Satellite-based winter wheat yield estimation with a newly parameterized LUE model based on crop water status and leaf chlorophyll content

Abstract

Water and nutrient availability are crucial factors influencing crop yield. However, the extent of their respective impacts on yield and the potential of remote sensing to clarify these effects remain insufficiently understood. This study explores the relative importance of satellite-derived crop water status (CWS) and leaf chlorophyll concentration (LCC) in determining crop yield production at the field scale. To address this question, we introduce a newly parametrized LUE model for winter wheat yield estimation. It leverages ETa and subsequently CWS from OPTRAM-ET, plus LCC from PROSAIL, to drive yield estimates. The LUE model was calibrated using big-plot field experimental data collected in 2021 and 2022 and was further validated on large areas across 125 farm fields from 2017 to 2021 in South Germany and Switzerland. Results showed that, under various nitrogen fertilization treatments in a region such as Germany with relatively favourable water availability, LCC showed a more dominant role in yield determination and was more sensitive to nitrogen availability than was CWS. Although the interplay between CWS and LCC was important, even slight improvements in the accuracy of LCC measurements considerably enhanced the precision of winter wheat yield estimates. Yield estimation using the LUE model had a high accuracy, with R2 of 0.89 and RMSE of 0.74 t/ha in the big-plot experiments. Subsequently, the model was validated in large fields in Germany and Switzerland. While the direct impact of CWS on yield was less pronounced, its derivation from optical data provided superior temporal resolution compared with thermal images, which further refined yield predictions by increasing R2 from 0.21 to 0.56 on the TUM fields and from 0.33 to 0.56 on the SWTZ fields, while decreasing RMSE from 1.22 to 0.91 t ha⁻¹ and from 1.50 to 1.22 t ha⁻¹ , respectively. These findings highlight the importance of taking into account both the CWS and LCC, as well as their derivation methods, in predicting crop yield, presenting a scientifically robust approach to spatially explicit yield estimation under varying nitrogen availability conditions.

Dr. Haibo Yang

Lecturer (current)

My research interests include hyperspectral remote sensing, precision agriculture, and nitrogen nutrient management.

Prof. Dr. Kang Yu

Professor of Precision Agriculture

My research interests include precision crop farming, hyperspectral remote sensing, and AI in agriculture.