Hassan II Institute of Agronomy and Veterinary Medicine, Rabat, Morocco.
Mohammed Premier University, Oujda, Morocco.
Environ Monit Assess. 2024 Aug 27;196(9):843. doi: 10.1007/s10661-024-13030-1.
Irrigated agricultural lands in arid and semi-arid regions are prone to soil degradation. Remote sensing technology has proven useful for mapping and monitoring the extent of this issue. To accurately discern soil salinity, it is essential to choose appropriate spectral wavelengths. This study evaluated the potential of the land degradation index (LDI) using the visible and near infrared (VNIR) and the short wavelength infrared (SWIR) spectral bands compared to that of soil salinity indices by integrating only the VNIR wavelengths. Landsat-OLI and Sentinel-MSI data, acquired 2 weeks apart, were rigorously preprocessed and used. This research was conducted over irrigated agricultural land in Morocco, which is well known for its semi-arid climate and moderately saline soil. Furthermore, a field soil survey was conducted and 42 samples with variable electrical conductivity (EC) were collected for index calibration and validation of the results. The results showed that the visual analysis of the derived maps based on the examined indices exhibited a clear spatial pattern of gradual soil salinity changes extending from the elevated upstream plateau to the downstream of the plain, which limits agricultural activities in the southwestern sector of the study area. The results of this study show that LDI is effective in identifying soil salinity, as indicated by a coefficient of determination (R) of 0.75 when using Sentinel-MSI and 0.72 with Landsat-OLI. The R value of 0.89 and root mean square error (RMSE) of 0.87 dS/m for soil salinity maps generated from LDI with Sentinel-MSI demonstrate high accuracy. In contrast, the R value of 0.83 and RMSE of 1.24 dS/m for maps produced from Landsat-OLI indicate lower accuracy. These findings indicate that high-resolution Sentinel-MSI data significantly improved the prediction of salinity-affected soils. Furthermore, this study highlights the benefits of using VNIR and SWIR bands for precise soil salinity mapping.
干旱和半干旱地区的灌溉农田容易发生土壤退化。遥感技术已被证明可用于绘制和监测这一问题的范围。为了准确识别土壤盐分，选择合适的光谱波长至关重要。本研究通过整合仅可见近红外（VNIR）和短波红外（SWIR）光谱波段，评估了土地退化指数（LDI）在识别土壤盐分方面的潜力，与仅使用 VNIR 波长的土壤盐分指数进行了比较。使用了相隔两周获取的 Landsat-OLI 和 Sentinel-MSI 数据，并进行了严格的预处理。该研究在摩洛哥的灌溉农田进行，该地区以半干旱气候和中等盐分土壤而闻名。此外，还进行了实地土壤调查，收集了 42 个具有不同电导率（EC）的样本，用于指数校准和验证结果。结果表明，基于所检查的指数得出的地图的直观分析显示了土壤盐分逐渐变化的明显空间模式，从上游高原延伸到平原下游，限制了研究区域西南部的农业活动。研究结果表明，LDI 可有效识别土壤盐分，使用 Sentinel-MSI 时的决定系数（R）为 0.75，使用 Landsat-OLI 时为 0.72。使用 Sentinel-MSI 生成的 LDI 生成的土壤盐分图的 R 值为 0.89，均方根误差（RMSE）为 0.87 dS/m，表明具有高精度。相比之下，使用 Landsat-OLI 生成的地图的 R 值为 0.83，RMSE 为 1.24 dS/m，表明精度较低。这些发现表明，高分辨率的 Sentinel-MSI 数据显著提高了对受盐分影响土壤的预测能力。此外，本研究强调了使用 VNIR 和 SWIR 波段进行精确土壤盐分制图的好处。
