Multi-scale sensitivity of Landsat and MODIS to forest disturbance associated with tropical cyclones

Multispectral data from satellites are widely used to study the effects of extreme weather events in forest ecosystems at a variety of spatial and temporal scales. Understanding the sensitivity of these data is important since these phenomena are projected to increase as climate changes. The Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat data were used to study the observed patterns of forest disturbance at different spatial scales in temperate forest (US Gulf Coast) produced by tropical cyclones Charley (2004), Katrina (2005), Rita (2005), and Gustav (2008), and in tropical rainforests (Australia) produced by cyclone Yasi (2011). The severity of forest disturbance was quantified by applying spectral mixture analysis to the MODIS and Landsat coverages. Field studies were used to verify and compare the results. At the local scale Landsat data was sensitive to forest disturbance both within and between forest types. Higher tree mortality was observed in tropical rainforests than in temperate forests. This observation may be explained by forest type characteristics such as stem density, forest adaptation, and depth of root systems. At the species level, Landsat showed a gradient of forest resilience to tropical cyclone winds that agreed with observational field studies. At the landscape scale, the observed topographic effects on disturbance patterns were well represented by the MODIS data. Positive covariance was observed between surface orientation and slope on the severity of disturbance. Greater levels of disturbance were observed on windward surfaces with steeper slopes. Finally, at the regional scale, MODIS reproduced the pattern of forest damage associated with cyclone winds. The highest level of forest disturbance was observed on the right side of the cyclone track in the northern hemisphere (US Gulf Coast forest ecosystems) and on the left side in the southern hemisphere (Australian rainforest). At the regional scale, forest disturbance was positively associated with the decrease of wind speeds in an inland direction. Tropical cyclone surface winds explained 20% of forest disturbance, although characteristics of cumulative processes could cause this to be underestimated.