Using AirMISR data to explore moisture-driven land use–land cover variations at the Howland Forest, Maine — A case study

Abstract

Surface moisture stress and vegetation condition can be derived from spectral vegetation indices such as the normalized difference vegetation index (NDVI). The combination of thermal infrared data with these vegetation indices is often plotted as a triangular scatterplot. One such relationship uses the radiant temperatures versus the NIR/blue ratio developed for a semiarid area from Landsat Thematic Mapper data. Given that the thermal range of this index was invalid when applied to a humid temperate region, this study evaluated the utility of using the NIR/blue ratio alone as an indicator of land use/land cover (LULC) separability under spatially varying moisture conditions. It also explored the additional information gained from using multiple view angles (MVA) of this ratio. Results indicate that for the Howland Forest, Maine study site in August 2003, the MVA version of the NIR/blue ratio discriminated among moisture-driven LULC variations, wetland types and moisture stress extent. Separability was best captured by forward view angles and the optimum multi-angular composite was the An/Af/Bf RGB. The multi-angular ratio was sensitive to variations in species type and vigour, water/vegetation proportions and moisture gradients across emergent wetlands. The study extends the original NIR/blue ratio to include wetlands, aquatic vegetation and other temperate LULC types and has the potential to avoid false change detection while contributing to improved mapping and land cover classifications.