Evaluación de índices de vegetación para identificar combustibles forestales

Resumen

The ability of different vegetation indices, derived from spectral bands of a very high spatial resolution remote sensor, was evaluated for classifying forest fuel types. A WorldView-2 image of a forest area in the North of Tenerife (Canary Islands) was used, quasi-simultaneous in time with a field campaign, where seven fuel types were identified in 83 plots defined in the study area. A total number of 110 vegetation indices were calculated for the pixels of the image inside these plots and correlated with the fuel types identified. A separability spectral analysis was performed using Jeffries-Matusita distance and Transformed Divergence. It was found that the bush types (vegetation height of between 0.6 to 2.0 and 2.0-4.0 meters) could not be differentiated from each other with any of the calculated indices. The spectral index with the best performance was the Atmospherically Resistant Soil and Vegetation Index (SARVI2), which is valid to discriminate herbaceous shrub from other fuels.