Abstract

During 30 days in May and June 2003, the R/V Southern Surveyor was operating in the Gulf of Carpentaria, northern Australia. Measurements of sea surface temperature (SST) were made with an accurate single-channel infrared radiometer as well as with the ship’s thermosalinograph. These ship-based measurements have been used to assess the quality of the SST derived from nine satellite-borne instruments. The satellite dataset compiled during this period also allows the intercomparison of satellite-derived SST fields in areas not covered by the ship’s track. An assessment of the SST quality from each satellite instrument is presented, and suggestions for blending ground and satellite measurements into a single product are made. These suggestions are directly applicable to the international Global Ocean Data Assimilation Experiment (GODAE) High Resolution SST Pilot Project (GHRSST-PP) that is currently developing an operational system to provide 6-hourly global fields of SST at a spatial resolution close to 10 km. The paper demonstrates how the Diagnostic Datasets (DDSs) and Matchup Database (MDB) of the GHRSST-PP can be used to monitor the quality of individual and blended SST datasets. Recommendations for future satellite missions that are critical to the long-term generation of accurate blended SST datasets are included.

1. Introduction

Sea surface temperature (SST) is one of the key parameters used in global numerical modeling of weather and climate processes. Many space agencies have recently launched satellites that include infrared and microwave instruments designed to provide global or regional data on the temperature of the ocean surface. Infrared systems that provide operational products include the Advanced Very High Resolution Radiometer (AVHRR) on the National Oceanic and Atmospheric Administration (NOAA) series of polar-orbiting satellites and the imaging radiometers on all geostationary operational meteorological satellites. In this study we have access to data from the Geostationary Operational Environmental Satellite-9 (GOES-9) Multispectral Imager (MSI) after it was moved from above the eastern Pacific Ocean to over the equator north of Australia. The Along-Track Scanning Radiometer (ATSR) series flying on European satellites has been specifically designed to provide accurate estimates of SST suitable for climate applications. Other infrared instruments have been launched on a series of environmental satellites, specifically the Global Imager (GLI) on the Advanced Earth Observing Satellite-II (ADEOS-II) and the Moderate Resolution Imaging Spectroradiometers (MODIS) on the Terra and Aqua satellites. Advanced Microwave Scanning Radiometer (AMSR) instruments have been included on both the ADEOS-II and Aqua satellites. In this study only data from the Earth Observing System (EOS) AMSR (AMSR-E) on Aqua are used. Unfortunately, ADEOS-II has not provided data since November 2003, but data from GLI have been made available for this study. Further details on the satellite instruments used in this study are included in a later section.

The instruments described above all supply estimates of SST that have a range of spatial resolutions and accuracies. However, in most cases, the accuracies of satellite-derived SST estimates are comparable to those for data collected from ground-based platforms. The suite of satellite instruments listed above now provides a global coverage that is not practical from ships and buoys. With all these data being available to research and application communities there is now a need to develop methods for getting the best SST possible using all the satellite and ground-based data that are available. This is the aim of the Global Ocean Data Assimilation Experiment (GODAE) High Resolution SST Pilot Project (GHRSST-PP) (www.ghrsst-pp.org). The available data include a range of satellite orbits and overpass times, several different infrared spectral configurations, and microwave-based estimates that can be used in cloudy conditions. A further complication is that infrared radiometers measure the radiometric skin temperature of the ocean, microwave radiometers give subskin measurements, and ship and buoy measurements of SST are typically at a depth of 1–3 m. Thus, when developing a blended SST from all available data all these factors need to be taken into account. Finally, any system developed under the GHRSST-PP program needs to be flexible enough to account for loss of sensors, the introduction of new sensors, and any development of new blending techniques.

In this paper satellite data from the nine instruments named above (and listed in Table 1) have been assembled for comparison with shipborne measurements of SST. The data analysis shows the performance of each instrument in tropical clear-sky conditions. The data also allow the intercomparison of the different satellite-derived SST estimates without reference to any ground-based data. The analysis techniques developed here demonstrate how, in GHRSST-PP, in situ and satellite datasets can be used to monitor the accuracy of both individual and blended SST datasets.

Table 1.

Satellite radiometers used in the analysis with the number of cloud-free coincidences with the ship location. The overpass times for each satellite over the Gulf of Carpentaria and the channels used for SST determination are also included. UTC = Gulf of Carpentaria overpass times (UTC) day (night); Resolution = spatial resolution (km).

Satellite radiometers used in the analysis with the number of cloud-free coincidences with the ship location. The overpass times for each satellite over the Gulf of Carpentaria and the channels used for SST determination are also included. UTC = Gulf of Carpentaria overpass times (UTC) day (night); Resolution = spatial resolution (km).
Satellite radiometers used in the analysis with the number of cloud-free coincidences with the ship location. The overpass times for each satellite over the Gulf of Carpentaria and the channels used for SST determination are also included. UTC = Gulf of Carpentaria overpass times (UTC) day (night); Resolution = spatial resolution (km).

2. Ship instrumentation

a. DAR011 infrared radiometer

The DAR011 radiometer is a single-channel, self-calibrating, infrared radiometer developed and built within the Commonwealth Scientific and Industrial Research Organisation (CSIRO). The radiometer has a heritage going back many years and is the culmination of developments leading to a reliable accurate instrument. Full details of the instrument are provided by Bennett (1998). A rotating 45° plane mirror sequentially views the sea, a hot blackbody calibration target, the sky, and finally an ambient temperature blackbody calibration target. The incoming radiation is physically chopped against a second ambient temperature blackbody target, and the chopped radiation is focused with a 45° parabolic front-surfaced mirror onto a pyroelectric detector. The detector is located behind an interference filter that passes radiation with wavelengths between 10.5 and 11.5 μm. The temperatures of the two calibration blackbody targets are accurately monitored, providing excellent absolute radiometric accuracy.

During 2001 the DAR011 radiometer was included in the Miami2001 infrared radiometer calibration and intercomparison. The radiometer was calibrated against a National Institute of Standards and Technology (NIST)-designed blackbody target and compared against other similar instruments used for the validation of satellite-derived surface temperatures, and was found to perform with a high degree of accuracy—better than 0.1 K. Results from the Miami2001 exercise are reported by Barton et al. (2004) and Rice et al. (2004). Since Miami2001, regular comparisons with laboratory-based blackbody targets and other ship-based measurements suggest that the DAR011 radiometer has maintained its accuracy. A further radiometer intercomparison is planned for the near future to ensure that shipborne infrared radiometers have maintained their capability of providing skin SST measurements suitable for validation.

On the R/V Southern Surveyor the infrared radiometer was mounted above the bridge and viewed the sea surface outside the ship’s wake with a nadir angle of 45°. The radiometer was generally operated for two periods of about 4 h each day throughout the voyage. On some occasions, the radiometer was not operated due to other operational requirements. Typical operating times were 0800–1200 and 2000–2400 LT (LT is 10 h ahead of UTC).

The radiometer measurements of the sea surface were corrected for nonunity surface emissivity using the sky-view measurements and an emissivity of 0.9875 that is interpolated from the tables given by Masuda et al. (1988). The data were then averaged over 1-min intervals for comparison with the satellite and thermosalinograph measurements.

b. Bulk SST and other ship measurements

The bulk SST on the R/V Southern Surveyor is measured continuously by the ship’s thermosalinograph with the water intake being at a depth of 3 m. The thermosalinograph temperature is accurate to within 1 mK. A full set of meteorological data are recorded, including wind velocity, air temperature, relative humidity, surface air pressure, and insolation. Ship location, heading, and speed are provided by GPS, and the voyage track is shown in Fig. 1.

Fig. 1.

Voyage track of the R/V Southern Surveyor during voyage SS0403.

Fig. 1.

Voyage track of the R/V Southern Surveyor during voyage SS0403.

3. Comparisons between ship and satellite SST estimates

Image data from nine different satellite instruments were obtained for comparison with the ship measurements of radiometric and bulk SST as supplied by the DAR011 radiometer and the thermosalinograph, respectively. Coincident thermosalinograph and satellite data used in the analysis were obtained within the same minute. For the radiometer the same coincident time applied, except when the radiometer was in a calibration mode, and then the coincidence was always less than 3 min. The satellite instruments are listed in Table 1 along with details of the number of images supplied and the number of cloud-free coincidences with the ship location.

It is important to note that the SST algorithms for the Advanced ATSR (AATSR) and ATSR-2 are derived theoretically and give a radiometric temperature of the sea surface, which is often called the skin temperature. In contrast, the SST algorithms for the other satellite instruments are tuned to give a bulk SST, even though infrared radiometers sense the skin temperature and the microwave radiometers sense the subskin temperature at a depth close to 1 mm.

a. AVHRR on NOAA-12 and NOAA-16

The AVHRR instruments on two NOAA satellites (NOAA-12 and NOAA-16) provided coincident data with the ship measurements during the voyage. AVHRR instruments have three infrared channels (3, 4, and 5) with a 1-km nadir spatial resolution and central wavelengths at 3.7, 10.8, and 12.0 μm. The data were received at the Australian Centre for Remote Sensing (ACRES) station at Alice Springs and relayed to the CSIRO laboratory in Hobart. For NOAA-12 the SST was derived using the nonlinear SST (NLSST) algorithm coefficients (available online at http://noaasis.noaa.gov/NOAASIS/pubs/SST/noaa12sst.asc). These coefficients are based on December 1993 drifting buoy matchups. For NOAA-16 the NLSST coefficients (available online at http://www2.ncdc.noaa.gov/docs/klm/html/g/app-g3.htm#a101901aa) are based on March 2001 global drifting buoy matches.

The algorithms were used to create fields of SST for comparison with the ship-based measurements. This processing includes an initial cloud-clearing based on the method of Saunders and Kriebel (1988). The satellite pixel covering the ship location was identified, and a mean value over 5 × 5 pixels about the ship location was derived. The standard deviation of the SST in this small area was calculated, and if this was greater than 0.5 K then the data were assumed to be contaminated by cloud and not used in the analysis. Figure 2 shows the differences between the AVHRR and thermosalinograph estimates of SST for the entire AVHRR dataset. Data for the two different instruments are identified in the figure, and the bias and standard deviation for the comparisons are, respectively, −0.41 and 0.79 K for NOAA-12 AVHRR and 0.07 and 0.51 K for the instrument on NOAA-16. For NOAA-12 the overpass times were close to 0515 and 1730 LT and for NOAA-16 they were 0230 and 1445 LT. The instrument on NOAA-12 does not perform as well as that on NOAA-16, which may not be surprising as the former instrument was launched in 1991 and has survived for a period well beyond its design life. Gaps in Fig. 2, for example between AVHRR pass numbers 91 and 95, are due to the removal of cloud-contaminated data.

Fig. 2.

Comparisons between AVHRR SST and the ship thermosalinograph bulk SST.

Fig. 2.

Comparisons between AVHRR SST and the ship thermosalinograph bulk SST.

b. MODIS on Terra and Aqua

MODIS is a 36-channel instrument flying on both the Terra and Aqua satellites launched by the National Aeronautics and Space Administration (NASA) in December 1999 and March 2002, respectively. The MODIS channels cover the spectral bands between wavelengths of 400 nm and 15 μm, with a nadir spatial resolution of near 1 km. Channels at 3.7, 4.0, 10.8, and 12.0 μm are used to derive the SST.

SST was derived using the algorithms supplied by the Rosenstiel School of Marine and Atmospheric Science at the University of Miami for each instrument, and the values were then averaged over 5 × 5 pixel areas. The development of the MODIS SST algorithms, similar to that used for the AVHRR Pathfinder Project, is described by Kilpatrick et al. (2001). Again those cases in which the standard deviation of the SST in the 5 × 5 pixel area was more than 0.5 were assumed to be cloud-contaminated. Finally, the SST average values were compared with the thermosalinograph measurements, and the results are shown in Fig. 3. The Terra satellite views the Gulf of Carpentaria at close to 1100 and 2330 LT, while the Aqua times are nearer 0200 and 1430 LT. For the MODIS on Terra the comparisons give a bias and standard deviation of −0.15 and 0.43 K, respectively, while for Aqua these values are 0.05 and 0.41 K.

Fig. 3.

Comparisons between MODIS SST and the ship TSG bulk SST.

Fig. 3.

Comparisons between MODIS SST and the ship TSG bulk SST.

c. ATSR-2 on European Remote Sensing Satellite-2 (ERS-2) and AATSR on Envisat

ATSR-2 and AATSR are satellite instruments that were designed specifically for the accurate measurements of SST. Each instrument has three infrared channels that view the earth’s surface at two different view angles by using an offset conical scan. Full details of this design are described by Prata et al. (1990). The instruments include two accurate calibration targets and provide data with 12-bit digitization. SST can be estimated by using different combinations of infrared channels and view angles, and algorithm coefficients have been derived using a theoretical radiative transfer model (see Závody et al. 1995). The 2- and 3-channel algorithms use the nadir view data at 11 and 12 μm and 3.7, 11, and 12 μm, respectively. The 4- and 6-channel algorithms use the same channels but with both the nadir and forward views and are the algorithms used in this analysis. The 3.7-μm data are only usable at night due to reflected solar contamination during the day.

The DAR011 radiometer measurements were corrected for nonunity sea surface emissivity and then averaged over 1-min intervals. Coincidences between the AATSR data and the ship measurements were isolated, and the AATSR brightness temperature fields were then scanned to ensure that the sky in the vicinity of the vessel was free from clouds.

After this analysis was completed ATSR-2 and AATSR each gave four occasions with coincident satellite and DAR011 data. For these coincidences the standard European Space Agency (ESA) AATSR and ATSR-2 SST algorithm coefficients were used with the brightness temperatures to produce the two SST estimates for the 4- and 6-channel algorithms. These estimates were then averaged over areas of 5 × 5 pixel windows for each ship–satellite coincidence. The differences between the DAR011 measurements and the satellite-derived estimates from ATSR-2 and AATSR are shown in Fig. 4. Both AATSR and ATSR-2 have overpass times of approximately 1030 and 2300 LT.

Fig. 4.

ATSR-2 and AATSR radiometric SST comparisons with the DAR011 measurements.

Fig. 4.

ATSR-2 and AATSR radiometric SST comparisons with the DAR011 measurements.

All SST values are obtained with the 6-channel algorithm except for the single daytime coincidence when the 4-channel algorithm is used. For the dataset used here DAR011 data were available for all cloud-free ATSR-2 and AATSR passes. The results in Fig. 4 show that all the estimates for both AATSR and ATSR-2 are within the 0.3-K target, and the mean difference is −0.12 K for both instruments. Insufficient data are available for any further statistical analysis. This situation highlights the major limitation of the ATSR instruments. With a narrow swath of 500 km and a 35-day repeat cycle, tropical areas may be viewed less than 14 times in the 35-day period and can have periods of up to 8 days without coverage.

d. GLI on ADEOS-II

The GLI on the ADEOS-II satellite is a 36-channel radiometer that is similar to NASA’s MODIS instrument. The radiometer has four infrared channels that can be used to derive SST. These channels (30, 34, 35, and 36) all have a spatial resolution of 1 km and have central wavelengths of 3.7, 8.6, 10.8, and 12.0 μm.

The latest version (1.6) of the GLI satellite data was kindly supplied by the Japan Aerospace Exploration Agency (JAXA). For the period between 1 May and 10 June, 112 GLI data files were provided, 22 of which provided clear-sky matchups with the ship measurements. Cloudless skies were confirmed by manual scanning of the satellite radiance fields and confirmed by a steady DAR011 sky-view radiance measurement when these data were available. All local ship times and dates were converted to UTC to match the satellite data. The overpass time of each ADEOS-II satellite pass was obtained (typically these were close to 1100 and 2315 LT), and the ship location at that time was then determined. The bulk temperature was obtained from the thermosalinograph, and the radiometric temperature of the sea surface was derived as described above.

SST was calculated using the official algorithms supplied by JAXA. These use channels 34, 35, and 36 during the daytime, and the same three channels plus channel 30 at night. Further details of the GLI SST algorithm development and validation are given by Sakaida et al. (2006).

The differences between the thermosalinograph measurements and the 5 × 5 pixel averages of the SST derived using the daytime and nighttime GLI algorithms are shown in Fig. 5. For the daytime data the comparisons gave a bias and standard deviation of −0.01 and 0.19 K, while for the night data these values were 0.30 and 0.36 K. When combined into a single dataset these statistics were 0.22 and 0.36 K. Radiometric data from the DAR011 radiometer are also available but have not been compared with the GLI data, as the JAXA algorithms are designed to provide bulk SST estimates. Suffice it to say that the difference between the bulk and radiometric SST (the skin effect) is, as anticipated, between 0.0° and 0.3°, with the radiometric temperature being cooler.

Fig. 5.

Comparisons between GLI SST and the ship TSG bulk SST.

Fig. 5.

Comparisons between GLI SST and the ship TSG bulk SST.

e. MSI on GOES-9

The MSI on the geostationary GOES-9 satellite has three thermal infrared channels at wavelengths of 3.8, 10.7, and 12 μm, with a spatial resolution of near 4 km. These channels are capable of providing an estimate of the sea surface temperature in the same manner as the AVHRR instruments on the NOAA polar-orbiting satellites. Full details and specifications of the infrared channels are given by May and Osterman (1998). The GOES-9 satellite was moved from its location over the western United States to 140°E early in May 2003, and the Australian Commonwealth Bureau of Meteorology started to receive data on 15 May. The early data were cloud-affected and sporadic, so the GOES-9 analysis in this work was started on 20 May (day 140). GOES-9 images were received approximately hourly, and 640 images were provided for the period between 15 May and 9 June.

The GOES-9 data were supplied as separate files for latitude, longitude, and the two thermal infrared brightness temperatures. The GOES SST estimates were calculated using the two brightness temperatures and the operational algorithm presented by May and Osterman (1998). When applying the GOES-9 SST algorithm the brightness temperature differences were averaged over a 5 × 5 pixel area about the central value. For comparison with other satellite and ship measurements, the SST values were finally averaged over a further 5 × 5 pixels. This averaging process was required because the GOES-9 brightness temperatures are provided as 8-bit data, and significant smoothing is required to reduce the digitization effects to an acceptable level. As a result of this process, each final individual SST value is then an average over approximately 0.20° in latitude and 0.12° in longitude (22 km × 12 km). When the GOES-9 SST values were plotted against time for different locations it was obvious that further smoothing of the results with time was required. A running mean of five consecutive values (usually 5 h) was found to give consistent estimates, with a steady variation of SST with time instead of the unreal sporadic variations from hour to hour.

A preliminary assessment of the diurnal variations in the GOES-9 SST estimates suggested that the data had not been corrected for the “midnight blackbody calibration” effect, as described by Weinreb et al. (1997). Because this facet of GOES-9 data analysis had not been implemented for the GOES-9 new location during the observation period, only daytime data from GOES-9 are used in this analysis.

Using daytime SST values, a comparison was then performed between the GOES SST and coincident measurements with the ship’s thermosalinograph. Data were not included in this analysis if the standard deviation of either GOES-9 11- or 12-μm brightness temperatures over a 5 × 5 pixel area exceeded 0.5 K. This simple test was used to exclude cloud-contaminated data and provided 217 coincidences for comparison with the thermosalinograph temperatures (see Fig. 6). The comparisons gave a bias and standard deviation of 0.51 and 0.50 K, respectively.

Fig. 6.

Comparison between coincident GOES-9 SST estimates and the ship TSG.

Fig. 6.

Comparison between coincident GOES-9 SST estimates and the ship TSG.

f. AMSR-E on Aqua

The AMSR-E is an 8-channel radiometer supplied to NASA by the National Space Development Agency of Japan (NASDA), now named JAXA, for inclusion in the payload of the Aqua satellite. (Fields of SST and other AMSR-E geophysical products are supplied by Remote Sensing Systems online at http://www.remss.com/, where information on algorithm derivation and validation is also available.) Orbital SST fields are provided daily as well as average values over 3 days, a week, and a month. The daily orbital data, which are used in this analysis, are mapped onto a 0.25° (approximately 28 km) latitude–longitude grid. Data that are within 100 km of shore are not included due to possible land surface contamination in the 38-km footprint of the radiometer. Thus the number of coincidences with the ship thermosalinograph is limited to eight cases, when the ship was in the north of the study area and well away from the coast. The differences between the thermosalinograph and AMSR-E estimates of SST are plotted in Fig. 7, and a statistical analysis gave a bias of 0.05 K and a standard deviation of 0.67 K.

Fig. 7.

Comparison between coincident AMSR-E SST estimates and the ship TSG.

Fig. 7.

Comparison between coincident AMSR-E SST estimates and the ship TSG.

g. Summary of satellite–ship intercomparisons

The previous sections contain an assessment of the performance of each satellite radiometer in the determination of the SST in the Gulf of Carpentaria during the north Australian “dry” season. While this is a limited geographic area and a short time period, the results give a valuable insight into how each radiometer performs and how it may contribute to a blended dataset that is compiled from all satellite and ground-based systems. The bias and standard deviation of each comparison between the ship and satellite estimates are given in Table 2. The polar-orbiting infrared instruments, except for the long-serving NOAA-12 AVHRR, all provide SST with accuracies near or better than 0.5 K. The benefit of providing data with 12-bit digitization (instead of the 10 bits of AVHRR) can be seen in the improved standard deviations of MODIS and GLI estimates. The MSI provides data with only 8-bit resolution, and temporal and spatial averages are required to obtain data with acceptable accuracy. With the advent of new geostationary radiometers on the European Meteosat Second Generation (MSG) satellites (www.esa.int/msg) and Japan’s Multifunctional Transport Satellite (MTSAT-1R) (www.jma.go.jp/en/gms), the provision of infrared data with 10-bit digitization will give a marked improvement in SST measurements from geostationary orbit.

Table 2.

Biases and standard deviations for the clear-sky comparisons between the ship and satellite estimates of SST. Note that all comparisons are between the satellite-derived values and those from the thermosalinograph except for ATSR-2 and AATSR, where the comparisons are with the radiometric temperatures measured by the DAR011 infrared radiometer.

Biases and standard deviations for the clear-sky comparisons between the ship and satellite estimates of SST. Note that all comparisons are between the satellite-derived values and those from the thermosalinograph except for ATSR-2 and AATSR, where the comparisons are with the radiometric temperatures measured by the DAR011 infrared radiometer.
Biases and standard deviations for the clear-sky comparisons between the ship and satellite estimates of SST. Note that all comparisons are between the satellite-derived values and those from the thermosalinograph except for ATSR-2 and AATSR, where the comparisons are with the radiometric temperatures measured by the DAR011 infrared radiometer.

The single microwave radiometer in this analysis provides SST with a low bias and an acceptable standard deviation. The ability of these radiometers to provide SST estimates in cloudy conditions ensures that they will be a valuable asset in the development of near-real-time global datasets—especially in those regions that endure long periods of cloud cover.

4. Satellite–satellite intercomparisons

The Gulf of Carpentaria dataset, with its assembly of images from nine different satellites, provides the opportunity of comparing satellite-derived SST values without using any surface-based data. This is a valuable technique to develop, as there are many geographical locations, especially in the Southern Ocean, where reliable surface-based data are nonexistent. Comparisons between the products provided by different instruments will provide useful information for the development of future data-blending techniques.

Cloud-free latitude–longitude areas of 1.5° × 1.5° have been selected from the dataset to compare the SST fields from different sensors. The areas were chosen to be clear of islands and to provide cloud-free data from each of the instruments used in the intercomparison. The SST estimate for each sensor has been remapped onto a 0.01° latitude–longitude grid giving an image of 151 × 151 pixels. For AMSR-E no interpolation was applied, and the SST value is constant over each (AMSR-E) pixel of 0.25° × 0.25°. For the other sensors a nearest-neighbor approach was used to select the SST value on each grid point. In each of the first three comparisons undertaken here the AVHRR (NOAA-16) estimates were available and were thus used as the base measurement. In the fourth comparison three images from the GOES-9 SMI were used to identify possible diurnal warming of the ocean surface in low-wind-speed conditions. Figure 8 shows images of the SST fields used in each comparison, while Fig. 9 shows histograms of the differences between the SST estimates at each grid point in the remapped images. Each case is discussed separately below.

Fig. 8.

Common SST fields from different sensors: (a) night of 25 May, (b) night of 22 May, (c) day of 23 May, and (d) three GOES-9 fields for different times on 23 May. The instrument and time of data collection are included below each image. Note: MYDIS refers to the MODIS instrument on the Aqua satellite.

Fig. 8.

Common SST fields from different sensors: (a) night of 25 May, (b) night of 22 May, (c) day of 23 May, and (d) three GOES-9 fields for different times on 23 May. The instrument and time of data collection are included below each image. Note: MYDIS refers to the MODIS instrument on the Aqua satellite.

Fig. 9.

Histograms of differences in the SST fields shown in Fig. 8. Note: MYDIS refers to the MODIS instrument on the Aqua satellite.

Fig. 9.

Histograms of differences in the SST fields shown in Fig. 8. Note: MYDIS refers to the MODIS instrument on the Aqua satellite.

a. Night of 25 May

1) Latitude 14.25°–15.75°S, longitude 139.50°–141.00°E

The AVHRR data are for approximately 0300 LT, with AATSR and GLI being four hours earlier. All three images show cooler water in the southeast of the area. The two difference histograms show a sharp peak, indicating low variations in the SST differences. In both comparisons the standard deviation of the differences is 0.13 K. The AVHRR–AATSR difference peak is close to 0.2 K, which is expected as the AATSR algorithms are derived to give the radiometric (skin) temperature, while the AVHRR algorithm gives the bulk temperature. The GLI shows a warm bias compared to AVHRR, in agreement with the results in Fig. 5 and Table 2. The minor striping in GLI that is aligned close to lines of latitude is due to scan mirror and detector array effects in the instrument. Some secondary diagonal striping is thought to be due to electronic noise in the instrument electronics. Both striping effects are minor anomalies, and the quality of the GLI SST images is very high—especially when averaged over a small array of 1-km pixels. Further details of these phenomena are given by Kurihara et al. (2004).

b. Night of 22 May

1) Latitude 13.75°–15.25°S, longitude 139.50°–141.00°E

The AVHRR, MYDIS, and AMSR-E data are all for near 0200 LT. (N.B., here the NASA notation is adopted with MYDIS referring to the MODIS instrument on Aqua.) The difference histograms show MYDIS to be 0.4 K warmer than AVHRR, and AMSR-E cooler by 0.02 K. These nighttime biases are somewhat different to those given for combined day and night in Table 2. The cause of the AVHRR–MYDIS discrepancy is not known at the moment but may be related to the AVHRR data being near the swath edge, while the MYDIS data are close to 750 km off the subsatellite track, but extra datasets would need to be collected to confirm this possibility. The large area pixels for AMSR-E are obvious in the image. The mirror and detector striping of the MYDIS descending node data are also evident in the image. Further details of the image striping are given by Antonelli et al. (2004). As with the GLI striping discussed above, the color scale is rather stretched and the striping anomalies are quite small.

c. Day of 23 May

1) Latitude 14.25°–15.75°S, longitude 139.00°–140.50°E

The daytime AVHRR, MYDIS, and GOES-9 data displayed here are all for between 1400 and 1430 LT. All three images show a warm feature in the northeast with a weak cooling trend toward the southwest. The difference histograms show peaks with absolute values of less than 0.1 K. MYDIS is slightly cooler than AVHRR and GOES-9 warmer. The warmer temperatures in the northern and eastern extremes of the GOES-9 image are only partly evident in the other instruments.

d. Day of 23 May

1) Latitude 14.25°–15.75°S, longitude 139.00°–140.50°E

The final three images are all from daytime measurements with the geostationary GOES-9. The first image at noon local time shows temperatures close to 302.5 K over the entire area, with a small section in the northeast warmer than 303.0 K. Two hours later, shown in the central image, temperatures are lower, nearer 302.0 K, except in the extreme north and east of the image. By 1800 LT, shown in the third image again, temperatures are lower over the entire area. One possible cause for these variations in temperature is the presence of some diurnal solar heating of the near-surface layer during a period of low wind speeds in the morning followed by an increase of wind speed during the early afternoon that progressed from the south of the area to the north. The middle image thus shows a wind-mixed surface layer except in the extreme north. The last image suggests that the wind-mixing had occurred over the entire area by the evening. During this day the R/V Southern Surveyor was operating some 160 km south of the southern edge of this area, and the wind measurements at the vessel confirmed a low wind speed near noon but showed no increase in strength during the afternoon.

5. Discussion

The analysis in this paper has shown that all of the nine satellite instruments investigated provide SST with a reasonable accuracy. The digitization of the radiance signal in the satellite data stream partly limits the accuracy of the derived SST using standard split-window algorithms (see, e.g., Dudhia 1989). GLI and MODIS data have 12-bit digitization, and comparisons with the ship measurements show standard deviations of between 0.35 and 0.45 K. The number of matchups for the ATSR instruments in this study is small, but other reports by Mutlow et al. (1994), Corlett et al. (2006), and O’Carroll et al. (2006) confirm that these sensors provide SST estimates with a standard deviation of less than 0.3 K when compared with in situ data. The AVHRR provides 10-bit data, and the standard deviations found in this analysis are close to the single-pixel theoretical limit of 0.5 K (Pearce et al. 1989). The 8-bit data of the MSI on GOES-9 severely limits the usefulness of the data in many applications because significant temporal and spatial averaging is required to get reasonable agreement with surface measurements. This places a severe limit on the usefulness of current geostationary observations, and future geostationary instruments must have at least 10-bit data streams, preferably 12 bit. The number of matchups between AMSR-E and the thermosalinograph in this study is small, but the derived SST accuracy is close to that reported by Wentz et al. (2003).

The satellite–satellite intercomparisons show evidence of detector (and mirror side) striping when the images are stretched. If future sensors are developed using this latest detector technology, then improved techniques for data analysis are required. This may require regional algorithms or a dynamic system for removing this extra source of error.

The results given in this paper suggest a way forward in developing steps to provide a blended SST product from the abundance of data that is currently available.

  • As mentioned above, the ATSR series of instruments have been specifically designed to provide accurate measurements of SST. With a dual-view capability, 12-bit digitization, accurate onboard calibration targets, and detectors cooled to 80 K, these instruments provide the most accurate satellite-derived SST in cloud-free conditions. The measurements reported in this paper, and the other reports mentioned above, support this claim. However, the narrow swath and 35-day repeat orbit limits their coverage in low latitudes. These data will best be used to validate (and perhaps to adjust) the more frequent SST fields provided by wide-swath instruments such as AVHRR and MODIS. This may not be as easy as it first appears, as the AATSR algorithms are theoretically derived to give a surface skin SST, while AVHRR and MODIS algorithms are derived through a regression analysis of satellite infrared brightness temperatures and in situ bulk SST data. Comparisons between these two different measurements of SST thus require an estimate of the skin–bulk temperature difference. Donlon et al. (2002) provide a simple empirical formulation that only applies when wind speeds are greater than 6 m s−1. More complicated formulations, such as that of Fairall et al. (1996), require ancillary data that are not always easily obtained.

  • The validated (adjusted) AVHRR and/or MODIS data can next be used to provide an SST in cloud-free regions. The comparisons in Figs. 8 and 9 show that the AVHRR on NOAA-16 gives excellent measurements of SST over a wide swath. In contrast, the in situ comparisons for the NOAA-12 AVHRR demonstrate that sensors can degrade with age and that continuous monitoring, validation, and intercomparisons are required to ensure that each sensor is in good health. For AVHRR and other wide-swath sensors it is important that composite SST fields over several days be developed to provide useful estimates in cloudy regions.

  • The comparisons between AVHRR and the MODIS instrument on Aqua in sections 4b and 4c show a difference between the night and day estimates. The time difference between the measurements in each case was less than an hour, and good agreement is obtained during the day. However, at night the MODIS data are colder by 0.4 K, and there is no obvious reason why this difference is so large. These one-off comparisons show that decisions on how to best blend the data should not be made on small samples; rather, it is important to collect datasets over different seasons, climates, and other conditions to fully understand any differences in the individual datasets.

  • For areas in which there is persistent cloud cover, and infrared estimates are not available for some time (say, greater than days), then microwave measurements should be used to complete a global dataset.

  • Geostationary data should be used in two ways: first, to confirm that anomalies in the infrared and microwave data are due to diurnal solar heating of the upper ocean layers in low-wind situations and, second, to provide input to short-term (<6 hourly) skin SST data products if these are required. The comparisons in section 4d demonstrate the importance of hourly estimates in monitoring the diurnal variations in SST. Such monitoring will be of great assistance in using daytime satellite data to provide accurate estimates of bulk SST under light-wind conditions.

  • In situ measurements from ships, mooring, and buoys should be continually used to validate the blended satellite–SST products as well as assessing the performance of each satellite instrument. The GHRSST-PP will provide many Diagnostic Datasets (DDSs) over selected geographical areas as well as a Matchup Database (MDB) of combined satellite and surface data for these purposes. Typically, a DDS covers an area of 2° latitude by 2° longitude that is centered on a region of interest (e.g., a mooring, a validation site, an anomalous SST location). Over this area a DDS data file is generated for each satellite dataset with SST estimates on a grid of 0.01° in latitude and longitude. The results from the SST comparisons undertaken in this study demonstrate how the GHRSST DDS data files may be used in the development and validation of future blended global SST datasets.

Finally, any scheme that is developed for blending data must be sufficiently flexible to account for loss of sensors and the addition of new instruments.

6. Conclusions and recommendations

The analysis of the in situ and satellite datasets obtained during a voyage of the R/V Southern Surveyor has provided an assessment of each satellite instrument’s estimates of SST in the study region. Comparisons of different satellite SST datasets over small geographic areas demonstrate how the GHRSST-PP DDS can be used to continually monitor the quality of individual and blended SST estimates. The analysis also gives a guide to future development of blending techniques to assemble a single global SST field using all available satellite and in situ measurements. It also identifies some shortcomings in the existing framework that should be addressed to assist with the development of the GHRSST program. These are listed below.

  • Future geostationary meteorological satellites should include sensors with split-window infrared channels at wavelengths of 11 and 12 μm. The instruments should have a spatial resolution of 5 km or less and provide data with 10- or 12-bit digitization.

  • With their two views, 12-bit digitization and accurate onboard calibration, the ATSR instruments provide the most accurate SST measurements from space. There is a concern within the wider SST and climate community that currently there are no approved plans for a follow-on instrument in the ATSR series. Data from such an instrument are vital in maintaining the absolute accuracy of future blended SST products.

  • Improved detector-stripe removal methods are required for instruments that use detector arrays.

  • Future microwave instruments with better spatial resolution would provide valuable data close to coastlines. The current restriction of more than 100 km from land limits the usefulness of this valuable data source.

  • Further research is required to enable better estimates of skin–bulk temperature differences and their dependence on short-term surface wind history and insolation. Future geostationary satellites with 12-bit infrared sensors should provide valuable data in this research. In the meantime, comparisons between AATSR and AVHRR skin and bulk SST measurements would be simplified if AATSR bulk SST algorithms were developed in the same manner as those for AVHRR. Of course, theoretical AVHRR skin SST algorithms should also be developed at the same time.

  • Following on from the last point, SST algorithms that are developed from a regression analysis of brightness temperatures with in situ bulk SST measurements should be done only with data collected at wind speeds greater than 5 m s−1. Inclusion of diurnal heating events in the derivation of these regression-based algorithms currently provides errors in bulk SST estimates that can be avoided. As a corollary, such algorithms should only be applied to satellite data when the local wind speed is above this threshold. Therefore, estimates of wind speed from satellite or other sources (including numerical model forecasts) are a vital ingredient in future satellite-derived SST data analyses.

Acknowledgments

Pamela Brodie and Stephen Thomas operated the DAR011 radiometer during the R/V Southern Surveyor voyage and have provided an excellent dataset. The satellite data have come from a variety of sources: AVHRR from the Australia Centre for Remote Sensing (ACRES), AATSR from ESA, ATSR-2 from the Rutherford Appleton Laboratory (United Kingdom), MODIS from NASA, AMSR-E from Remote Sensing Systems (California), GLI from JAXA, and the MTI from the Australian Government Bureau of Meteorology. All these agencies and institutes are acknowledged for their assistance in data provision. Ken Suber and Chris Rathbone provided the SST analyses from the MODIS and AVHRR data. Part of this work was funded by JAXA through ADEOS-II Contract A2GCF003.

REFERENCES

REFERENCES
Antonelli
,
P.
,
M.
di Bisceglie
,
R.
Episcopo
, and
C.
Galdi
,
2004
:
Destriping MODIS data using IFOV overlapping. Proc. Geoscience and Remote Sensing Symp. (IGARSS’04), Anchorage, AK, IEEE, 4568–4571
.
Barton
,
I. J.
,
P. J.
Minnett
,
K. A.
Maillet
,
C. J.
Donlon
,
S. J.
Hook
,
A. T.
Jessup
, and
T. J.
Nightingale
,
2004
:
The Miami2001 infrared radiometer calibration and intercomparison. Part II: Shipboard results.
J. Atmos. Oceanic Technol.
,
21
:
268
283
.
Bennett
,
J. W.
,
1998
:
CSIRO single channel infrared radiometer—Model DAR011. CSIRO Atmospheric Research Internal Paper 8, 22 pp
.
Corlett
,
G. K.
,
Coauthors
2006
:
The accuracy of SST retrievals from AATSR: An initial assessment through geophysical validation against in situ radiometers, buoys and other SST data sets.
Adv. Space Res.
,
37
:
764
769
.
Donlon
,
C. J.
,
P. J.
Minnett
,
C.
Gentemann
,
T. J.
Nightingale
,
I. J.
Barton
,
B.
Ward
, and
M. J.
Murray
,
2002
:
Toward improved validation of satellite sea surface skin temperature measurements for climate research.
J. Climate
,
15
:
353
369
.
Dudhia
,
A.
,
1989
:
Noise characteristics of the AVHRR infrared channels.
Int. J. Remote Sens.
,
10
:
637
644
.
Fairall
,
C. W.
,
E. F.
Bradley
,
D. P.
Rogers
,
J. B.
Edson
, and
G. S.
Young
,
1996
:
Bulk parameterization of air–sea fluxes for Tropical Ocean–Global Atmosphere Coupled Ocean–Atmosphere Response Experiment.
J. Geophys. Res.
,
101
:
3747
3764
.
Kilpatrick
,
K. A.
,
G. P.
Podesta
, and
R. H.
Evans
,
2001
:
Overview of the NOAA/NASA Pathfinder algorithm for sea surface temperature and associated matchup database.
J. Geophys. Res.
,
106
:
9179
9198
.
Kurihara
,
S.
,
H.
Murakami
,
K.
Tanaka
,
T.
Hashimoto
,
I.
Asanuma
, and
J.
Inoue
,
2004
:
Calibration and instrument status of ADEOS-II Global Imager.
Proc. SPIE,
,
5234
:
11
19
.
Masuda
,
K.
,
T.
Takashima
, and
Y.
Takayama
,
1988
:
Emissivity of pure and sea waters for the model sea surface in the infrared window regions.
Remote Sens. Environ.
,
24
:
313
329
.
May
,
D. A.
and
W. O.
Osterman
,
1998
:
Satellite-derived sea surface temperature: Evaluation of GOES-8 and GOES-9 multispectral imager retrieval accuracy.
J. Atmos. Oceanic Technol.
,
15
:
788
797
.
McMillin
,
L. M.
and
D. S.
Crosby
,
1984
:
Theory and validation of the multiple window sea surface temperature technique.
J. Geophys. Res.
,
89
:
3655
3661
.
Mutlow
,
C. T.
,
A. M.
Zavody
,
I. J.
Barton
, and
D. T.
Llewellyn-Jones
,
1994
:
The Along Track Scanning Radiometer (ATSR) on ESA’s ERS-1 satellite—Early results and performance analysis.
J. Geophys. Res.
,
99
:
22575
22588
.
O’Carroll
,
A. G.
,
J. G.
Watts
,
L. A.
Horrocks
,
R. W.
Saunders
, and
N. A.
Rayner
,
2006
:
Validation of the AATSR Meteo product sea surface temperature.
J. Atmos. Oceanic Technol.
,
23
:
711
726
.
Pearce
,
A. F.
,
A. J.
Prata
, and
C. R.
Manning
,
1989
:
Comparisons of NOAA/AVHRR-2 sea surface temperatures with surface measurements in coastal waters.
Int. J. Remote Sens.
,
10
:
37
52
.
Prata
,
A. J.
,
R. P.
Cechet
,
I. J.
Barton
, and
D. T.
Llewellyn-Jones
,
1990
:
The Along Track Scanning Radiometer for ERS-1—Scan geometry and data simulation.
IEEE Trans. Geosci. Remote Sens.
,
28
:
3
13
.
Rice
,
J. P.
,
Coauthors
2004
:
The Miami2001 infrared radiometer calibration and intercomparison. Part I: Laboratory characterization of blackbody targets.
J. Atmos. Oceanic Technol.
,
21
:
258
267
.
Sakaida
,
F.
,
K.
Hosoda
,
M.
Moriyama
,
H.
Murakami
,
A.
Mukaida
, and
H.
Kawamura
,
2006
:
Sea surface temperature observation by Global Imager (GLI)/ADEOS-II—Algorithm and accuracy of the product.
J. Oceanogr.
,
62
:
311
319
.
Saunders
,
R. W.
and
K. T.
Kriebel
,
1988
:
An improved method for detecting clear sky and cloudy radiances from AVHRR data.
Int. J. Remote Sens.
,
9
:
123
150
.
Weinreb
,
M. P.
,
M.
Jamieson
,
N.
Fulton
,
Y.
Chen
,
J. X.
Johnson
,
C.
Smith
,
J.
Bremer
, and
J.
Baucom
,
1997
:
Operational calibration of GOES-8 and -9 imagers and sounders.
Appl. Opt.
,
36
:
6895
6904
.
Wentz
,
F. J.
,
C.
Gentemann
, and
P.
Ashcroft
,
2003
:
On-orbit calibration of AMSR-E and the retrieval of ocean products. Preprints, 12th Conf. on Satellite Meteorology and Oceanography, Long Beach, CA, Amer. Meteor. Soc., CD-ROM, P5.9
.
Závody
,
A. M.
,
C. T.
Mutlow
, and
D. T.
Llewellyn-Jones
,
1995
:
A radiative transfer model for sea surface temperature retrieval for the along-track scanning radiometer.
J. Geophys. Res.
,
100
:
937
952
.

Footnotes

Corresponding author address: Dr. I. J. Barton, CSIRO Marine and Atmospheric Research, P.O. Box 1538, Hobart, Tasmania 7001, Australia. Email: ian.barton@csiro.au