Sentinel‐2 Accurately Estimated Wheat Yield in a Semi-arid Region Compared With Landsat‐8

Abstract

Wheat and barley are among the primary food resources of the world population; therefore, their growth and observation are essential in farms to enhance food security worldwide. On top of that, careful observation of the product is essential to find solutions for the issues faced during their production and to reduce the impacts of weather changes. With the advancement of Remote Sensing technology, the observation and estimation process has increased. In this study, numbers of spectral vegetation indices was used along with canopy biophysical properties ( LAI ) and biochemical properties (chlorophyll), there calculated from (Landsat 8 and Sentinel-2) satellite data. The wheat and barley samples were collected before were be ready for harvest, and a relation with the vegetarian indices was established using the Multi-Linear Regression module, in which the equations used in predicting the harvest were developed and used to create a graph for expected harvest. The result indicated that there is a strong relationship between the vegetation indices of Sentinel-2 and Landsat images and the actual grain yield with R2 of 0.77 and 0.71, respectively. The results show that the strongest correlation is observed between the LAI data obtained from Sentinel data and cereal yield data, with an R2 0.68, and the highest correlation for the indices of Landsat images is observed in the NDWI with R2 0.59 and the lowest degree of error was in the root mean square error (RMSE) for the Sentinel-2 and Landsat 8 with 0.57 and 1.54. In addition, this study also showed that the least relationship for grain yield prediction was observed between the NDRI for Sentinel-2 (R2 0.1) and SAVI for Landsat image (R2 0.47).

Plant phenotyping relevance

衛星リモートセンシングと回帰モデルにより穀物収量を推定し、Sentinel-2とLandsat-8の精度を比較・評価しており、植物形質の取得・推定手法が中心である。

Code and data availability

The supplied blocks contain no paper-specific public assets: no phenotype/trait dataset deposit, no field-survey data availability statement, no author code/workflow URL, and no supplement with images or models. All URLs in the text are citations to prior work, and processing (GEE, R) is mentioned without any deposit.

No evidence-backed public reproduction asset is currently recorded.