Improving Site-Specific Maize Yield Estimation by Integrating Satellite Multispectral Data into a Crop Model

Vijaya R Joshi,Gregg A Johnson,Kelly R Thorp,Axel Garcia Y Garcia,Paul M Porter,Jeffrey S Strock,Jeffrey A Coulter

doi:10.3390/agronomy9110719

Abstract

Integrating remote sensing data into crop models offers opportunities for improved crop yield estimation. To compare site-specific yield estimation accuracy of a stand-alone crop model with a data-integration approach, a study was conducted in 2016–2017 with nitrogen (N)-fertilized and unfertilized treatments across a heterogeneous 7-ha maize field. For each treatment, yield data were grouped into five classes resulting in 109 spatial zones. In each zone, the Crop Environment Resource Synthesis (CERES)-Maize model was run using the GeoSim plugin within Quantum GIS. In the data integration approach, maize biomass values estimated using satellite imagery at the five (V5) and ten (V10) leaf collar stages were used to optimize the total soil nitrogen concentration (SLNI) and soil fertility factor (SLPF) in CERES-Maize. Without integration, maize yield was simulated with root mean square error (RMSE) of 1264 kg ha−1. Optimization of SLNI improved yield simulations at both V5 and V10. However, better simulations were obtained from optimization at V10 (RMSE 1026 kg ha−1) as compared to V5 (RMSE 1158 kg ha−1). Optimization of SLPF together with SLNI did not further improve the yield simulations. This study shows that integrating remote sensing data into a crop model can improve site-specific maize yield estimations as compared to the stand-alone crop modeling approach.

Highlights

Estimation of crop yield is important to farmers, government agencies, and policy makers in improving crop production efficiency and detecting potential biotic and abiotic risks that affect crop yield [1,2]
Sensors mounted on combine harvesters calculate the mass of grain per unit of area harvested, which together with global positioning systems (GPS) receivers provide grain yield measurements at geo-referenced points to produce yield maps that are effective in visualizing spatial variability of crop yield [18]
Spatial optimization of soil nitrogen concentration (SLNI) only using V10-estimated biomass reduced the root mean square error (RMSE) by 18.8% and gave the lowest RMSE of 1026 kg ha−1

Summary

Introduction

Estimation of crop yield is important to farmers, government agencies, and policy makers in improving crop production efficiency and detecting potential biotic and abiotic risks that affect crop yield [1,2]. Mounting pressures to address environmental problems resulting from crop production [3,4] and increasing competition for greater economic efficiency [5] have directed research efforts for site-specific crop yield estimation. Explicit estimation of crop yield helps to explain the spatial variability of crop growth within a field and to optimize crop management efforts and reduce risks [6]. Agronomy 2019, 9, 719 and geographic information systems (GIS) have greatly enabled digital data-driven approaches for site-specific crop yield estimation [7,8,9,10]. Historical yield maps help to locate high and low yielding regions within a field and are useful in estimating site-specific yield [12,19]. In this regard, are advantageous for monitoring in-season crop growth patterns in response to the effects of weather, pests, disease, and other management issues

Objectives

Methods

Results

Conclusion