Assimilation of leaf Area Index from multisource earth observation data into the WOFOST model for sugarcane yield estimation

ABSTRACT

Crop Growth Models (CGM) have been widely used in estimating crop yield at a local scale while Remote Sensing (RS) data has the advantage of retrieving crop parameters such as leaf Area Index (LAI) at a range of spatial scales. Data Assimilation (DA) is highly useful tool that integrates CGM and RS data derived from satellite imageries to improve the simulated crop state variables and consequently model outputs such as crop total biomass and yield. In this study, we assimilated LAI with the WOrld FOod STudies (WOFOST) model to estimate sugarcane yield using an Ensemble Kalman Filter (EnKF) algorithm. The LAI was retrieved from Landsat 8 (L8) optical and Sentinel 1A (S1A) Synthetic Aperture Radar (SAR) imageries using a Gaussian Process regression (GPR) method. The Deterministic Modeling (DM), independent assimilations of LAI retrieved from L8 and S1A, and assimilation of LAI retrieved from a combined SAR-optical data were tested and validated using field observation data in the Wonji-Shoa sugar plantation, Ethiopia. The results demonstrate that the accuracy of sugarcane yield estimated by the WOFOST model was significantly improved after DA using combined L8 and S1A data. Compared to the DM estimation, the root mean square error (RMSE) was decreased by 2.13 t/ha for the independent assimilations of LAI retrieved from L8, 3.96 t/ha for the independent assimilations of LAI retrieved from S1A and 5.94 t/ha for combined assimilation of L8 and S1A LAI. A coefficients of determination (R²) of 0.36, 0.48, 0.53, and 0.69 and Normalized Root mean square error (NRMSE) of 14.72%, 11.67%, 10.55%, and 8.44% were obtained for DM, L8 LAI assimilation alone, S1A LAI assimilation alone and combined L8 and S1A LAI assimilation, respectively. The results show that combined L8 and S1A LAI DA has better performance because SAR and optical data have complementary effects. Hence, the assimilation of LAI from combined L8 and S1A data into the WOFOST model provides a robust technique to improve crop yield estimations.

KEYWORDS:

• Data assimilation
• EnKF
• LAI
• Landsat 8
• Sentinel 1A
• WOFOST
• sugarcane yield

Acknowledgements

The authors are thankful to the Ethiopian Space Science and Technology Institute (ESSTI) for financially supporting this research. We thank the Wonji research centre staffs for their cooperation in the field surveys and data collection. We also thank the Alaska Satellite Facility (ASF) and the United States Geological Survey (USGS) for the open source Sentinel-1A and Landsat 8 OLI data respectively. Our gratitude is extended to the Python development team for the open source python crop simulation environment (PCSE) package for building crop simulation models.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author [[email protected]] upon reasonable request.

Authors’contribution

All authors contributed to the study conception and design. GA analyses and interprets the data and wrote the manuscript. TT and BG revised and edited the manuscript. All authors read and approved the final manuscript.

Additional information

Funding

The work was supported by the Ethiopian Space Science and Technology Institute [-].