Vogel, RL, Privette, JL, Yu, Y (2008). Creating Proxy VIIRS Data From MODIS: Spectral Transformations for Mid- and Thermal-Infrared Bands. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 46(11), 3768-3782.
Prior to the launch of a new satellite, simulated sensor data are often desired to develop and test the new sensors and algorithms. Ideally, these data closely approximate the data that will be collected on-orbit. Although radiative-transfer models can be employed for this purpose, all models have biases, and none can completely mimic the complex heterogeneity of Earth's environmental system. An alternative approach is to derive proxy data sets by transforming real observations collected from past or current sensors. Proxy data inherently contain both natural Earth radiation characteristics and sensor noise as the data from the new sensor will. In preparation for the National Polar-orbiting Operational Environmental Satellite System (NPOESS) and NPOESS Preparatory Project missions, we developed a methodology to create proxy data for the mid- and thermal-infrared bands of the Visible-Infrared Imager-Radiometer Suite (VIIRS). Specifically, by combining radiative-transfer modeling and data from NASA's Atmospheric Infrared Sounder (AIRS), we developed spectral transformation equations to convert real Moderate Resolution Imaging Spectroradiometer (MODIS) data into proxy VIIRS data. The functional forms of the equations were determined through regression analysis. Typically, the best spectral transformation equation for a given VIIRS band was a function of multiple MODIS bands, sensor/solar geometry, and surface type. All transformation equations are for clear-sky conditions. Our daytime midinfrared transformation equations have an accuracy that is below the predicted sensor noise for all surface types. Our thermal-infrared equations over land are most accurate for vegetated covers; our ocean equations are accurate for most bands. Validation of this approach with the Advanced Very High Resolution Radiometer suggests that this method may provide higher accuracy proxy data than other methods. Although the advantage in using AIRS is its hyperspectral design, allowing simulation of MODIS and VIIRS bands, its coarse spatial resolution presented a disadvantage in identifying pure land-cover and cloud-free pixels for generating the equation coefficients. Our primary intent with this paper is to offer a methodology for consideration by other sensor teams. Our provisional MODIS-to-VIIRS spectral transformation equations are included for some example surface types.