A recent paper entitled “Mapping burn severity in a disease-impacted forest landscape using Landsat and MASTER imagery” was accepted by the International Journal of Applied Earth Observation and Geoinformation. This project was led by Dr. Gang Chen, with collaborators from NC State, UC Davis and Lewis & Clark College.
Abstract
Global environmental change has increased forest vulnerability to the occurrence of interacting disturbances, including wildfires and invasive diseases. Mapping post-fire burn severity in a disease-affected forest often faces challenges because burned and infested trees may exhibit a high similarity in spectral reflectance. In this study, we combined (pre- and post-fire) Landsat imagery and (post-fire) high-spectral resolution airborne MASTER data [MODIS (Moderate Resolution Imaging Spectroradiometer) / ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer)] to map burn severity in a California coastal forest environment, where a non-native forest disease sudden oak death (SOD) was causing substantial tree mortality. Results showed that the use of Landsat plus MASTER bundle performed better than using the individual sensors in most of the evaluated forest strata from ground to canopy layers (i.e., substrate, shrubs, intermediate-sized trees, dominant trees and average), with the best model performance achieved at the dominant tree layer. The mid to thermal infrared spectral bands (3.0–12.5 μm) from MASTER were found to augment Landsat’s visible to shortwave infrared bands in burn severity assessment. We also found that infested and uninfested forests similarly experienced moderate to high degrees of burns where CBI (Composite Burn Index) values were higher than 1. However, differences occurred in the regions with low burn severity (CBI values lower than 1), where uninfested stands revealed a much lower burn effect than that in infested stands, possibly due to their higher resilience to small fire disturbances as a result of higher leaf water content.
Comparison of coefficient of determination (R2) using Landsat, MASTER, and the Landsat and MASTER bundle for modeling burn severity in infested and uninfested forests.