Soil salinity has become a highly disastrous phenomenon responsible for crop failure worldwide, especially in countries with low farmer incomes and food insecurity. Soil salinity is often due to water accumulation in fields caused by improper flood irrigation whereby plants take up the water leaving salts behind. It is, however, the subsurface soil salinity that affects plant growth. This soil salinity prevents further water intake. There have been very few studies conducted for subsurface soil salinity estimation. Therefore our study aimed to estimate subsurface soil salinity (at 60 cm depth) for the early stage of wheat crop growth in a simplified manner using freely available satellite data, which is a novel feature and prime objective in this study. The study utilises SENTINEL-1 SAR (synthetic aperture RADAR) data for backscatter coefficient generation, SENTINEL-2A multispectral data for NDSI (normalised differential salinity index) generation and on-ground equipment for direct collection of soil electrical conductivity (EC). The data were collected for two dates in November and December 2019 and one date in January 2020 during the early stage of wheat crop growth. The dates were selected keeping in mind the satellite pass over the study area of Rupnagar on the same day. Ordinary least squares regression was used for modelling which gave R²-statistics of 0.99 and 0.958 in the training and testing phase and root mean square error (RMSE) of 1.92 and mean absolute error (MAE) of 0.78 in modelling for soil salinity estimation.

中文翻译：

---

使用 SENTINEL-1 SAR 和 SENTINEL-2 多光谱卫星数据的协同作用对印度旁遮普省鲁普纳加尔小麦作物生长的早期阶段进行简化的地下土壤盐度估算

土壤盐分已成为导致全球农作物歉收的高度灾难性现象，特别是在农民收入低和粮食不安全的国家。土壤盐分通常是由于不适当的洪水灌溉导致田间积水，植物吸收水分而留下盐分。然而，影响植物生长的是地下土壤盐分。这种土壤盐分阻止了进一步的水摄入。对地下土壤盐分估计进行的研究很少。因此，我们的研究旨在使用免费提供的卫星数据以简化的方式估计小麦作物生长早期的地下土壤盐度（60 厘米深度），这是本研究的一个新特征和主要目标。该研究利用 SENTINEL-1 SAR（合成孔径雷达）数据生成反向散射系数，利用 SENTINEL-2A 多光谱数据生成 NDSI（标准化差分盐度指数），并利用地面设备直接收集土壤电导率 (EC)。收集了 2019 年 11 月和 12 月两个日期以及 2020 年 1 月小麦作物生长早期的一个日期的数据。选择日期时牢记卫星在同一天通过 Rupnagar 研究区。普通最小二乘回归用于建模，给出了 R 收集了 2019 年 11 月和 12 月两个日期以及 2020 年 1 月小麦作物生长早期的一个日期的数据。选择日期时牢记卫星在同一天通过 Rupnagar 研究区。普通最小二乘回归用于建模，给出了 R 收集了 2019 年 11 月和 12 月两个日期以及 2020 年 1 月小麦作物生长早期的一个日期的数据。选择日期时牢记卫星在同一天通过 Rupnagar 研究区。普通最小二乘回归用于建模，给出了 R² - 训练和测试阶段的统计数据为 0.99 和 0.958，土壤盐度估计建模中的均方根误差 (RMSE) 为 1.92，平均绝对误差 (MAE) 为 0.78。