Remote Sensing Evapotranspiration Model: Study And Application In The Arid Oasis

Evapotranspiration is an important component in the mass and energy exchange between land surface and atmosphere. It is also the final type of water consumption. So it has attracted many researchers from hydrology, water resources management, climatology and agriculture fields. The traditional methods to compute evapotranspiration are based on field measurements. These results are difficult to extend to a larger spatial scale because of the heterogeneity in land surface. Remote sensing data are continuous in space and dynamic in time. So it is potential to use them to estimate regional evapotranspiration. This thesis did further researches on some aspects in evapotranspiration retrieval, and applied these results to analyze water consumption within the arid oasis.Soil heat flux is an important component during evapotranspiration retrieval. It is estimated empirically in existing remote sensing models. This thesis introduced a more theoretical method—the harmonic method—to estimate soil heat flux. This method was modified with fractional vegetation cover. Then researches were focused on how to obtain input variables of the harmonic method. On one hand, a typical curve to describe the diurnal pattern of surface temperature was introduced to determine the harmonic term. On the other hand, the harmonic method was coupled with the Two-Source Energy Balance (TSEB) model through surface soil water content. Comparisons with field measurements indicated that errors in latent heat flux simulations are limited through model coupling.In order to obtain remote sensing data with high spatiotemporal resolution, this thesis put forward downscaling surface temperature data with the triangle method. This method was compared with the polynomial method, the fractional vegetation cover method, and the local similarity method using Landsat 7 ETM+ data. Simulations with the triangle method are more reasonable. Further analysis showed that these downscaling methods based on vegetation index are only applicable to coarse pixels with large heterogeneity in vegetation density. Evapotranspiration retrieved by the remote sensing model is usually instantaneous. In order to extend this value to a longer temporal scale, this thesis carried out some researches on land surface data assimilation, including analyzing sensitivity of the data assimilation system and assimilating surface soil water content, surface temperature, latent and sensible heat fluxes retrieved from remote sensing data. It was found that the systemic bias in measurements can deeply affect assimilation results. When the precision of measurements is controlled within a certain range, data assimilation can improve simulations of the measured variable. For the other variables, assimilation results may be worse than the full open loop case.Based on researches above, this thesis designed a process to estimate evapotranspiration at different temporal scales using multi-temporal MODIS data. It was applied to the Yerqiang Oasis. The spatiotemporal distribution of water consumption in this oasis in 2007 was analyzed. Water consumption proportions by different land use types were also calculated. These results can provide scientific basis for water consumption research and water resources allocation in the oasis.