Evapotranspiration Estimates Derived Using Multi-Platform Remote Sensing in a Semiarid Region

Kyle Knipper,Russell Scott,Terri Hogue,Kristie Franz

doi:10.3390/rs9030184

Abstract

Evapotranspiration (ET) is a key component of the water balance, especially in arid and semiarid regions. The current study takes advantage of spatially-distributed, near real-time information provided by satellite remote sensing to develop a regional scale ET product derived from remotely-sensed observations. ET is calculated by scaling PET estimated from Moderate Resolution Imaging Spectroradiometer (MODIS) products with downscaled soil moisture derived using the Soil Moisture Ocean Salinity (SMOS) satellite and a second order polynomial regression formula. The MODis-Soil Moisture ET (MOD-SMET) estimates are validated using four flux tower sites in southern Arizona USA, a calibrated empirical ET model, and model output from Version 2 of the North American Land Data Assimilation System (NLDAS-2). Validation against daily eddy covariance ET indicates correlations between 0.63 and 0.83 and root mean square errors (RMSE) between 40 and 96 W/m2. MOD-SMET estimates compare well to the calibrated empirical ET model, with a −0.14 difference in correlation between sites, on average. By comparison, NLDAS-2 models underestimate daily ET compared to both flux towers and MOD-SMET estimates. Our analysis shows the MOD-SMET approach to be effective for estimating ET. Because it requires limited ancillary ground-based data and no site-specific calibration, the method is applicable to regions where ground-based measurements are not available.

Highlights

Evapotranspiration (ET) is fundamental to understanding the regional water balance, in semiarid and arid regions where ecosystem processes are often limited by the availability of water and where water loss is dominated by ET [1]
This method is generally suitable when remotely sensed images are available at regular intervals and each image captures the overall pattern of variation in ET [94]
Time between cloud/error free images is six days on average, with the largest gaps in data occurring during the monsoon season where cloudy days increase significantly

Summary

Introduction

Evapotranspiration (ET) is fundamental to understanding the regional water balance, in semiarid and arid regions where ecosystem processes are often limited by the availability of water and where water loss is dominated by ET [1]. Conventional ground-based ET measurement techniques (eddy covariance, Bowen ratio) are constrained to relatively small homogeneous footprints that rarely exceed 1–2 km [5,6] As such, they are limited in their capability to estimate fluxes on larger spatial scales due to the inherent heterogeneity of the land surface and hydroclimatological forcing. Different methods have been proposed to estimate ET through satellite remote sensing which allows acquisition of large-scale distributed data at various spatial and temporal resolutions. These methods vary in complexity, with tradeoffs between empirical and physically-based models [7].

Objectives

Methods

Results

Conclusion