Relating TRMM Precipitation Radar backscatter to water stage in wetlands

Information on water stage over an extended area is important for hydrological and ecological studies. Microwave remote sensing provides an opportunity to measure changes in water stage from space because of its sensitivity to land surface characteristics; it reduces the need to monitor water stage at multiple locations. In this research, a linear model is developed which relates variation in water stage measurements (ws) to Tropical Rainfall Measuring Mission Precipitation Radar backscatter (σ°). The estimated water stage from the model is compared with the observed water stage in the wetlands of South Florida. The model performance is assessed by comparing the correlation coefficient (R), the root mean square error (RMSE), and the non-exceedance probability of mean absolute error between observed and modeled water stage measurements for various landcovers. The model works reasonably well in the regions with tree heights greater than 5 m. For example, over woodlands R ranges between 0.59–0.93 and the average RMSE = 19.8 cm. Similarly, for wooded grassland, R ranges between 0.54–0.93 and the average RMSE = 19.8 cm. For other relatively shorter height vegetation landcovers such as grassland (R = 0.57–0.85, RMSE = 20.1 cm) and cropland (R = 0.69–0.79, RMSE = 18.2 cm), the model also performs reasonably well. The research presents a novel use of TRMMPR data and gives an insight into the effect of water level in partially inundated vegetation on radar backscatter.