Performance of spectral indices for soil properties: a case study from Redland farm, south Florida

Proximate interpretation of soil properties is essential for sustainable agriculture, demonstrating this for a possum trot farm located in South Florida, Miami Dade County (MDC), known for diverse agronomic regions over the decade. In this work, we explore the capabilities of multispectral images (Sentinel 2A and Landsat 8) for accessing the dynamic soil properties of the study site. The predefined combinations of spectral band values (spectral indices) of Sentinel 2A and Landsat 8 image on the study area were used for evaluation. The correlation coefficient and linear regression models were demonstrated to assess the relationship between the derived spectral indices and five topsoil properties (Bulk Density (BD), Soil Organic Matter (SOM), Electric Conductivity (EC), pH, and Water Content). The results illustrated that specific soil properties (SOM, EC, pH, and BD) correlated well with different spectral indices with both images. Eight spectral bands combinations were found good with three soil properties with maximum correlation coefficient (R = 0.623) for Sentinel 2A, and Landsat 8 has maximum correlation coefficient (R = 0.463) of three spectral indices for two soil properties. The influence of distinct spectral bands of multispectral satellite images in soil surface properties involved in the best-suited indices algorithms was discussed in this article. Overall, we found that the spectral indices demonstrated promising results for this study, and hence they can be accounted for in soil investigation in agriculture.