An improved fusion of Landsat-7/8, Sentinel-2, and Sentinel-1 data for monitoring alfalfa: Implications for crop remote sensing

Precision agriculture management with remote sensing big data provides a promising solution to monitor crops. Alfalfa is an important forage crop for various livestock around the world. Unlike corn and soybean, alfalfa growth is difficult to describe using a typical phenological curve since it is characterized by monthly harvest and rapid regrowth. Limited by the availability of high spatio-temporal resolution data from single satellite sensor, there have been few studies reported on remotely monitoring alfalfa. Fusion of Landsat/Sentinel-2 optical and Sentinel-1 Synthetic Aperture Radar (SAR) data can improve the spatio-temporal resolution of satellite data. However, to form a unified optical and SAR image from multisource satellites, there are still four issues during data fusion. To overcome these challenges, this study proposes an improved fusion of Landsat-7/8, Sentinel-2, and Sentinel-1 data for alfalfa. The improved fusion framework includes three models, namely Landsat2Sentinel2, SAR2Optical, and Optical2SAR. The results indicate that using random forest (RF) algorithm and the known spectral bands, the Landsat2Sentinel-2 model improves Landsat-based surface red edge reflectance fusion with root mean square error (RMSE) reduced 28.22-31.16 % compared to multiple linear regression models. In addition, the RMSE was further reduced by 22.61-23.58 % if considering solar angles in the RF algorithm. Compared to the traditional linear regression model, the Landsat2Sentinel-2 model more accurately fused four surface optical vegetation indices (VIs) derived from Landsat and Sentinel-2 with a reduced RMSE of 46.81-51.16 %. The SAR2Optical model improved VIs fusion with a reduced RMSE of 32.18-34.59 % benefited from the involved climate data. The Optical2SAR fusion model accurately generated optical-based SAR data, proving that optical and SAR data can be reciprocally fused. Moreover, the developed models present improved accuracy in fusing SAR and optical data at both surface and top-of-atmosphere (TOA) levels. On the other hand, the proposed framework can be used to build a unified data fusion model across different crop types. Overall, the proposed optical and SAR data fusion framework demonstrated has great potential for monitoring alfalfa and other crops using multisource satellites.