[M-AG33] Satellite Land Physical Processes Monitoring at Medium/High/Very High Resolution

[MAG33-P05] Crop yield estimation at different growing stages using a synergy of SAR and optical remote sensing data

*Natacha Kalecinski¹、Sergii Skakun¹、Nathan Torbick³、Xiaodong Huang³、Belen Franch⁴、Jean-Claude Roger^1,2、Eric Vermote² (1.University of Maryland, College Park、2.NASA Goddard Space Flight Center, Greenbelt、3.EOMRV, NH, USA、4.Global Change Unit, Image Processing Laboratory, Universitat de Valencia)

キーワード：Agriculture, remote sensing, optical data, SAR, Yield estimation

Crop yield forecasting and assessment is an essential component of crop production assessment and optimization. So far, mainly optical satellite data at high-temporal resolution (1-3 days) has been actively used for crop yield monitoring, whereas the use of synthetic aperture radar (SAR) data has been lagging. In this paper, we assess efficiency of SAR data acquired by various sensors to explain inter- and intra-field crop yield variability. We used optical imagery acquired by Planet/Dove-Classic, Sentinel-2, Landsat 8, to establish a baseline performance of satellite-based indicators to explain yield variability and assess dual- and quad-polarimetric SAR data for crop yield assessment (Sentinel-1, UAVSAR, RADARSAT-2) of corn, soybean and rice in Arkansas, US (258 fields, 2019). In terms of polarimetric indices, the results showed that in general the results for rice were mostly stable and better than the other crops (R²_adj ~ 0.4 on average). The best results were obtained for the Sentinel-1 VH_ascwith R²_adj=0.42 and for RADARSAT-2 phase difference with R²_adj=0.42. The results for corn were the worst with an R²_adj <0.35 for all the indices. The results for soybean were more variable and were highly correlated with certain indicators such as RADARSAT-2 HV, RADARSAT-2 Volume and RADARSAT-2 Pauli HV with R²_adj>0.5. The maximum correlation for optical features occurs in a short time was between DOY 155 (June 4) and 185 (July 5) for corn and rice, and DOY 190 (July 9) and DOY 211 (July 30) for soybean, and these results were consistent across various optical-based sensors. On the contrary, the maximum of correlation for SAR-derived features varied significantly and was between DOY 120 (April 30) to DOY 225 (August 13). A study of the time series parameters cross-correlation showed that the optical parameters were highly correlated, but the SAR parameters showed strong temporal decorrelation. Using a linear model combining SAR parameters with NIR spectral band, we improved the error by 50% in comparison of the error using the common difference vegetation index (DVI) for corn, soybean, and rice with R²_adj>0.7. We also found that both optical and SAR models were unable to explain yields for high values, which corroborates previous findings.