Evaluation of Ecosystem Water Use Efficiency Based on Coupled and Uncoupled Remote Sensing Products for Maize and Soybean

Abstract

Accurate quantification of ecosystem water use efficiency (eWUE) over agroecosystems is crucial for managing water resources and assuring food security. Currently, the uncoupled Moderate Resolution Imaging Spectroradiometer (MODIS) product is the most widely applied dataset for simulating local, regional, and global eWUE across different plant functional types. However, it has been rarely investigated as to whether the coupled product can outperform the uncoupled product in eWUE estimations for specific C4 and C3 crop species. Here, the eWUE as well as gross primary production (GPP) and evapotranspiration (ET) from the uncoupled MODIS product and the coupled Penman–Monteith–Leuning version 2 (PMLv2) product were evaluated against the in-situ observations on eight-day and annual scales (containing 1902 eight-day and 61 annual samples) for C4 maize and C3 soybean at the five cropland sites from the FLUXNET2015 and AmeriFlux datasets. Our results show the following: (1) For GPP estimates, the PMLv2 product showed paramount improvements for C4 maize and slight improvements for C3 soybean, relative to the MODIS product. (2) For ET estimates, both products performed similarly for both crop species. (3) For eWUE estimates, the coupled PMLv2 product achieved higher-accuracy eWUE estimates than the uncoupled MODIS product at both eight-day and annual scales. Taking the result at an eight-day scale for example, compared to the MODIS product, the PMLv2 product could reduce the root mean square error (RMSE) from 2.14 g C Kg−1 H2O to 1.36 g C Kg−1 H2O and increase the coefficient of determination (R2) from 0.06 to 0.52 for C4 maize, as well as reduce the RMSE from 1.33 g C Kg−1 H2O to 0.89 g C Kg−1 H2O and increase the R2 from 0.05 to 0.49 for C3 soybean. (4) Despite the outperformance of the PMLv2 product in eWUE estimations, both two products failed to differentiate C4 and C3 crop species in their model calibration and validation processes, leading to a certain degree of uncertainties in eWUE estimates. Our study not only provides an important reference for applying remote sensing products to derive reliable eWUE estimates over cropland but also indicates the future modification of the current remote sensing models for C4 and C3 crop species.

Integrating Meteorological and Remote Sensing Data to Simulate Cropland Nocturnal Evapotranspiration Using Machine Learning

“…This could lead to errors in the measurement of climate conditions, vegetation indexes, etc., at the site when using FLUXNET data and MODIS data, which do not match each other better. It has been noted that finer species mapping is needed to explore data matching quantitatively [85], and this needs to be followed up with more in-depth studies.…”