A Review of the Thermal Components on

MODIS and ASTER

Julia Barsi

Center for Imaging Science

Rochester Institute of Technology

With NASA’s pledge to provide for long term studies of the Earth and its systems, a number of new instruments are being developed and built to observe changes remotely. Specifically, the difficulties of imaging in the thermal range are slowly being overcome and new sensors are taking advantage of these advances in imaging in the long-wave infrared (LWIR). However, in order for this data to be useful and absolute, calibration efforts must achieve some minimal error never before attained, possible for airborne instruments, difficult for orbiting systems.

Currently, the most interesting new thermal sensors under construction are ASTER and MODIS. Both have multiple thermal channels in the LWIR, something not yet attempted on satellite instruments and only newly used with airborne systems. This paper will summarize the systems proposed for pre-launch and on-orbit calibration. A comparison to what RIT is implementing for Landsat ETM+ (with only one thermal channel) will be made and the advantages and disadvantages of the various plans will be discussed.

Since the of advent public sector remote sensing with the launch of the first Landsat in 1972, studying the Earth and its processes from space has become increasingly more popular and useful for a number of reasons. Monitoring long term effects on a planetary scale becomes difficult to do from the microscopic view we get from the Earth’s surface. Removing the observations to outside the system allows for a less intrusive measurement; the data collector is not affecting the environment being monitored in any way. And the lifetime of a study depends on the stamina of those collecting; wide spread, extensive studies, with many data points requires many man-hours in just collecting the observations.

The Landsat program opened a whole new method of study, allowing scientists to use information in images to retrieve the data that otherwise would have taken difficult days, months, even years to collect. Since the first Landsat, many other instruments have been put into orbit for global monitoring. With the recent concern about global warming, ozone depletion, deforestation, and many other environmental issues, the use of satellite imagery has been brought to the forefront of research methods.

NASA has developed a 18 year international program, the Earth Observing System (EOS), committed to introducing a host of new instruments for the express purpose of monitoring global change (Butler 1997). At present, eleven major platforms are planned, each with between one and five instruments on board through 2008. The first EOS satellite to be put into orbit will be EOS-AM1, in 1998. This platform is a ground breaker; aboard it are five instruments, all taking advantage of technologies not yet used on space-based platforms. Within three years, follow-on instruments will be launched, building on what was learned with AM1.

This paper will focus on two of the EOS-AM1 instruments, MODIS and ASTER. Both are of particular interest due to their multiple thermal channels. EOS-AM1 is scheduled for launch just before the new Landsat instrument. Landsat ETM+ still only has one thermal band but with the Landsat calibration validation being taken on here are RIT and the capabilities of MISI and the calibration facility, multi-spectral thermal imaging is an area of specialization within the Center for Imaging Science.

The EOS-AM1 platform will carry five different instruments (Collier 1997):

• CERES (Clouds and the Earth’s Radiant Energy System) - a scanning bolometer to measure the radiation budget of the Earth, with particular interest in studying how clouds affect the budget.
• MISR (Multiangle Imaging Spectroradiometer) - radiometer measures sunlight reflected by Earth into space
• MOPITT (Measurement of Pollution in the Troposphere) - radiometer using pressure and length modulation to measure chemicals in the lower atmosphere
• LIS (Lightning Imaging System) - a CCD array monitoring only a specific bandpass relevant to lightning
• ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) - radiometer with high spatial, spectral and radiometric resolution
• MODIS (Moderate Resolution Imaging Spectrometer) - radiometer with moderate resolution but the high spectral resolution of any of these instruments

The satellite is schedule for launch in June 1998. It will have an average altitude of 705km in a near polar, sun-synchronous orbit with a 10:30am descending equatorial crossing. It has an orbital period of 98.9 minutes and a repeat cycle of 16 days.

The Moderate Resolution Imaging Spectrometer (MODIS) is the corner stone of the EOS-AM1 platform. It will be the most rigorously calibrated instrument in the EOS fleet and have the highest spectral resolution on a spaceborne system thus far (Barbieri 1997). The imaging spectroradiometer carries 490 detectors in 36 spectral channels covering the visible and infrared, in line scanner design.

The main component of the thermal OBC is a Planckian blackbody, an aluminum v-groove plate with a very high, well known emissivity and precise temperature. The temperature of the blackbody can be related to the radiance by integrating Planck’s function over the bandpass of the individual detector array being calibrated. This results in a radiance for a corresponding temperature, one point on the calibration curve. This blackbody can be varied in temperature between 273K and 315K (Barbieri 1997). Most of the time, the temperature of the blackbody will only be monitored, not controlled in the cavity. Every few weeks the blackbody will heat to 315K to check a high temperature gain and bias.

The second point on the curve comes from the instrument’s deep space look (Barbieri 1997). Since stars are too small to be resolved by the detectors, the space view provides a constant very cold temperature of 4K (except for the few times a year when the moon is in the view, in which case it will be used for reflective calibration). Converting this to a radiance using Planck’s function provides a low signal on the calibration curve.

Knowing absolutely what radiance corresponds to what temperature only helps if the pre-launch system characterizations can be trusted. As well as a system can be characterized and calibrated on the ground, putting the instrument into orbit can wreck havoc on the equipment. The shaking and banging can cause unpredictable change. And over the lifetime of the instrument, the components can degrade, individually affecting the system characterization.

To monitor these systematic changes, a ground truth method must be used. If the radiance leaving the ground is known, it can be extrapolated to the radiance reaching the sensor. The calculated at-satellite radiance should be the same as the measured at-satellite radiance, within the error of the extrapolation routine. The method of vicarious calibration chosen to be used for this system is an underflight campaign. A well calibrated airborne imaging system with similar characteristics as MODIS will be put in a aircraft and image the same scene as MODIS is imaging at the same time. Since the error in the airborne system is much less than that of the satellite instrument, the airborne measurements are taken as truth and the OBC can be modified to correct for the difference between calculated and measured.

The selection of targets is of extreme importance in a ground truth vicarious calibration. The target must be large enough to encompass at least three pixels (3000m), preferably all pixels; uniform, lacking any highly varying features that might change the average radiance from the pixel; and have a well behaved emissivity so that the radiance integrated over the bandpass doesn’t have any unexpected variations (Biggar 1996). Also of consideration is the relative ease of getting the team and equipment to the site and the magnitude of the corrections for the atmospheric effects. Such targets are difficult to find. The calibration team has chosen Lake Tahoe in California and Nevada. The lake is a large uniform temperature water surface, at a high elevation.

The altitude of the lake allows for the plane to fly higher, decreasing the amount of atmosphere above the aircraft image. Since the remaining atmosphere has to be accounted for using a radiative transfer code, like MODTRAN, the thinner the atmosphere above the airborne instrument, the less opportunity for error in the extrapolation (Slater 1996). This predicted top of the atmosphere (TOA) radiance can be compared to the radiance MODIS is measuring, according to the calibration curve.

If the two radiances do not match, within the expected error, the instrument calibration can be updated to account for this shift in readings (Biggar 1996). This underflight is scheduled for once a year beginning in the first few months of operation. The periodic updates will keep track of instrument degradation over the lifetime of MODIS.

The Advanced Spaceborne Temperature and Emissivity Resolution Radiometer (ASTER) is the highest combination of spatial, spectral and radiometric resolution instrument in space to date (ASTER 1995). The Japanese imager consists of three separate subsystems, divided into fourteen spectral bands from visible to longwave infrared.

Each thermal band is an array of 10 HgCdTe photoconductor detectors. The five bands are mechanically scanned in the cross-track direction, a standard whiskbroom scanning design (ASTER 1995). The spectral bands are defined by interference filters just in front of the individual detectors. The 12 bit quantization of the LWIR channels increases the dynamic range and precision of the detectors. The ground instantaneous field of view is 90m, considerably better than MODIS.

Since ASTER and MODIS are on the same platform, the same vicarious calibration methods can be used (Ono 1996). Although the instruments are slightly different, this can be taken into account in the calculations of predicted top of the atmosphere radiance. The same airborne will be taking the measurements for two instruments of slightly different spectral characteristics. As long as the airborne instrument has spectral characteristics encompassing the spectral range of the orbiting instruments, the differences can be factored into the integration over bandpass of Planckian spectral exitance.

The onboard calibration is probably the most important component of the entire system. As long as the OBC is working and providing a calibration curve, changes in the optics and detectors can be accounted for. As soon as the OBC stops working, the degradation cannot be measured and only relative measurements can be made. It is surprising to see, then, that both of these new generation instruments are doing less to ensure a thorough OBC than instruments in the past.

The inclusion of only one blackbody on board each system is cause for immediate concern. MODIS attempts to compensate by using the cold space view. However, the temperature of cold space is out of the range of temperatures that the instrument would be seeing from Earth. The calibration curve for MODIS is created over a huge range of temperatures, some of which it will never see. It has already been mentioned that the response of the thermal detectors is not quite linear but more of a slight quadratic shape. Over a small range of temperatures the approximation of linearity is appropriate. Expanding the linearity over an extra 200K unnecessarily is an immediate source of error. The use of a second blackbody, set to a temperature at one end of the range of temperatures seen on Earth, would reduce the uncertainty of the linearity approximation and make the calibration curve more accurate.

ASTER does not even make a pretense of creating a two point calibration curve. ASTER assumes the preflight instrument characterization remains consistent, so the slope of the calibration curve is constant over the life of the mission. The one point calibration will be used to change the intercept of the calibration using the pre-determined slope of the calibration curve. System designers have known for years that pre-flight characterization does not hold for the lifetime of the instrument and therefore calibration procedures are necessary. ASTER is attempting to ignore this fact.

The system of multiple instruments on one platform and multiple platforms provides novel opportunity for system cross calibration. Since MODIS and ASTER will be imaging the same scene at the same instant, a method can be developed to absolutely calibrate the instruments to each other. The MODIS measurement can then be used as a second point on the ASTER calibration curve. However, this still relies on the shaky MODIS thermal calibration.

Other plans for system cross calibration are (Ono 1996):

• same footprint size and spectral bands and image the same scene simultaneously, i.e. the two HRV cameras on SPOT
• similar spectral bands that cover part or all of the same spectral range and image the same scene simultaneously but with different footprint sizes, i.e. ASTER, MODIS and MISR
• same footprint sizes but acquisition at different imaging angles, i.e. the two VNIR stereo cameras on ASTER
• similar spectral bands and footprint sizes but imaging at different times, i.e., ASTER, Landsat and SPOT

These methods will allow all EOS data to be used interchangeably, each system being calibrated to another. However, this is no excuse for poor individual instrument design.

The vicarious calibration and validation of calibration using underflight is a standard method of updating the calibration curve. The procedure has been used in numerous studies and will be used by the DIRS group at RIT to verify Landsat ETM+’s calibration. The calibration team has selected Lake Tahoe as the target. A large body of water is an excellent target due to its high emissivity and thermal mixing. However, they are limiting themselves to a one temperature check of the calibration unless underflight is planned more than once a year. The one point calibration is similar to the problem with ASTER’s calibration; if the slope of the calibration curve has changed, it will not appear in this method.

The target selected for our own calibration verification of Landsat is also a large body of water, Lake Ontario, but the Great Lakes provide a unique thermal opportunity. The thermal bar, a phenomenon of large temperate zone bodies of water, provides a separation of warm and cool water in the spring warming season. Underflying a satellite during this period provides at least two points, perhaps more depending on the resolution of the system, to check the calibration curve. Lake Tahoe does not exhibit this phenomena so makes the calibration validation less complete.

In all the papers written about the planned vicarious calibration of MODIS and ASTER, the underflight thermal imager was not mentioned. The instrument must have similar characteristics to the on orbit instrument. The one fact that was brought up was the recording of the signal on video to assist in registration issues (Biggar 1996). Hopefully, this video is only for cosmetic purposes, not the only method of recording the signal. Video is highly nonlinear and would introduce a host of errors into the process.

The multitude of new science and NASA’s commitment to Earth observation over the next fifteen years is exciting for all in the field of imaging science. New techniques, new technologies, and new systems are being tried out faster than ever before. However, this commitment also includes a "smaller, faster, cheaper" objective, meaning well thought out, extensively characterized systems are obsolete. The system now being put into orbit are quick-and-dirty versions of past glories. ASTER and MODIS are just two examples of this. The systems ideally will provide much needed scientific data for the EOS program but as complete imaging systems, they are lacking. It must be mentioned that MODIS and ASTER are only first generation instruments of their type. Another two MODIS instruments are planned for later missions and the successes and failures of the first instrument will be taken into consideration in developing the second and third. Unfortunately, when MODIS claims to be the most rigorously calibrated satellite system, it is referring to the reflective bands. A major goal for the next 18 years will be getting the thermal standards to where the reflective standards are now. The work being done at RIT is a beginning but multispectral thermal imaging will become increasingly useful as more systems provide more data and hopefully the instrument design and implementation will get increasingly more rigorous.