Assessing the potential of parcel-level barley yield estimation in Spain: a case study using the process-based model SAFYE-C02 and the ACEO processing chain

Details

Supervisors

Defourny, Pierre ; Houdmont, Pierre

Faculty

Faculté des bioingénieurs

Degree label

Master [120] en sciences agronomiques et industries du vivant, à finalité spécialisée: Geo-Information Science and Earth Observation for Environmental Modelling and Management

Abstract

enCrop yield estimates are valuable information for a wide range of stakeholders, from farmers to national governments and traders, to ensure smooth functioning of the food system and to combat food insecurity. Currently, most yield estimates are produced at national or regional scales, with governments primarily relying on cost- and time-intensive survey sampling methods, to inform agricultural policies. However, technological advances in remote sensing and modeling techniques offer opportunities to improve the accuracy and level of insight on yield estimates. Therefore, this study explores the potential of generating parcel-level crop yield estimates for barley production in Castilla y León, Spain, by applying the latest version of the process-based SAFY model series, in combination with the assimilation of the remote-sensing-derived biophysical variable Leaf Area Index (LAI). Additionally, this study presents one of the first applications of the AgriCarbon-EO (ACEO) processing chain, a new hybrid tool designed to efficiently produce yield estimation at the pixel level across large regions. The SAFYE-CO2 model was applied at the parcel level, achieving an RMSE of 1192 – 1318 kg/ha and an R2 of 0.16 – 0.21. These results are comparable to the performance of the statistical Sen4Stat model, which yielded an RMSE of 962 – 1182 kg/ha and R2 of 0.29 – 0.37. However, the application of the ACEO processing chain revealed considerable limitations in its ability to distinguish between parcels with low and high LAI, resulting in a lower predictive performance (RMSE: 1212 kg/ha, R2: 0.01). The study demonstrates that while SAFYE-CO2 is capable of producing parcel-level yield estimates, improvements are needed to enhance its robustness and scalability, particularly when used in conjunction with ACEO. Key areas for improvement include the need to address equifinality issues in the calibration of SAFYE-CO2 and enhancing the parameterization and integration of remotely sensed data for ACEO. These insights contribute to ongoing efforts to develop operational, scalable tools that support parcel-level crop monitoring.