MODIS + ASTER = CAMEL

Networking and making connections at conferences isn’t unusual. And in that regard, the International TOVS Study Conferences (ITSC) organized by the International TOVS Working Group (ITWG) are no different. However, attending the 16th ITSC in Angra dos Reis, Brazil in May 2008 proved pivotal for University of Wisconsin-Madison Space Science and Engineering Center (SSEC)/Cooperative Institute for Meteorological Satellite (CIMSS) scientist Eva Borbas.

SSEC scientist Eva Borbas. Credit: Bill Bellon

Following her presentation on the land surface emissivity global database that she had developed in collaboration with late SSEC/CIMSS colleague Suzanne Wetzel Seemann, Roger Saunders of the United Kingdom’s Met Office tracked Borbas down to discuss incorporating it into the UK’s radiative transfer model (RTTOV); RTTOV is widely used in Europe in numerical weather prediction (NWP) to assimilate satellite data. More accurate land surface emissivity leads to more accurate temperature and water vapor retrievals and radiance simulation over clear skies.

The UW–Madison emissivity database (UWIREMIS) was initially included in RTTOV, version 10, released in January 2011. Because of his familiarity and experience with RTTOV, Ben Ruston of the Naval Research Laboratory (NRL) was instrumental in helping Borbas integrate and test the emissivity database. This collaboration between SSEC, NRL, and the Met Office was supported by EUMETSAT’s NWP Satellite Application Facility (NWP-SAF).

In 2017 with help from SSEC/CIMSS and NASA Jet Propulsion Laboratory colleagues, Glynn Hulley and Simon Hook, Borbas provided a significant upgrade for the most recent RTTOV release, version 12. They replaced the UWIREMIS database — which uses laboratory measurements and data from the MODerate-resolution Imaging Spectroradiometer (MODIS) — with the Combined ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) and MODIS emissivity over Land (CAMEL) product. According to Borbas, merging the UW database with JPL’s ASTER emissivity database allowed them to highlight their advantages, leading to an improved emissivity product.

CAMEL Broad Band Emissivity (between 3.6 and 14.1 μm) for September 2009 (left) and its deviation (right) from the constant value of 0.98. Setting BBE as a constant value over the globe is still common practice in land surface or radiative models. Differences larger than 5% are seen over non-vegetated areas like the Saharan Desert and Arabian Peninsula. Credit: Eva Borbas

“ASTER is more accurate in the 8.3–12 micron region, especially on non-vegetated areas, and MODIS is more accurate on the vegetated areas. In addition, MODIS provides emissivity values in the midwave IR region while ASTER does not,” stated Borbas.

Despite the complementary nature of the two datasets, combining them required navigating a few challenges, some of which they are still resolving – such as how this monthly database can accurately characterize emissivity in areas with snow coverage that is not constant throughout the month.

Borbas and SSEC/CIMSS colleagues, Michelle Feltz and Bob Knuteson, are continuing to improve the CAMEL product, in particular the error uncertainty estimates. They are taking into account improvements in spatial resolution from 50km to 5km, as well as providing the user with more flexibility and control in managing the uncertainty estimates.

This work was supported by EUMETSAT’s NWP Satellite Application Facility (NWP-SAF).

By Leanne Avila