Tracking crop phenology in a highly dynamic landscape with knowledge-based Landsat–MODIS data fusion

Biniam Sisheber,Andrew Nelson,Daniel Mengistu,Michael Marshall

doi:10.1016/j.jag.2021.102670

Abstract

Earth observation image data are regularly used to capture surface conditions over large areas, but there is a trade-off between high (or low) spatial and low (or high) temporal resolution. The Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM) overcomes this trade-off by fusing high spatial and temporal resolution multisource image data. However, ESTARFM requires additional modifications in order to provide reliable estimates of surface conditions showing large spectral differences in highly dynamic and fragmented agricultural systems. We modified ESTARFM by taking a knowledge-based approach to track maize and rice phenology in a highly dynamic and fragmented agricultural landscape in Ethiopia in 2019. The two major improvements included: (i) Selection of Landsat-MODIS imageries based on crop sowing and harvesting information and (ii) generation and use of a land cover map to select similar pixels. We assessed model performance with the enhanced vegetation index (EVI) derived from independent Landsat image data and in-situ leaf area index (LAI) data. The improved ESTARFM workflow resulted in reliable Landsat-MODIS prediction (R2 = 0.67, RMSE = 0.07) compared to the standard ESTARFM workflow (R2 = 0.54 RMSE = 0.01) during the rapid growth stage. Our modifications outperformed the standard implementation of ESTARFM according to LAI magnitude (R2 = 0.73–0.84 versus R2 = 0.58–0.64) and phenological timing (RMSE = 8 days verses RMSE = 12 days). Our modified application of ESTARFM serves as a basis for monitoring crop growth and development in highly dynamic and fragmented agricultural systems.

Highlights

Sustainable agricultural production faces enormous challenges due to the intensification of extreme weather events, increased soil infer tility, and rising food demand from a growing population and middle class (Karthikeyan et al, 2020)
When we evaluated performance using ground data, better data fusion and crop phenology detection were obtained at field and pixel scales compared with the standard enhanced Spatial and Temporal Adaptive Reflectance Fusion Method (STARFM) (ESTARFM) workflow
Input selection and similar pixel selection based on crop cal endar information and land cover classification were effective strategies for tracking the crop cycles and determining field-level crop phenology in the study area

Summary

Introduction

Sustainable agricultural production faces enormous challenges due to the intensification of extreme weather events, increased soil infer tility, and rising food demand from a growing population and middle class (Karthikeyan et al, 2020). Unmixing methods such as MERIS–Landsat fusion (Zurita-Milla et al, 2008) and the Unmixing-based Spatiotemporal Reflectance Fusion Model (Huang and Zhang, 2014) apply linear spectral mixing theory to unmix coarse resolution pixels They assume no land cover change during the period of study, so the class proportion is constant for each coarse image, even though this may not be the case (Rao et al, 2015). Weight function methods, such as the Spatial and Temporal Adaptive Reflectance Fusion Method (STARFM: Gao et al (2006)), combine in formation from neighbouring pixels exhibiting similar spectral re sponses in high resolution pixels to account for changes in coarse resolution pixels.

Results

Discussion

Conclusion