Investigating the Impact of Temporal Compositing on Remotely Sensed Based Evapotranspiration

Abstract:

Evapotranspiration (ET) is an important component of the hydrological cycle and is one of the most difficult to quantify. ET can be measured at the surface directly using the eddy covariance method which provides continuous measurements but is limited in representing ET over areas larger than a few square kilometers. To obtain ET data over a large region satellite data can be utilized. While ET can be calculated from daily satellite imagery daily images can have missing data due to conditions such as cloud cover. To obtain a complete image of ET multiple day composites are constructed. The composited images offer a more complete picture but may induce bias in the surface radiometric temperature, thus biasing flux estimates. Here, methods for obtaining composite satellite images of ET in the grasslands of the United States are compared. The approach for obtaining ET will be based on the Priestly Taylor (PT) equation. The advantage of using the PT equation is that the only input is net radiation (R_n) and the soil heat flux (G). In addition to the inputs of R_n and G the PT equation also incorporates a dimensionless parameter α that accounts for aerodynamic resistance. The surface temperature – vegetation index (T_s-VI) triangle method is also used to estimate α using daily surface temperature and 16 day NDVI images from the Moderate Resolution Imaging Spectrotadiometer (MODIS) aboard the Aqua satellite. Fluxes are computed daily and compared with eddy covariance observations in the region. In order to investigate the bias induced by temporal compositing, α is calculated from surface temperature data composited at the 2, 4, 6, 8, 16 and 32 day scales. Flux estimates are computed at each scale and compared to daily flux estimates aggregated to the same time scale. Differences in the seasonal and spatial trends of the methods are presented focusing on the impacts of climate anomalies such as drought. Implications for using composite images are discussed.