Crop yield estimation based on assimilation of crop models and remote sensing data: A systematic evaluation

Abstract

CONTEXTWith the warming trend and the increasing frequency of extreme weather events, accurate crop yield estimation is becoming urgent. Crop yield estimation mainly consists of two methods: crop model simulation and remote sensing observations. Crop models can achieve accurate simulations of crop growth at field scales. However, in regional applications, they are limited by the spatial heterogeneity of certain input parameters. Remote sensing observations can obtain crop status over large areas quickly and conveniently, while lacking knowledge of crop growth processes. By combining the advantages of crop models and remote sensing, crop yield estimation with spatiotemporal continuity can be achieved using data assimilation methods. OBJECTIVEThe research progress of the three elements of data assimilation system has been quantitatively reviewed in this paper. And the relevant literature was quantitatively screened and reviewed to provide a systematic overview of the application of data assimilation in crop yield estimation. METHODSIn this study, the scientific background of the data assimilation system for crop yield estimations was described, and basic principles of data assimilation were introduced. A second part of this review screened and reviewed the relevant literature quantitatively. The answers to problems on: the most widely used crop model, the assimilation algorithm, and the assimilation variables were reported. Finally, a synthesis of the emerging directions and challenges of data assimilation systems for crop yield estimation were discussed. RESULTS AND CONCLUSIONSThe results show that: a) the sequential assimilation method is the most widely used algorithm in the field of data assimilation, especially EnKF. b) WOFOST, DSSAT, AquaCrop and SAFY are the most common models in the research of data assimilation for yield estimation. c) In terms of assimilation variables, LAI (leaf area index), SM (soil moisture), and VIs (vegetation indexes) are relatively common assimilation variables. Research progress of data assimilation system for crop yield estimation is summarized from the aspects of algorithm improvement, model coupling research, multi-source remote sensing data assimilation and multiple assimilation variables. SIGNIFICANCEThis review quantitatively examined and contrasted the research progress of data assimilation systems providing researchers with a more comprehensive background. It also suggests ideas and references for higher resolution, more accurate and reliable yield estimation for future research.