Development of the LinZI Method for Merging MODIS and Landsat-based Evapotranspiration Maps

Abstract:

Evapotranspiration (ET) plays a key role in transporting water and energy in the soil-plant-atmosphere continuum. Remotely sensed images from Landsat and Moderate Resolution Imaging Spectroradiometer (MODIS) are increasingly used for estimating spatial and temporal distribution of ET using different modeling algorithms. Landsat and MODIS have different spectral, spatial, and temporal resolutions thus resulting in ET maps of different visual and analytical qualities. Downscaling is a great way of utilizing the combined benefits of the high temporal resolution of MODIS images and fine spatial resolution of Landsat images. We have evaluated the performance of the output regression with intercept method for data downscaling in the Colorado River basin. Our results showed that application of regression coefficients computed over large areas such as the Colorado River Basin are not able to produce accurate ET maps. Thus we developed the Linear with Zero Intercept (LinZI) method for downscaling MODIS-based monthly ET maps to the Landsat-scale ET maps for the Colorado River Basin. We obtained a high agreement between downscaled monthly ET maps using the LinZI method and the eddy covariance measurements from seven flux sites within the Colorado River Basin. The mean absolute error of the monthly ET ranged from 8 mm to 25 mm, and the coefficient of determination varied from 0.53 to 0.88. Most of the discrepancies between measured and downscaled monthly ET were mainly at two flux sites due to the prevailing flux footprint. Downscaled monthly ET using the LinZI method nicely captured the temporal variation in the sampled land cover classes. We plan to use the LinZI method for estimating ET at finer temporal resolution (such as 8 days) with further evaluation. The proposed downscaling method will be very useful in advancing the application of remotely sensed images in water resources planning and management.