Using Field-Scale Evapotranspiration to Study Impacts of Drought and Disturbance on a Forested Landscape in North Carolina, United States

Forests provide many important environmental services, including animal habitat, timber production, watershed water regulation and many others. The decline of forests can happen under many causality agents, which can be categorized into two groups, disturbance and drought. Disturbance includes harvest, thinning, wind throw, fire and insect damage. Drought can serve as a modulator of forest disturbances. For example, drought can make forest more susceptible to insect attack and wildfire, while prolonged drought can directly cause tree mortality. Hence, improved understanding of the impacts of drought and disturbance on forests at high spatiotemporal resolution is extremely important for forest management. As an important parameter used to monitor forest health and regional water use, evapotranspiration (ET) can be estimated using surface energy balance models based on thermal infrared (TIR) imagery. One challenge in using TIR remote sensing for monitoring applications is the need for both high spatial and temporal resolution imagery. While Landsat TIR data can provide high spatial resolution (~100 m), the long revisiting time (>16 days) limits the frequency of ET estimation. Moderate Resolution Imaging Spectroradiometer (MODIS) TIR data can provide daily information but only at relatively coarse scale (1-km resolution). To get daily Landsat-scale ET data, Landsat and MODIS retrieved ET was estimated using ALEXI/DisALEXI and then fused using the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM). The study area was over a forested landscape (~900km²) on the humid lower coastal plains in North Carolina, USA. The simulation period was from 2006 to 2012, with 2007 and 2008 as severe drought years. The model results were evaluated at two AmeriFlux sites (US-NC1 and US-NC2) dominated by a mature and a recently clearcut pine plantation, respectively. The simulated long-term ET shows good agreement with observed fluxes. Evaporation (E) and transpiration (T) were also estimated in the model and were used to investigate changes in water use patterns in response to land cover type, forest stand age, climatic forcing and disturbance. T shows a clear ascending trend with the growth of young pine plantations. Thinning decreases leaf area index and T. Analyses show differential response to extreme drought events from different land cover types, with young plantations and short vegetation showing larger impacts than mature pine plantations with significantly deeper rooting systems. Observed groundwater table depth was also analyzed and the lowest depth since 1988 was found after the drought year, which shows the lag of drought impact and a slow recovery time of groundwater.