Can VIIRS continue the legacy of MODIS for near real-time monitoring of tropical forest disturbance?

Abstract

Near real-time monitoring of forest disturbance is a critical tool for combating illegal logging. The only feasible data sources for near real-time monitoring over large area are remote sensing data, and historically the only sensors with repeat times sufficient for near-real time monitoring are from coarse-resolution sensors such as MODIS. The MODIS sensors onboard the Terra and Aqua satellites have provided near daily global observations in the red and NIR spectral bands at 250 m resolution since 1999. While their longevity is impressive, the sensors are nearing the end of their lifespan and the MODIS data record will be continued by the VIIRS sensor onboard Suomi NPP. But because of the spatial and spectral differences between MODIS and VIIRS, it is worth investigating whether VIIRS can replace MODIS as input data for near real-time monitoring. In this article, we compared the performance of the NRT-CCDC (Near Real-Time Continuous Change Detection and Classification) algorithm using the Terra/ and Aqua/MODIS composites at 250 m resolution, Terra/MODIS 250 m and 500 m surface reflectance, Aqua/MODIS 250 m and 500 m surface reflectance, and VIIRS 500 m surface reflectance. We found that the algorithm using composites from both MODIS sensors and Terra/MODIS data at 250 m resolution outperformed the results obtained when using VIIRS data. The results were similar when using 500 m data from Terra/MODIS alone compared to VIIRS data. We found that a morning overpass, like that of Terra, is preferable to an afternoon overpass, which is employed by VIIRS and Aqua. Spatial and temporal resolution of the input data are two key factors in successful near real-time monitoring of forest disturbance.