Classification of burn severity using Moderate Resolution Imaging Spectroradiometer (MODIS): A case study in the jarrah-marri forest of southwest Western Australia

Abstract

The southwest of Western Australia is a fire-prone landscape. In this Mediterranean region, prescribed fuel reduction burning is applied as a management tool by the state government's Department of Conservation and Land Management (CALM). Remote sensing data from the Moderate Resolution Imaging Spectroradiometer (MODIS) with multiple observations per day are investigated for operational monitoring of prescribed burning activities.The Normalized Burn Ratio (NBR) is sensitive to the amount of biomass, soil exposure and equivalent water content. The differenced Normalized Burn Ratio (ΔNBR) shows the greatest response of landscape change due to fire. The ratios, originally applied to 30-m Landsat 7 ETM+ data, have been transferred to 250–500 m MODIS data. The high temporal resolution and direct broadcast capability of MODIS are considered favorable for monitoring prefire and postfire conditions, in particular in near-real time. This study applies the ΔNBR to classify burn severity using MODIS data with various levels of preprocessing. On the basis of field studies, four burn severity classes are distinguished with best discrimination for high burn severity where the top layer of the vegetation canopy is altered. As expected, the spatial detail of the classifications from MODIS is reduced when compared to results from Landsat 7 ETM+, but the large-scale spatial patterns are similar. NBR time series of daily data showed that classes of burn severity can be separated for each acquisition date. Large temporal variations of the NBR limit class separation with absolute thresholds, in particular for data uncorrected for effects due to varying viewing geometries. However, MODIS top of atmosphere data allow near-real-time assessment of burn severity, important to fire managers for monitoring postfire conditions.