Multisensor Translation and Continuity of Vegetation Indices Using Hyperspectral Data

The earth surface is monitored periodically by numerous satellite sensors which have different spectral response functions, image acquisition heights, atmosphere correction schemes, overpass times, and sun/view angle geometries. Temporal and spatial variations of land surface properties, such as vegetation index, Leaf Area Index (LAI), land surface temperature, and soil moisture, have been provided by long-term time series of various remote sensing datasets. Inter-sensor translation equations are required to build long-term time series by the combination of multiple sensors from historical to advanced and new satellite datasets. In the first chapter, inter-sensor translation equations of band reflectances and two vegetation indices (e.g. Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI)) were derived using linear regression equations relative to Moderate Resolution Imaging Spectroradiometer (MODIS) values. The consistency and validation of inter-sensor transforms were investigated through statistical student's t-test and the root mean square error (RMSE).In the second chapter, cross-sensor extension of EVI and a 2-band EVI (without the blue band; EVI2) were investigated based on the continuity of both EVI's. Sensor specific red-blue coherencies were examined for the possibility of the EVI and EVI2 extension from MODIS sensor. The EVI continuity to MODIS was particularly problematic for the Visible Infrared Imager / Radiometer Suite (VIIRS) and the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) that have dissimilar blue bands from that of MODIS. The cross-sensor extension and compatibility of EVI2 were improved and provided the possibility to be lengthened to the Advanced Very High Resolution Radiometer (AVHRR) using its translation equation.Finally, we evaluated the use of sensor-specific EVI and NDVI data sets, using a time sequence of Hyperion images over Amazon rainforest in Tapajos National Forest, Brazil for the 2001 and 2002 dry seasons. We computed NDVI, EVI, and EVI2 with the convolution data of different global monitoring and high temporal resolution sensor systems (AVHRR, MODIS, VIIRS, SPOT-VGT, and SeaWiFS) from Hyperion, and evaluated their spectral deviations and continuity in the characterization of tropical forest phenology. Our analyses show that EVI2 maintains the desirable properties of increased sensitivity in high biomass forests across all sensor systems evaluated