Using a data fusion method to estimate daily stand-scale evapotranspiration over a managed pine plantation in North Carolina, USA

Abstract:

Within the context of a globally changing climate, efficient management of freshwater resource management is becoming an increasingly critical issue. As an indicator of vegetation health and soil moisture status, remotely sensed estimates of evapotranspiration (ET) estimation can provide valuable information about water usage in managed landscapes. The two source energy balance (TSEB) model has been widely applied to quantify field scale ET over agricultural systems using thermal infrared remote sensing data. However, limitations on the spatial and temporal resolution of the satellite data combined with the effects of cloud contamination constrain the amount of detail that a single satellite can provide. Fusing multi-satellite data with varying spatial and temporal resolutions can give a more continuous estimation of daily ET at field scale. In this study, we used the regional TSEB modeling system (Atmosphere-Land Exchange Inverse; ALEXI) to map ET at 4-km resolution over the continental U.S. using imagery from geostationary satellites. These 4-km regional estimates were disaggregated to 1-km (daily timesteps) using Moderate Resolution Imaging Spectroradiometer (MODIS) input data and down to 30-m with Landsat data 8 scenes during the growing season) using the disaggregation scheme DisALEXI, applied over a managed pine plantation in North Carolina, USA. The MODIS and Landsat ET retrievals were then combined using the Spatial-Temporal Adaptive Reflectance Fusion Model (STARFM) to estimate daily ET estimates at 30-m resolution. The modeled daily ET was in good agreement compared with observations at two Ameriflux eddy covariance flux tower sites (US-NC2, mid-rotation site and US-NC3, recently clear cut site). Seasonal water use patterns varied significantly with stand age, with higher rates of ET from mid-rotation stands in comparison with younger stands. Water use also varies with land cover types, with higher rates of ET from forested areas than from agricultural fields. The results demonstrate the capability of using ALEXI/DisALEXI with multi-satellite data to estimate stand-scale water usage associated various land cover types and stand ages, which can provide useful information for water resource management in forested landscapes.