In this study we evaluate a Random Forest (RF) model for characterizing the spatial variability of soil moisture based on model derived from in situ soil moisture samples, geophysical data and RADAR observations. The RF model is run with and without C-band SAR backscatter to understand the importance of the inclusion of SAR data for mapping of soil moisture at field scale. The inclusion of SAR data in the RF resulted in a modest improvement however the geophysical parameters (e.g. soil types and terrain properties) were of greater importance.
In order to calibrate and validate the SMAP soil moisture products, networks of ground-based soil moisture sensors have been deployed. Measurements collected from the networks must be upscaled to the radiometer footprint scale (30-40 km) for comparison with the SMAP radiometer-based retrievals. The upscaling is typically performed as a weighted average of individual sensor measurements within the SMAP grid. Since different weighting schemes have been found to result in different upscaled soil moisture estimates, an independent method of assessing soil moisture estimation biases is needed. We therefore present a method for calculating estimation biases at each SMAP Core Validation Site (CVS). The estimation was enabled by networks of enhanced soil moisture sampling that were deployed at four CVSs for a limited time. Based on Random Forests, our method offers a straightforward, systematic, and unified approach to bias estimation across a variety of sites. The method was applied to estimate biases at the four SMAP CVSs.
Wetlands act as major sinks and sources of important atmospheric greenhouse gases and can switch between atmospheric sink and source in response to climatic and anthropogenic forces in ways that are poorly understood. Despite their importance in the carbon cycle, the locations, types, and extents of northern wetlands are not accurately known. We have used two seasons of L-band synthetic aperture radar (SAR) imagery to produce a thematic map of wetlands throughout Alaska. The classification is developed using the Random Forests decision tree algorithm with training and testing data compiled from the National Wetlands Inventory (NWI) and the Alaska Geospatial Data Clearinghouse (AGDC). Mosaics of summer and winter Japanese Earth Resources Satellite 1 (JERS-1) SAR imagery were employed together with other inputs and ancillary datasets, including the SAR backscatter texture map, slope and elevation maps from a digital elevation model (DEM), an open-water map, a map of proximity to water, data collection dates, and geographic latitude. The accuracy of the resulting thematic map was quantified using extensive ground reference data. This approach distinguished as many as nine different wetlands classes, which were aggregated into four vegetated wetland classes. The per-class average error rate for aggregate wetlands classes ranged between 5.0% and 30.5%, and the total aggregate accuracy calculated based on all classified pixels was 89.5%. As the first high-resolution large-scale synoptic wetlands map of Alaska, this product provides an initial basis for improved characterization of land-atmosphere CH4 and CO2 fluxes and climate change impacts associated with thawing soils and changes in extent and drying of wetland ecosystems.
In order to provide a reliable source of ground-based validation data for the SMAP mission at spatial scales of 3 km, 9 km and 36 km, we have developed a new regression-based method capable of yielding highly-accurate upscaled soil moisture estimates based on sparse, irregularly-spaced soil moisture measurements.
A map of vegetated wetlands in Alaska generated by applying an automated classification routine to ALOS PALSAR data from 2007 and ancillary layers. The method used to generate the map and associated accuracy is described in the following, open access, publication: Clewley, Daniel; Whitcomb, Jane; Moghaddam, Mahta; McDonald, Kyle; Chapman, Bruce; Bunting, Peter. 2015. "Evaluation of ALOS PALSAR Data for High-Resolution Mapping of Vegetated Wetlands in Alaska." Remote Sens. 7, no. 6: 7272-7297. The data is a raster in KEA format (http://kealib.org/) with associated map (figure 4 from paper).
