Wheat is a strategic crop in Iran. In the current research, we applied FAO land evaluation frameworks for land suitability assessment of the agricultural land of Kurdistan province for irrigated wheat crop. Land quality index (LQI) was calculated for 105 point observations by comparing of land requirements and land characteristics. Calculated LQI was then digitally mapped. Here, we used Boruta algorithm to identify the most relevant auxiliary variables (i.e. remote sensed data and terrain attributes) for spatial modelling of LQI. Then, three machine learning techniques (random forest, RF, support vector machine, SVM, and artificial neural networks, ANN) were implemented to correlate LQI to the most relevant auxiliary variables. The results of Boruta algorithm indicated that the normalized difference vegetation index and the wetness index were the most important auxiliary variables in order to capture the spatial variation of LQI in the study area. Furthermore, the root mean square error (RMSE) of the spatial models of all machine learning algorithms revealed the superiority of ANN relative to RF and SVM for predicting LQI. The ANN improved prediction accuracies of LQI about ~10% and ~14%, respectively, compared to those of RF and SVM models. The digital map of LQI in the study area was divided into four zones: highly suitable (S1), moderately suitable (S2), marginally suitable (S3) and unsuitable (N) for agricultural purpose. Importantly, it was found that about ~7%,~41% and ~52% of the study area falls under S2, S3 and N zone, respectively, for the irrigated wheat crop purpose. Keywords: Land evaluation, land quality index, machine learning, digital mapping.
