Joint Rosenstiel School of Marine & Atmospheric Science (RSMAS)
Committee on Earth Observation Satellites (CEOS) Validation Workshop

-- Bob Kannenberg (rkannenb@pop900.gsfc.nasa.gov), Science Systems & Applications, Inc.
-- Frank Palluconi (frank.d.palluconi@jpl.nasa.gov), Jet Propulsion Laboratory

Participants

The workshop was held at RSMAS March 5 and 6 and was chaired by Ian Barton (CSIRO). Participants included Otis Brown (RSMAS), Peter Minnett (RSMAS), Walt McKeown (NRL), Craig Donlon (CCAR, University of Colorado), Jennifer Hanafin (RSMAS), Goska Szczodrak (UBC, Vancouver), Andy Jessup (Applied Physics Lab [APL], University of Washington), Gary Wick (University of Colorado), Andy Harris (UK Met Office), E. Theocharous (NPL), Fred Prata (CSIRO), Tim Nightingale (Rutherford Appleton Lab), Tom Sheasby (University of Leicester), Jim Butler (NASA/GSFC), Bob Evans (RSMAS), Frank Palluconi (JPL), Carol Johnson (NIST), and Bob Kannenberg (NASA/GSFC).

Introduction

Barton welcomed participants and announced that this workshop is intended to be a forum in which the various groups involved in the validation of satellite measurements of surface temperatures can discuss sharing data and participating in one another's field activities. He encouraged participants to look beyond their respective areas of expertise when considering inter-comparison activities (i.e., how can the validation community most effectively work together as a whole?).

Objectives and Outputs

Barton cited the following objectives and outputs for the workshop:

SST Validation

Advanced Along Track Scanning Radiometer (AATSR)

Nightingale reported that the AATSR instrument, the third in the ATSR series, is to be a payload instrument on ESA's ENVISAT-1 polar-orbiting mission (scheduled for launch in 1999). AATSR has the same signal channels and embodies exactly the same viewing principle as ATSR-2 (i.e., thermal channels at 3.7, 10.8, and 12-µm wavelengths, and reflected visible/near infrared (IR) channels at 0.555, 0.659, 0.865, and 1.61-µm wavelengths). The main objective of AATSR is to contribute to the long-term climate record of global SST by extending the current ATSR-1 and -2 global data sets well into the next decade. Nightingale stated that requests for AATSR data must be made through ESA. McKeown asked what AATSR's approximate SST accuracy will be, and Nightingale replied that it will be 0.3 K. Sheasby provided an overview of the AATSR Validation Plan, available online at: http://www.le.ac.uk/physics/research/eos/aatsr/val1.html.

The group debated the merits of providing raw data to the validation community at large. Brown cited the example of brightness temperatures, where the original counts are also provided. In the future somebody may come up with a better way to compute the brightness temperatures from the counts so, from that standpoint, it makes sense to provide the counts as well. Barton asked that AATSR and other instruments publish detailed validation campaign plans on the Web, to facilitate inter-comparison of data; most instruments currently have some kind of plan available. Brown added that the information should include not only where and when campaigns will be held, but what kinds of products and measurements will be involved, and how these can be obtained. An exchange needs to be mediated somewhere as to how data changes hands, and CEOS is probably the right place to begin this discussion. Brown stressed that the IR community needs to look at long-term cross-comparison and cross-validation plans so as to have consistent data sets for use by future environmental scientists.

MODIS

Minnett reported that the MODIS IR SST Validation Plan is available online at: http://www.rsmas.miami. edu/modis. Minnett pointed out that MODIS includes both land and atmosphere components, and the MODIS Ocean group will be collaborating with the Atmosphere group (ER-2 flights, etc.) as well as other instrument teams. The MODIS SST Validation Plan is really a validation of the atmospheric correction. The plan calls for comparison of "like with like" (i.e., IR radiometry from ships and aircraft). Measurements will be taken with the Marine-Atmosphere Emitted Radiance Interferometer (M-AERI), band-pass radiometers, and conventional in situ sensors (buoys). Issues for analysis include regional and seasonal effects, the skin effect, and diurnal heating. Minnett listed the objectives of MODIS field programs as follows: use M-AERI; take ancillary measurements in the atmosphere and ocean; make measurements in varied regions; study ocean thermal skin effects; improve robustness of SST retrieval (e.g., variations in water vapor and aerosols); and validate SST retrievals. Upcoming M-AERI cruises include a voyage from Seattle to either New Zealand or Tasmania in Fall 1998. Minnett indicated that the results of the previous three-day IR radiometer workshop (refer to summary on page 51) will help to determine the type of band-pass radiometer selected by MODIS to supplement M-AERI measurements.

Donlon noted that so far discussion has not touched on a sea state measurement. Currently this is a very subjective measurement, perhaps taking the form of a half-hourly visual report from a ship's bridge. McKeown stated that an NRL scientist has developed an algorithm to calculate sea state in the visible rangeperhaps this can be modified and applied to IR?

Global Imager (GLI)

Barton briefly discussed the National Space Development Agency's (NASDA) GLI validation plans, as NASDA representatives were unable to attend the workshop. GLI is very similar to MODIS in both design and validation strategy, and there has been a good deal of synergy between the two instrument teams. GLI validation activities will be conducted from test sites in the Bering Sea and the Japan Sea, as well as coastal sites in Japan. The Japanese fishing industry has so far provided strong support for validation activities. Bulk SST (BSST) and skin SST (SSST) measurements are among the highest priorities for GLI validation.

CSIRO

Barton indicated that CSIRO is working with MODIS, AATSR, and GLI. CSIRO has placed instruments aboard commercial ferry boats (Townsville and Perth) in order to collect daily data over a fixed area. CSIRO also plans to place instruments aboard ships of opportunity operating in Australian waters. Barton presented examples of ATSR and AVHRR data, looking at the skin/bulk temperature difference. In situ data taken by CSIRO (using both a radiometer and a thermosalinograph) compared well with the satellite data. Barton noted that so far modellers have relied primarily on BSST measurements, but the true SST picture is much more complex than that, and requires an algorithm that factors in the SSST.

Land Surface Temperature (LST) Validation

CSIRO

Prata stated that he has been asked to derive IR LST algorithms for both AATSR and GLI. The LST algorithm will be a regression-based algorithm that will factor in 16 types of existing land classifications, thus bypassing the surface emissivity problem. (Prata will not derive a snow or ice algorithm.) He explained that LST is not really a skin temperature, as radiation is being emitted from a variety of sources (deep within the canopy, leaves, ground, etc.). There are also angular effects to consider. Analysis of ATSR-2 data shows that there can be significant variations in emissivity depending on viewing angle. Prata presented some ATSR images taken over Australian validation sites, as well as visible LST data collected in April 1997 during a collaborative effort between CSIRO and the University of Nottingham. The purpose of this activity was to validate ATSR-2 spectral radiance measurements. Overall ground measurements compared well with the ATSR-2 measurements.

Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER)

Palluconi explained that the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) validation effort focuses on LST. ASTER will validate the radiance at sensor (Level 1), and then the surface leaving radiance (Level 2). ASTER consists of three separate instruments, or components: Visible Near Infrared (VNIR), Shortwave Infrared (SWIR) and Thermal Infrared (TIR). There will be no space-view calibration for ASTER. Jessup stated that he may be interested in obtaining some high-resolution ASTER data, and he will discuss this with Palluconi. Primary ASTER validation sites are Lake Tahoe, Salton Sea, Railroad Valley, and White River Valley. Barton suggested that ASTER could benefit greatly from collaborative validation efforts with MODIS. Prata noted that ASTER could use SST data to help validate its atmospheric correction measurements. Palluconi indicated that six months after launch anybody can request ASTER data, and he encouraged participants to do so.

University of Washington APL

Jessup reported that APL will participate in a NOAA cruise in May and June of 1998. He hopes to have his prototype radiometer unit built by July 1998 and then conduct testing from the Hood Canal Bridge and the ferry in Puget Sound. Platform Harvest in 1999 may be another possible testing site. Jessup is also interested in taking measurements in the Gulf of Mexico in 1999, and this is the experiment for which he would like to obtain high-resolution ASTER data. Minnett suggested that Jessup may want to investigate putting a radiometer package aboard one of the ships that travels between Seattle and Yokohama.

Radiometer Designs

Ship of Opportunity Sea Surface Temperature Radiometer (SOSSTR)

Donlon asserted that there is presently a serious lack of geographically widespread and temporally dense in situ SST measurements. This hinders the validation, long-term stability, and interpretation of SST satellite measurements. It also hinders the development of skin-to-bulk temperature transfer models. Donlon cited the high cost and limited availability of extremely accurate radiometers as another limiting factor for in situ SST measurements. He presented data gathered with Tasco THI-500L radiometers, which are inexpensive but require constant calibration. When compared at sea to the precisely calibrated sea surface temperature radiometer (SISTeR), the Tascos provided good data even without protection from the elements, although the data did not reach the desired 0.1°C accuracy . To improve the accuracy of the Tascos, Donlon has housed two of them within an enclosed (but not sealed) unit called the SOSSTR. One radiometer looks at the sea, while one looks at the sky. SOSSTR also contains two blackbodies, one ambient and one hot. Donlon believes that the SOSSTR units can be constructed inexpensively enough to be practical for a large deployment.

Proposed University of Washington Applied Physics Lab (APL) Radiometer Design

Jessup reviewed his proposed radiometer design, which calls for enclosing the sensors and blackbodies within a sealed, temperature-controlled housing. (The radiometers inside the unit would be Heimann KT-15 models.) The housing would be filled with dry nitrogen, and the sensors would look out through an IR-transparent window, equipped with a wiper and fluid reservoir to periodically clean salt, spray, etc. Harris asserted that adding the window means that there is no true end-to-end calibration. Jessup acknowledged this, but said that if he can find an IR-transparent window that can be very accurately characterized, the protection it would afford the sensors could more than account for the added uncertainty introduced. He added that he may be over-engineering in response to the worst-case scenario (e.g., a huge wave hits the unit), and asked participants for their feedback. Barton and Donlon suggested that the error introduced by the window may well eat up the 0.1° C accuracy, but agreed that the window is worth checking out so long as the design of the unit allows the window to be removed later. Jessup has begun investigating IR window materials (e.g., zinc selenide), and will look at testing windows and the effects of various cleaners.

JPL ASTER Radiometer

Palluconi explained that the ASTER unit contains an Everest 4000.4GL radiometer, housed within a very well-insulated box. The attempt to isolate the radiometer passively with the insulation appears to have caused temperatures to drift up significantly.

Scanning Infrared Sea Surface Temperature Radiometer (SISTeR)

Nightingale described the SISTeR, developed at Rutherford Appleton Lab. SISTeR is based on an ellipsoid mirror that allows a small foreoptics window. This permits smaller internal blackbody calibration targets and a narrow exit slit, which protects against the elements. The radiometer's optical path is chopped at 100 Hz, and the radiation is passed through one of six possible filters before arriving at a DLTGS detector. The chopper is close to the detector, and the filter is between chopper and detector. Barton pointed out that the chopper could change temperature when the hot and cold blackbody are viewed. For an SST measurement including a sky scan, the measurement cycle is about 2 minutes, which includes a look at the two blackbodies. In situ calibration with the CASOTS blackbody indicates no bias and a peak-to-peak noise of about 0.1 K.

NRL Buoy Radiometer

McKeown stated that the NRL hopes to install IR radiometers on some 300 buoys to make SSST measurements. There is a proposal to mount a system at 20-25 m with a rain shield, spray shield, and a rotating radiometer (possibly a Tasco model) with a single calibration blackbody. A conical mirror above the radiometer is to be used to feed the radiometer and create a large footprint. A full-sky radiometer will be used to obtain sky irradiance measurement.

Improving Radiometer Design and Testing

Donlon advised that radiometer developers try to keep their units as lightweight as possible. Also, for most point-and-shoot radiometers, a look angle of less than 40 is desirable; otherwise, roll and pitch must be sampled very precisely. Jessup suggested that at the next radiometer comparison workshop a surface disrupter and an imager be used for the measurements taken from the rooftop platform. He indicated that for a future workshop instruments might be mounted on the Floating Instrument Platform (FLIP), which is unique in that it orients itself with the wind and remains very stable. This platform would be safe and stable enough so that the NIST or APL blackbody could be used onboard. Barton stated that one of these two blackbodies, plus an M-AERI, should be incorporated into future workshops and field activities whenever possible.

Surface Temperature Validation Web Site

Barton suggested that a CEOS IR Cal/Val Web page be established to provide a link to the data sets that are maintained by individuals or organizations. Individual Web pages could be established for instrumentation description, contacts, field campaigns (oceans and land, with protocol descriptions), meetings (CASOTS, RSMAS, etc.), and metadata data sets. Evans indicated that RSMAS will have all AVHRR and MODIS data, and for cal/val purposes, they would be willing to do the extractions to match the cal/val data. Barton indicated that the general agreement of the CEOS member agencies is that data for climate research, including cal/val, should be freely available. He added that it may be necessary to respond to the recent ENVISAT AO to ensure timely delivery of data and data of the highest spatial resolution.

Measurement Database

The question of a database of measurements was discussed and Evans offered to use the resources at RSMAS to help set up a Web page to point to the sources of data themselves (especially for the large continuing data sets). He pointed out that some countries do not have a problem with individuals using data; however, where the interest is to provide data to the general community, much more extensive negotiations (i.e., several years) may be required. Barton suggested that there be a separation of data used for cal/val and that used for research, in that cal/val data should be made widely available as soon as it is reduced and understood by the generator. Cal/val data would be released quickly, but if the same data were used for research, CEOS could recommend that the originators be contacted and informed of this use. Minnett proposed a 9-month submission period for data; after 12 months the database administrator would release any data in the database to qualified users.

Conclusion

Participants agreed that it would be useful to hold another IR Instrument Comparison Workshop and CEOS Validation Workshop within the next 1-to-2 years. The next CEOS Cal/Val meeting will be held in July 1998 in Tokyo, and Barton will report on these workshops there. He thanked Bob Kannenberg (NASA/GSFC) and Frank Palluconi (JPL) for providing meeting minutes. He also thanked Otis Brown, Peter Minnett, and RSMAS for hosting the workshops.

Recommendation

As a result of the workshops, Barton will make the following recommendation to the CEOS Plenary:

"The thermal infrared validation community met in Miami during March 1998 for a round-robin instrument inter-comparison and a satellite validation workshop. The meeting agreed to foster close international collaboration in the collection and analysis of ground truth data for the validation of geophysical products derived from thermal infrared satellite data. To ensure that validation data sets are available to the satellite operators in a timely manner, it is recommended that the appropriate satellite data be made available to the ground data providers in near real time and free of any charges. In return, the ground data collectors will provide validation data sets to the satellite community in a timely manner. It is also recommended that both the ground data supplied by the data collectors and the associated satellite data are only initially available for use in satellite instrument calibration and data product validation, and should not be used in any other manner unless agreed to by the data producer."