Empirical evapotranspiration modeling in a chaparral ecosystem using statistical methods

Five years' worth of optical, evapotranspiration, and micro-meteorological data were collected at Sky Oaks, a chaparral-dominated site in San Diego County. The site underwent years of normal precipitation, a 100-year drought, a wildfire, and post-fire recovery, providing a wide range of conditions to study. Initially, correlation analyses were run to identify promising optical, micro-meteorological variables, and flux aggregations for use in models. The FAO-56 Penman-Monteith model was then compared against simple linear regression models based on optical indices and then on optical indices with micro-meteorological variables, both for split time periods and the entire study time-frame. A dummy variable to test for fire and equipment artifacts and an fAPAR-based vegetation cover index were also incorporated into the crop coefficient of the Penman-Monteith based model. The model that captured evapotranspiration dynamics best incorporated the vegetation cover index. Model quality was also affected by flux aggregations and whether the flux and micro-meteorological measurements were mid-day or daily averages. Between the pre- and post- fire conditions, the most statistically significant optical indices in modeling evapotranspiration also changed: WBI, mWBI, and PRI were most significant before the fire, and PRI and NDVI were most significant after the fire. The overall results indicate that with disturbance and recovery, the controls in evapotranspiration also change based on vegetation status and the models developed in this study can be used as a base-line for further refinement.