The relationship between effective plant area index and Landsat spectral response across elevation, solar insolation, and spatial scales in a northern Idaho forest

Leaf area index (LAI) is an important forest characteristic related to photosynthesis and carbon sequestration, and gains in efficiency for LAI measurements are possible using remotely sensed imagery. However, the potential effects of complex topography on this measurement system are not well understood. Our objective was to understand how complex terrain and measurement aggregation influence the relationship between LAI and remotely sensed vegetation indices across a mountainous conifer forest. We identified NDVIc, a middle-infrared (MIR) correction to NDVI (Normalized Difference Vegetation Index), as the vegetation index providing the best prediction of effective plant area index (PAIe), used to approximate LAI. We tested formal hypotheses to identify how elevation and solar insolation gradients and spatial scale of measurement aggregation affected the PAIe–NDVIc relationship and found that it changed across elevation at one spatial scale. Comparisons of NDVIc with NDVI revealed that vegetation index choice is important in complex terrain, and we concluded that the MIR correction improves the PAIe–NDVI relationship by explaining variation related to solar insolation. Our results suggest that NDVIc calculated from Landsat ETM+ provides a practical estimate of PAIe across our northern Idaho study area and potentially other conifer forests in complex terrain.