Jason L. McVay | USGS Science

USGS2017

Climate, wildfire, and erosion ensemble foretells more sediment in western USA watersheds

The area burned annually by wildfires is expected to increase worldwide due to climate change. Burned areas increase soil erosion rates within watersheds, which can increase sedimentation in downstream rivers and reservoirs. However, which watersheds will be impacted by future wildfires is largely unknown. Using an ensemble of climate, fire, and erosion models, we show that post-fire sedimentation is projected to increase for nearly nine-tenths of watersheds by > 10% and for more than one-third of watersheds by > 100% by the 2041 to 2050 decade in the western USA. The projected increases are statistically significant for more than eight-tenths of the watersheds. In the western USA, many human communities rely on water from rivers and reservoirs that originates in watersheds where sedimentation is projected to increase. Increased sedimentation could negatively impact water supply and quality for some communities, in addition to affecting stream channel stability and aquatic ecosystems.

Geophysical Research Letters↗

USGS2017

UAV lidar and hyperspectral fusion for forest monitoring in the southwestern USA

Forest vegetation classification and structure measurements are fundamental steps for planning, monitoring, and evaluating large-scale forest changes including restoration treatments. High spatial and spectral resolution remote sensing data are critically needed to classify vegetation and measure their 3-dimensional (3D) canopy structure at the level of individual species. Here we test high-resolution lidar, hyperspectral, and multispectral data collected from unmanned aerial vehicles (UAV) and demonstrate a lidar-hyperspectral image fusion method in treated and control forests with varying tree density and canopy cover as well as in an ecotone environment to represent a gradient of vegetation and topography in northern Arizona, U.S.A. The fusion performs better (88% overall accuracy) than either data type alone, particularly for species with similar spectral signatures, but different canopy sizes. The lidar data provides estimates of individual tree height ( R 2 = 0.90; RMSE = 2.3 m) and crown diameter ( R 2 = 0.72; RMSE = 0.71 m) as well as total tree canopy cover ( R 2 = 0.87; RMSE = 9.5%) and tree density ( R 2 = 0.77; RMSE = 0.69 trees/cell) in 10 m cells across thin only, burn only, thin-and-burn, and control treatments, where tree cover and density ranged between 22 and 50% and 1–3.5 trees/cell, respectively. The lidar data also produces highly accurate digital elevation model (DEM) ( R 2 = 0.92; RMSE = 0.75 m). In comparison, 3D data derived from the multispectral data via structure-from-motion produced lower correlations with field-measured variables, especially in dense and structurally complex forests. The lidar, hyperspectral, and multispectral sensors, and the methods demonstrated here can be widely applied across a gradient of vegetation and topography for monitoring landscapes undergoing large-scale changes such as the forests in the southwestern U.S.A.

Arizona↗