the Intertropical Convergence Zone (ITCZ) and the Southern Pacific Convergence
Zone (SPCZ), leaving an insolation maximum along the
equator. Net longwave radiation flux is large in the western North Pacific, especially around Japan and in the central North Pacific, although it is not so large in other regions compared with shortwave radiation and latent heat
flux. It should be noted that the asymmetry between the
Northern and Southern Hemispheres is remarkable for
shortwave and net longwave radiation fluxes. Finally, the
largest ocean heat gain is found in the equatorial region,
while largest heat loss is found over the Kuroshio and
Kuroshio Extension Regions.
It is not difficult to validate instantaneous values of
satellite-derived flux data by comparison with in-situ data
such as buoy data. However, the validation of grid data
like J-OFURO is very difficult, because it is impossible
to obtain a true value for each grid. Therefore,
intercomparison between each data set is very important
in order to make clear the differences in characteristics
between each data set. In this section, the data of radiation, turbulent heat, and momentum fluxes in J-OFURO
are basically compared with climatological data (da Silva
et al., 1994) and the NCEP reanalysis (NRA) product and
the ECMWF reanalysis (ERA) product. The advantage
of a reanalysis over operational NWP data assimilation is
that it offers several years of fields from an unchanging
system. We use usual ECMWF product for comparison
of momentum flux in the present study.
The meridional distribution at the dateline of monthly
mean insolation of J-OFURO is compared with that of
the data sets of da Silva et al. (1994), NRA, and ERA for
January and July 1993 (Figs. 2a and 2b). The data set of
da Silva et al. is based on Reed’s (1977) empirical formula modified with integral constraints on oceanic heat
transport. The NRA and ERA data are calculated using
numerical models without assimilation of radiation data.
The most significant difference in insolation data between
J-OFURO and the NWP reanalysis products is seen around
35°N in northern summer, July 1993, as the J-OFURO
data set shows much larger meridional variation than the
reanalysis products. The NRA data may underestimate
insolation in cloudless regions in the subtropical area
south of 35°N and may overestimate insolation in mostly
overcast regions in the subarctic area north of 35°N. The
data of da Silva et al. show similar values to J-OFURO
in the subtropical area but seem to overestimate insolation in the subarctic area. Values in the equatorial region
are different from one data set to another, perhaps reflecting large intraseasonal to interannual variability of insolation. In contrast, the insolation data south of 15°S in all
the data sets are quite coincident. Differences among the
Fig. 1. Maps of heat fluxes in J-OFURO averaged from October 1992 to September 1993. a. Latent heat flux, b. Sensible heat
flux, c. Shortwave radiation flux, d. Net longwave radiation flux, e. Net heat flux.
data sets in January are smaller and less systematic than
in July. This holds at high latitudes in the South Pacific
where none of the data sets is validated using a satisfactory reference due to a lack of observations. A more detailed comparison among these and other available data
sets is ongoing for longer-period data.
A comparison for longwave radiation flux is shown
in Figs. 2c and 2d. Differences among the four data sets
in January are large at high latitudes in the Southern Hemisphere. In particular, J-OFURO data show large values
compared with other data sets between 30°S and 50°S.
On the other hand, J-OFURO gives small values from
10°N to 40°N in the Northern Hemisphere. Moreover,
NRA gives relatively small values between 30°S and 50°S
and between 45°N and 55°N, while the da Silva et al.
data gives relatively large values between 20°S and 5°N.
In July, J-OFURO shows a small flux from the ocean south
of 10°N and a large flux north of 30°N compared with
other data. As a result, J-OFURO presents a small meridional trend. The differences between each data set may
partly be attributable to the nature of the estimation
scheme of the downward longwave flux. The data of da
Silva et al. are computed with COADS data based on a
bulk method, and there must be a large estimation error
in the tropics and the Southern Hemisphere where ship
observations are generally sparse. Also, the estimation
scheme of the longwave flux used in da Silva et al. is too
simple to obtain a precise estimate and is not suitable for
data-sparse regions. On the other hand, the scheme used
in J-OFURO tends to underestimate the downward flux
under a clear sky and overestimate under a cloudy sky.
The character in the estimation scheme in J-OFURO may
contribute the differences. Another important reason for
the difference may be the capability to reproduce clouds
in the models. The spatial and temporal resolutions of
the global atmospheric models are too coarse to compute
cloud formation and change. Parameterization of cloud
processes is therefore essential to reproduce clouds in the
models and is developed semi-empirically in different
ways from model to model. We also suspect that the surface longwave radiation flux in the models might be computed as a residual of the atmospheric longwave radiation budget, because there is no constraint on the downward longwave radiation at the ocean surface. The residual
Introduction of J-OFURO
219
Fig. 2. Meridional profiles of radiation fluxes along 180° in January 1993 and July 1993. a. Shortwave radiation flux (January),
b. Shortwave radiation flux (July), c. Longwave radiation flux (January), d. Longwave radiation flux (July).
of the longwave radiation budget is estimated from boundary conditions at the top of the atmosphere based on satellite observations and the longwave radiation budget in
each layer of the atmosphere determined by temperature
profiles.
Figures 3a and 3b shows meridional profiles of latent heat flux along the dateline in January and July 1993,
respectively. In January, the maximum heat loss from the
ocean at 10°N–20°N and the minimum heat loss around
60°S and 50°N in January are common in all the data sets,
but the amplitudes are different. J-OFURO shows a large
heat loss between 10°N and 20°N and south of 15°S, and
a small heat loss north of 20°N, compared with other data.
Josey (2001) assessed the accuracy of surface heat flux
estimates from ERA and NRA by comparison with moored
buoy measurements in the subduction region of the Northeast Atlantic. He pointed out that the NRA and ERA data
overestimate latent heat loss, especially in winter. If his
result is applicable to the Pacific Ocean, latent heat loss
220
M. Kubota et al.
between 20°N and 40°N may be overestimated by the
NRA, ERA and da Silva et al. data sets, and J-OFURO
may give more accurate values in the subtropics. In the
Southern Hemisphere, some peaks in J-OFURO cannot
be found in NWP products, and J-OFURO presents a
structure with a small scale compared with NWP products. In July the most remarkable feature is that the flux
by da Silva et al. is extremely small at 30°S–10°S and
10°N–30°N, with a difference from other data reaching
about 200 Wm–2 around 10°S. The other data sets show
considerable coincidence with each other. Although there
is a small systematic difference between 15°S and the
equator, the overall difference between the J-OFURO and
the NWP products is extremely small, less than 10
Wm –2.
Figures 3c and 3d shows meridional profiles of sensible heat flux. All the data sets show that sensible heat
loss in January is large in the Northern Hemisphere with
sharp peaks around 40°N and 60°N. The difference among
Fig. 3. Same as Fig. 2, except for turbulent heat fluxes. a. Latent heat flux (January), b. Latent heat flux (July), c. Sensible heat
flux (January), d. Sensible heat flux (July).
the data sets is relatively large in the Northern Hemisphere; the flux in J-OFURO is smaller north of 20°N
than the other data sets, while the NRA flux is larger than
the others at 15°N–25°N and 35°N–45°N. The significant difference around the 40°N peak may be due to a
large difference in zonal wind speed among the data sets
around the peak at 35°N (Fig. 5a). Sensible heat flux in
July is large in the Southern Hemisphere, but does not
show a sharp peak, unlike January. The NRA flux is larger
at 40°S–30°N and smaller north of 40°N than the other
data sets, and the flux of da Silva et al. is larger at 60°S–
40°S and 0°–10°N. The NRA flux may be overestimated
in particular for 35°S–10°S. This cannot be attributed to
zonal wind speed, which is not stronger than the other
data sets (Fig. 5b).
Meridional profiles of the net heat flux are shown in
Figs. 4a and 4b. A net heat gain (loss) of ocean occurs
almost in the Southern (Northern) Hemisphere in January and the Northern (Southern) Hemisphere in July. In
January, the heat loss of J-OFURO is larger at 10°N–20°N
and smaller north of 20°N than the other data sets. The
former reflects the characteristics of longwave radiation
and latent heat fluxes, and the latter reflects the characteristics of latent and sensible heat fluxes. In the Southern Hemisphere, small-scale variations are remarkable and
are different among the data sets. For example, the data
of da Silva et al. show heat loss around 10°S, while the
other data sets show heat gain there. This may primarily
be due to overestimation of zonal wind speed at 25°S–0°
by da Silva et al. (Fig. 5a). In addition, the accuracy of
NWP products in the Southern Hemisphere may be worse
than that in the Northern Hemisphere, owing to fewer insitu data being assimilated into the NWP models. This
may be a reason for the various small-scale variations of
total heat flux in the Southern Hemisphere. In July, the
difference in net heat flux among the four data sets is
very large around 20°S and 20°N. The data of da Silva et
al. show a small net heat loss around 20°S and a large net
Introduction of J-OFURO
221
Fig. 4. Meridional profiles of net heat flux along 180° in (a) January 1993 and (b) July 1993.
Fig. 5. Same as Fig. 4, except for zonal wind speed.
heat gain around 20°N, while the NRA data show the
opposite. The da Silva et al. data tend to underestimate
latent heat loss (Fig. 3b), and the NRA data tend to underestimate shortwave radiation flux (Fig. 2b) and overestimate sensible heat loss (Fig. 3d).
Figures 5a and 5b shows meridional profiles of zonal
wind speed. All the data sets show similar maxima, the
westerly maxima around 55°S–45°S, 35°S, and 35°N and
the easterly maxima around 15°S–25°S and 10°N–20°N,
although the location and strength of the westerly maxima
in the Southern Hemisphere are quite different between
January and July. In January, the data of da Silva et al.
are much larger at 25°S–0°and around 10°N than the other
data sets, and show a remarkable easterly maximum near
2°S which is not seen in the others. This discrepancy may
be due to an error involved in the data of da Silva et al.,
because they are based on measurement by Volunteer
222
M. Kubota et al.
Observing Ships (VOS) and are relatively sparse in the
equatorial region. The data sets of NRA and da Silva et
al. conclude that there are much stronger maxima of the
easterly and westerly in the Northern Hemisphere than
that in J-OFURO. The difference reaches 3 m s –1 around
35°N. On the other hand, wind speed in the ECMWF data
is weaker than J-OFURO at all latitudes. South of 30°S,
the strength of zonal wind and the latitude of the
southernmost maximum of the westerly are different in
all data sets. Figure 5a shows that there is a discrepancy
of 20–30% between data sets in January. The discrepancy
may be a measure of uncertainty in the zonal wind field;
a large discrepancy suggests a large error in the data sets.
Differences among the data sets are relatively small in
July. The discrepancy is mostly within ±2 m s–1 at all latitudes. An exception is the much smaller wind speed at
25°S–25°N of the ECMWF data. This also holds in other
months, as indicated by Hackert et al. (2001) also who
compared some types of wind product during the 1997–
98 El Niño. Thus, the ECMWF wind data seem to underestimate the easterlies.
4. Summary and Discussion
We have constructed new data sets related to ocean
surface flux mainly using satellite data, named Japanese
Ocean Flux data sets with Use of Remote sensing Observations (J-OFURO). J-OFURO includes data sets of
shortwave and longwave radiation, turbulent heat fluxes,
momentum flux, sea surface dynamic topography,
geostrophic currents and surface air temperature. The data
of turbulent heat flux and air temperature are global in
their coverage. The radiation flux data cover 60°S–60°N,
80°E–160°W, because they are calculated with the data
from the Japanese satellite, Geostationary Meteorological Satellite (GMS). The momentum flux data cover 60°S–
60°N, 30°E–70°W.
The average fields derived from J-OFURO data are
shown for latent and sensible heat fluxes, shortwave and
net longwave radiation fluxes and total heat flux. A large
latent heat loss from the ocean is found around 20° and
over the western boundary currents. Sensible heat flux is
negligible in most regions, except over the western boundary current. The insolation pattern is nearly zonal, the
maximum insolation being located along the zonal subtropical High centered approximately at 20°N. A large
flux of net longwave radiation from the ocean is found in
the western North Pacific, especially around Japan and
in the central North Pacific. Finally, the largest ocean heat
gain is found in the equatorial region, while the largest
heat loss can be found over the Kuroshio and Kuroshio
Extension Regions. Also, all of the data sets in J-OFURO
succeed very well in detecting annual and interannaul
variability. In particular, interannual variability associated with El Niño and La Niña events in the tropical Pacific is detected remarkably well (not shown here).
Nevertheless, a quantitative validation of J-OFURO
should be done. However, the validation of grid data is
difficult, because we cannot obtain a true value for each
grid since errors in making grid data are considerable for
in-situ data due to their low density and the biased distribution of data. We have compared J-OFURO with three
existing data sets of da Silva et al. (1994), NCEP
reanalysis data (NRA) and ECMWF reanalysis data
(ERA). Intercomparison between these data sets has been
done for meridional profiles along the dateline of heat
fluxes (radiation, turbulent heat and total fluxes) and zonal
wind speed in January and July 1993. The data of da Silva
et al. are based on in-situ data such as ship and buoy data.
Shortwave radiation in January and July and longwave
radiation and latent heat flux in January of J-OFURO
show much greater meridional contrast than the other data
sets. WCRP (2000) pointed out that satellite estimates of
shortwave radiation flux may be more reliable than other
global estimates and the NRA and ERA data are smaller
in the tropics and larger at higher latitudes than satellitederived flux, as shown in the present study. The small
meridional contrast for the NRA and ERA data may be
due to a failure in reproducing cloudiness in the models.
As a result, the reanalysis data may underestimate insolation in cloudless regions in the subtropical area and
overestimate insolation in overcast regions in the subarctic
area. The difference in longwave radiation flux between
J-OFURO and the other data sets is considerable, especially in the Southern Hemisphere in July. J-OFURO may
overestimate longwave radiation flux south of 15°N in
July. WCRP (2000), however, pointed out that the zonal
mean longwave radiation flux estimated from satellite data
is lower than that according to other estimates. J-OFURO
shows an underestimation at 10°N–40°N in January and
35°N–50°N in July, but this is not so remarkable, except
at 10°N–30°N in January.
The data set of da Silva et al. very much underestimates the latent heat loss from the ocean in the subtropics around 20°S and 20°N in July. This is consistent with
WCRP (2000), which showed an underestimate of the
COADS-based product compared with ERA and NRA.
The latent heat flux of da Silva et al. seems unreliable in
the subtropics in northern summer. The latent heat flux
in January of J-OFURO overestimates at 10°N–20°N and
underestimates at 25°N–40°N. The NRA data set greatly
overestimates sensible heat flux in the subtropics of the
Southern Hemisphere in July. WCRP (2000) showed the
same feature of NRA data in terms of zonal mean sensible heat flux. The sensible heat flux of NRA may not be
reliable in the southern subtropics in northern summer.
J-OFURO shows a larger net heat loss from the ocean
at 10°N–20°N and a smaller loss north of 20°N in January than the other data sets. The former is caused by large
fluxes of longwave radiation and latent heat, while the
latter is caused by small fluxes of latent and sensible heat.
Josey (2001) pointed out that the ERA and NRA data sets
overestimate net heat flux partly due to an overestimate
of latent heat flux, in comparison with measurements of
buoys moored at mid-latitudes in the northeast Atlantic
in the subduction program. Our results also suggest an
overestimation by ERA and NRA products compared with
J-OFURO data. Small-scale variations are remarkable in
the Southern Hemisphere compared with the Northern
Hemisphere, and are considerably different among all data
sets, probably due to inaccuracy caused by sparse in-situ
observations. In July, the difference in net heat flux among
the four data sets is very large around 20°S and 20°N.
The data of da Silva et al. show a small net heat loss
around 20°S and a large net heat gain around 20°N, while
the NRA data show the opposite.
Introduction of J-OFURO
223
In terms of zonal wind speed, all the data sets show
a similar location and strength of westerly and easterly
maxima. Nevertheless, the data of da Silva et al. are very
different from the other data sets in the equatorial zone;
they show an easterly maximum near 2°S that is not found
in the other products, and greatly underestimate at southern low latitudes.
As a result of the intercomparison, it is probable that
the reliability of COADS-based grid data like da Silva et
al. is lower due to few in-situ observations than that of
reanalysis data like the NWP data and satellite-derived
data like J-OFURO, although one cannot easily conclude
which data set is best. In order to validate various data
sets and find the best one, we need much more in-situ
data in the ocean, and should make an effort to increase
the number of in-situ observation stations which provide
high quality data by, for example, the Improved Meteorology (IMET) buoys. In addition, we need to develop other
methods for data validation, because surface-mooring
buoys cannot cover the whole ocean. One of the methods
for making more reliable data sets is to assimilate in-situ
and satellite data into general circulation models of ocean
and atmosphere. ECMWF has already begun to assimilate satellite data observed by ERS and SSM/I into atmospheric general circulation model. For this purpose, it
is crucial to open observed data and calculated results
and to increase the chance of interaction between data
providers and users. Intercomparison among open data
sets is still an effective way to clarify the characteristics
of each product. In these activities, we should keep in
mind that both the good and bad points of a data set depend strongly on the purpose of data usage, and cannot
be common for every purpose. Finally, we hope that our
J-OFURO data set will be useful for scientific research,
and will make many people comfortable, like OFURO in
Japanese, which means a Japanese-style bath filled with
hot water.
Acknowledgements
This research was partly supported by the Grant-Aid
for Science Research on Priority Areas (No. 08241111)
of the Ministry of Education, Science, Sports and Culture, Japan and by Research and Development Applying
Advanced Computational Science and Technology, Japan
Science and Technology Corporation. We would like to
thank Dr. M. Kawabe for helpful comments about this
paper. One of the authors (N. Iwasaka) expresses his sincere gratitude to Dr. K. Ohtsuka of Obayashi Corporation Technical Research Institute for his help in developing the computer code for calculating the longwave radiation flux. He also addresses his thanks to Mr. A. Shinpo
and Dr. H. Nakamura of the Graduate School of Science,
University of Tokyo, for providing the NCEP reanalysis
data which they processed for research use from the origi-
224
M. Kubota et al.
nal data. The original NCEP data was provided by the
NCAR library. One of the authors (K. Kutsuwada) expresses his appreciation to the Institut Français de Recherche pour l’Exploitation de la Mer (IFREMER) for
kindly supplying the scatterometer data in the calculations of wind and wind-stress vectors. We also thank for
their great encouragement Drs. H. Masuko and N. Ebuchi
and other members in the Japanese Science Team, which
was organized by the National Space Development
Agency of Japan. Finally we thank H. Tomita for his assistance in preparing the figures.
References
Chou, M.-D., P.-K. Chan and M. M.-H. Yan (2001): A sea surface radiation data set for climate applications in the tropical western Pacific and South China Sea J. Geophys. Res.,
106, 7219–7228.
da Silva, A. M., C. C. Young and S. Levitus (1994): Atlas of
Surface Marine data 1994. Vol. 1, Algorithms and Procedures. U.S. Department of Commerce, 74 pp.
Ebuchi, N., H. C. Graver and R. Vakkayil (1996): Evaluation
of ERS-1 scatterometer winds with wind and wave ocean
buoy observations. Tech. Rep., CAOS, Tohoku Univ., CAOS
96-1, Sendai, 69 pp.
Esbensen, S. K. and V. Kushnir (1981): The heat budget of the
global ocean: An atlas based on estimates from surface
marine observations. Tech. Rep. 29, Clim. Res. Inst., Oreg.
State Univ.
Gautier, C., P. Peterson and C. Jones (1998): Ocean surface air
temperature derived from multiple data sets and artificial
neural network. Geophys. Res. Lett., 25, 4217–4220.
Goldman, A. and T. G. Kyle (1968): A comparison between
statistical model and line calculation with application to the
9.6 micro meter ozone and the 2.7 micro-meter water vapor.
J. Meteor., 12, 272–286.
Goody, R. M. (1964): Atmospheric Radiation. I: Theoretical
Basis. Oxford Univ. Press, London, 436 pp.
Graver, H. C., N. Ebuchi and R. Vakkayil (1996): Evaluation
of ERS-1 scatterometer winds with wind and wave ocean
buoy observations. Tech. Rep., Univ. of Miami, RSMAS
96-003, Miami, 69 pp.
Gupta, S. K. (1989): A parameterization for longwave surface
radiation from sun-synchronous satellite data. J. Climate,
2, 305–320.
Hackert, E. C., A. J. Busalacchi and R. Murtugudde (2001): A
wind comparison study using an ocean general circulation
model for the 1997–1998 El Niño. J. Geophys. Res., 106,
2345–2362.
Husar, R. B., J. M. Prospero and L. L. Stowe (1997): Characterization of Tropospheric aerosols over the oceans with the
NOAA advanced very high resolution radiometer optical
thickness operational product. J. Geophys. Res., 102, 16889–
16909.
Iqbal, M. (1983): An Introduction to Solar Radiation. Academic
Press, Canada, 390 pp.
Iwasaka, N. and K. Hanawa (1990): Climatologies of marine
meteorological variables and surface fluxes in the North
Pacific computed from COADS. Tohoku Geophys. J., 33,
185–239.
Jones, C., P. Peterson and C. Gautier (1999): A new method for
deriving ocean surface specific humidity and air temperature: An artificial neural network. J. Appl. Meteorol., 38,
1229–1245.
Josey, S. A. (2001): A comparison of ECMWF, NCEP/NCAR,
and SOC surface heat fluxes with moored buoy measurements in the subduction region of the northeast Atlantic. J.
Climate, 14, 1780–1789.
Kano, A. and M. Kubota (2000): Accuracy of satellite-derived
turbulent heat fluxes. Proceedings of the 5th Pacific Ocean
Remote Sensing Conference 2000, p. 613–616.
Kizu, S. (1998): Systematic errors in estimation of insolation
by empirical formulas. J. Oceanogr., 54, 165–177.
Kizu, S. (1999): Estimation of surface shortwave radiation over
the western Pacific based on satellite measurements. Report of Sci. Res. on Priority Areas of Better Understanding
of Water and Energy Cycles by Satellites (BUWECS) by
the Japanese Ministry of Education, Science, Sports and
Culture, 1–10 (in Japanese).
Kizu, S. (2001a): Degradation of VISSR visible radiometers
aboard three GMS-series satellites from 1987 to 1999. Int’l
J. Remote Sens. (submitted).
Kizu, S. (2001b): A simple method to estimate surface insolation using Japanese Geostationary Meteorological Satellites.—Method and validation—. J. Oceanogr. (submitted).
Konda, M., N. Imasato, K. Nishi and T. Toda (1994): Measurement of the sea surface emissivity. J. Oceanogr., 50, 17–
30.
Konda, M., N. Imasato and A. Shibata (1996): A new method
to determine the near sea surface temperature by using satellite data. J. Geophys. Res., 101, 14349–14360.
Kondo, J. (1975): Air-sea bulk transfer coefficients in diabatic
conditions. Bound.-Layer Meteor., 9, 91–112.
Kubota, M. and S. Mitsumori (1997): Sensible heat flux estimated by using satellite data over the North Pacific. p. 127–
136. In Remote Sensing of Subtropical Ocean, ed. by C. T.
Liu, Elsevier.
Kubota, M. and A. Shikauchi (1995): Air temperature at ocean
surface derived from surface-level humidity. J. Oceanogr.,
51, 619–634.
Kutsuwada, K. (1998): Impact of wind/wind-stress field in the
North Pacific constructed by ADEOS/NSCAT data. J.
Oceanogr., 54, 443–456.
Kutsuwada, K. and T. Kazama (2000): Intraseasonal variation
of surface wind in the tropical Pacific-Indian Oceans during 1997–98 El Niño event: Use of satellite scatterometer
data—. Proceedings of the 5th Pacific Ocean Remote Sensing Conference 2000, p. 244–247.
Large, W. G. and S. Pond (1981): Sensible and latent heat flux
measurements over the ocean. J. Phys. Oceanogr., 12, 464–
482.
Levy, G. and R. Brown (1986): A simple, objective analyses
scheme for scatterometer data. J. Geophys. Res., 91, 5153–
5158.
Liu, Q., C. Simmer and E. Ruprecht (1997): Estimating
longwave net radiation at sea surface from the Special Sen-
sor Microwave/Imager (SSM/I). J. Appl. Meteor., 36, 919–
930.
Liu, W. T. (1986): Statistical relation between monthly mean
precipitable water and surface-level humidity over global
ocean. Month. Wea. Rev., 114, 1591–1602.
Nunez, M. (1988): A comparison of three approaches to estimate daily totals of global solar radiation in Australia using GMS data. Aust. Meteor. Mag., 36, 25–33.
Oberhuber, J. M. (1988): An atlas based on the ‘COADS’ data
set: the budgets of heat, buoyancy, and turbulent kinetic
energy at the surface of the global ocean. Max-Plank-Institute for Meteorology, Report No. 15, 200 pp.
Pierson, W. J., Jr. (1990): Examples of, reasons for, and consequences of the poor quality of wind data from ships for the
marine boundary layer: implications for remote sensing. J.
Geophys. Res., 95, 13313–13340.
Prihodko, L. and S. N. Groward (1997): Estimation of air temperature from remotely sensed surface observations. Remote
Sens. Environ., 60, 335–346.
Reed, R. (1977): On estimating insolation over the ocean. J.
Phys. Oceanogr., 7, 482–485.
Reynolds, R. W. and T. M. Smith (1994): Improved global sea
surface temperature analyses using optimum interpolation.
J. Climate, 7, 929–948.
Reynolds, R. W. and T. M. Smith (1995): A high resolution global sea surface temperature climatology. J. Climate, 8,
1571–1583.
Roberts, R. E., J. A. Selby and L. M. Bibermann (1976): Infrared continuum absorption by atmospheric water vapor in
the 8–12 micro meter window. Appl. Opt., 15, 2085–2090.
Rogers, C. D. and C. D. Walshaw (1966): The computation of
infrared cooling rate in planetary atmospheres. Quart. J.
Roy. Meteoro. Soc., 92, 67–92.
Rossow, W. B. and Y.-C. Zhang (1995): Calculation of surface
and top of atmosphere radiative fluxes from physical quantities based on ISCCP data sets. 2 Validation and first results. J. Geophys. Res., 100, 1167–1197.
Schlüssel, P., L. Schanz and G. Englisch (1995): Retrieval of
latent heat flux and longwave irradiance at the sea surface
from SSM/I and AVHRR measurements. Adv. Space Res.,
16, 107–116.
Schulz J., P. Schlüssel and H. Graßl (1993): Water vapor in the
atmospheric boundary layer over oceans from SSM/I measurements. Int. J. Remote Sens., 14, 2773–2789.
Tanimoto, Y. (1993): Variations with multiple time scales of
the global sea surface temperature fields and their relationship with the atmospheric general circulation. Doctoral
Thesis, Tohoku University, 219 pp.
WCRP (2000): Intercomparison and validation of ocean-atmosphere energy flux fields. Working Group on air-sea fluxes,
WCRP-112, 303 pp.
Wentz, F. J. (1994): User’s manual SSM/I-2 geophysical tapes,
Remote Sensing Systems. Tech. Rep. 070194, 44 pp.
Weymouth, G. and J. Le Marshall (1994): An operational system to estimate insolation over the Australian Region. Proceedings of the Pacific Ocean Remote Sensing Conference,
Melbourne, Australia, 1–4 March 1994, p. 443–449.
Introduction of J-OFURO
225