PROJECT TITLE :

Investigation of Remote Sensing Imageries for Identifying Soil Texture Classes Using Classification Methods

ABSTRACT:

In this paper, classification trees were used to assess the utility of remote sensing imageries for detecting soil texture classes using one-against-one (OAO), one-against-all (OAA), and all-together approaches. Cloud-free Landsat photos across a small mountainous watershed yielded a set of normalized difference vegetation indices (NDVIs).

A digital elevation map was used to calculate terrain indicators (elevation, slope, and topographic moisture index) (30 m). Models were created with various input parameters (solely NDVI, purely topography and stratum, and NDVI with topography and stratum).

The classification accuracy was assessed using the overall accuracy, kappa statistic, receiver operating characteristics (ROC), and the area under the ROC curve (AUC). The classification strategy had a significant impact on the outcomes, according to the findings.

With averaged overall accuracy, kappa statistic, and AUC of 0.949, 0.821, and 0.87, the models under the OAO approach performed better. Overall accuracy, kappa statistic, and AUC of 0.975, 0.918, and 0.907, respectively, were the best for the model with NDVI plus topography and stratum.

The model with only NDVI and the model with purely topography and stratum both had similar results. With the use of NDVI, more clay and sand samples were detected. For clay, loam, and sand, the contributions of NDVI to explain soil texture class variability were 144 percent, 0 percent, and 14 percent, respectively.

The ideal period for distinguishing soil texture classes in the watershed was when NDVI was measured during the stem and leaf growth cycle of sweet potatoes. Our findings will be useful in evaluating the quality of the ecological environment utilizing remote sensing data.