Dong, TF; Liu, JG; Qian, BD; He, LM; Liu, J; Wang, R; Jing, Q; Champagne, C; McNairn, H; Powers, J; Shi, YC; Chen, JM; Shang, JL (2020). Estimating crop biomass using leaf area index derived from Landsat 8 and Sentinel-2 data. ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING, 168, 236-250.

The availability of Landsat 8 and Sentinel-2 has led to a steady increase in both temporal and spatial resolution of satellite data, offering new opportunities for large-scale crop condition monitoring and crop yield mapping. This study investigated the potential of using Landsat 8 and Sentinel-2 data from the harmonized Landsat 8 and Sentinel-2 (HLS) products for crop biomass estimation for six crops in Manitoba, Canada. Crop biomass was estimated using remotely sensed leaf area index (LAI) to reparametrize a simple crop growth model. The results showed that the LAI of six different crops can be estimated using a generic relationship between LAI and red-edge based vegetation indices (VIs, e.g., modified simple ratio red-edge (MSRRE) and red-edge normalized difference VI (NDVIRE)) for the Multispectral Instrument (MSI) of Sentinel-2. For the Operational Land Imager of Landsat 8 without the red-edge band, LAI can be best estimated using a VI derived from Near-infrared (NIR) and short-wave infrared (SWIR) bands (Normalized Difference Water Index, NDWI1). Above-ground dry biomass of these six crops was more accurately estimated from the assimilation of LAI derived from both satellites (R-2 (the coefficient of determination) = 0.81, RMSE (the root-mean-square-error) = 135.4 g/m(2), nRMSE (the normalized RMSE) 37.9%, RPD (the ratio of percent deviation) = 2.26) than that of LAI derived from MSI-data (R-2 = 0.80, RMSE = 136.7 g/m(2) , nRMSE = 38.3%, RPD = 2.23) or that from LAI derived from OLI-data (R-2 = 0.68, RMSE = 191.0 g/m(2), nRMSE = 53.5%, RPD = 1.16). Further analysis showed that these three assimilation cases (MSI and OLI; MSI alone; OLI alone) with a different number of LAI observations resulted in differences in parameter optimization, particularly the parameters relevant to crop phenology and biomass partitioning. Both crop growth stage (e.g., the emergence date for crop growth) and leaf dry biomass estimated from the assimilation of LAI derived from MSI and OLI, or MSI alone, produced the most accurate estimates. These results are likely attributed to the improved temporal coverage associated with Sentinel-2 and the availability of a red-edge band on this sensor.