Identification of soil type in Pakistan using remote sensing and machine learning

Abstract

Soil study plays a significant role in the cultivation of crops. To increase the productivity of any crop, one must know the soil type and properties of that soil. The conventional soil type identification, grid sampling and hydrometer method require expert intervention, more time and extensive laboratory experimentation. Digital soil mapping, while applying remote sensing, offers soil type information and has rapidity, low cost, and spatial resolution advantages. This study proposes a model to identify the soil type using remote sensing data. Spectral data of the Upper Indus Plain of Pakistan Pothwar region and Doabs were acquired using fifteen Landsat eight images dated between June 2020 to August 2020. Bare soil images were obtained to identify the soil type classes Silt Loam, Loam, Sandy Loam, Silty Clay Loam and Clay Loam. Spectral data of band values, reflectance band values, corrective reflectance band values and vegetation indices are practiced studying the reflectance factor of soil type. Regarding multi-class classification, Random Forest and Support Vector Machine are two popular techniques used in the research community. In the present work, we used these two techniques aided with Logistic Model Tree with 10-fold cross-validation. The classification with the best performance is achieved using the spectral data, with an overall accuracy of 86.61% and 84.41% for the Random Forest and Logistic Model Tree classification, respectively. These results may be applied for crop cultivation in specific areas and assist decision-makers in better agricultural planning.

Identification of Soil Types and Salinity Using MODIS Terra Data and Machine Learning Techniques in Multiple Regions of Pakistan

“…Moreover, the RF model exhibited higher precision, recall, and f1-score compared to GB, AB, DT, KNN, and ET, as depicted in Figure 5. Similarly, the effectiveness of RF in soil type classification was evident in the preliminary study [41]. Additionally, the RF model proved to be the most suitable for soil salinity mapping, achieving R 2 of 0.90, MAE of 0.56, MSE of 0.98, and RMSE of 0.97, as shown in Figures 6 and 7.…”

“…The RFR model demonstrated superior performance compared to the GB, AB, DT, KNN, and ET, primarily because of its utilization of random sampling [77], its effective fitting on smaller datasets [78], and the resulting improvement in decision-making accuracy [55]. Indeed, the preliminary studies provided clear evidence of the effectiveness of RF in predicting soil salinity [38,79].…”

“…Huete [47] Salinity Indices Abbas and Khan [53] After calculating the indices, the combined dataset was made from reflectance of bands (1-7), vegetation indices, spectral indices, and laboratory results. Several preprocessing techniques [41] were implemented before the training of ML models. To classify soil types, the ML models utilized input features consisting of reflectance values from seven bands and four vegetation indices, with the target class labeled as "soil type".…”

“…To avoid overfitting and to control the trade-off between margins and training mistakes, tuning parameters C, also known as the penalty factor, were used [70]. Using SVM, any problem can be solved by choosing a good kernel function [41]. In this research, The Radial Basis Function (RBF) kernel function was picked for both classification and regression problems.…”

“…They mapped the soil salinity with R 2 = 0.54 and also generated the salinity maps. Recently, Haq et al [41] proposed a model for determining the soil types using remote sensing data. Landsat 8 images were taken in order to classify soil types using three machine learning models RS, SVM, and Logistic Model Tree (LMT).…”