Soil texture is one of the most important variables in crop production that can influence agriculture plans. Preparing a map showing the different zones of soil texture classes is very important in agricultural planning. The map also serves as a basic map in many soil studies and land suitability. In the present research, the efficiency of different methods of interpolation of soil particle size distribution such as inverse distance weighting (IDW) and geostatistical methods such as kriging, co-kriging, and artificial neural networks (ANNs) was evaluated to predict the soil texture. We collected 105 soil samples from depths of 0–15 cm from Gavoshan dam basin in Kurdistan Province, Iran, and measured their texture by hydrometer method. Next, the map of soil particles groups was prepared using different methods of interpolation in ArcGIS environment. Afterward, the longitude and latitude, height, slope percent, and soil texture particles of training points were introduced to an ANN to estimate soil texture particles by MATLAB software. The accuracy of each method was assessed by statistical indicators such as root mean square error (RMSE), geometric mean error ratios (GMER), and the correlation coefficient (R) with the scoring method. Results showed that the accuracy of the co-kriging was greater than that of kriging in estimating the clay percentage. Although the efficiency and accuracy of ANNs in the estimation of particle size spatial distribution in the sand, silt, and clay was greater than three inverse distance weighting, kriging, and co-kriging methods, the correlation of the model was less than 50% using the ANNs. In this regard, the IDW had a lower accuracy among other methods.
