On the relationship between thermal emissivity and the normalized difference vegetation index for natural surfaces

TLDR

The study measured spatial variation of thermal emissivity (8–14 µm) and NDVI across natural surfaces in a Botswana savanna, then applied the derived relationship to AVHRR and Landsat data to compare spatial variability at multiple scales. Measurements with an emissivity box and a red/near‑infrared radiometer revealed a strong logarithmic correlation (R = 0.94) between thermal emissivity and NDVI, suggesting the empirical relationship can aid energy‑balance studies with thermal infrared remote sensing.

Abstract

Abstract The spatial variation of both the thermal emissivity (8–14/mu;m) and Normalized Difference Vegetation Index (NDVI) was measured for a series of natural surfaces within a savanna environment in Botswana. The measurements were performed with an emissivity-box and with a combined red and near-infrared radiometer, with spectral bands corresponding to NOAA/AVHRR. It was found that thermal emissivity was highly correlated with NDVI after logarithmic transformation, with a correlation coefficient of R = 0·94. This empirical relationship is of potential use for energy balance studies using thermal infrared remote sensing. The relationship was used in combination with AVHRR (GAC), AVHRR (LAC), and Landsat (TM) data to demonstrate and compare the spatial variability of various spatial scales.