Impact of Spatial Sampling on Continuity of MODIS–VIIRS Land Surface Reflectance Products: A Simulation Approach

Nima Pahlevan, Sudipta Sarkar, Sadashiva Devadiga, Robert E. Wolfe, Miguel Roman, Eric Vermote, Guoqing Lin, Xiaoxiong Xiong
2017 IEEE Transactions on Geoscience and Remote Sensing  
With the increasing need to construct long-term climate-quality data records to understand, monitor, and predict climate variability and change, it is vital to continue systematic satellite measurements along with the development of new technology for more quantitative and accurate observations. The Suomi National Polar-orbiting Partnership mission provides continuity in monitoring the Earth's surface and its atmosphere in a similar fashion as the heritage MODIS instruments onboard the National
more » ... Aeronautics and Space Administration's Terra and Aqua satellites. In this paper, we aim at quantifying the consistency of Aqua MODIS and Suomi-NPP Visible Infrared Imaging Radiometer Suite (VIIRS) Land Surface Reflectance (LSR) and NDVI products as related to their inherent spatial sampling characteristics. To avoid interferences from sources of measurement and/or processing errors other than spatial sampling, including calibration, atmospheric correction, and the effects of the bidirectional reflectance distribution function, the MODIS and VIIRS LSR products were simulated using the Landsat-8's Operational Land Imager (OLI) LSR products. The simulations were performed using the instruments' point spread functions on a daily basis for various OLI scenes over a 16-day orbit cycle. It was found that the daily mean differences due to discrepancies in spatial sampling remain below 0.0015 (1%) in absolute surface reflectance at subgranule scale (i.e., OLI scene size). We also found that the MODIS-VIIRS product intercomparisons appear to be minimally impacted when differences in the corresponding view zenith angles (VZAs) are within the range of −15 • to −35 • (VZA V − VZA M ), where VIIRS and MODIS footprints resemble in size. In general, depending on the spatial heterogeneity of the OLI scene contents, per-grid-cell differences can reach up to 20%. Further spatial analysis of the simulated NDVI and LSR products revealed that, depending on the user accuracy requirements for product intercomparisons, spatial aggregations may be used. It was found that if per-grid-cell differences on the order of 10% (in LSR or NDVI) are tolerated, the product intercomparisons are expected to be immune from differences in spatial sampling.
doi:10.1109/tgrs.2016.2604214 fatcat:6mn5tnxukjb7nocqefrfjv5llu