Abstract Phosphorous (P) is one of the most important nutritional and mineral elements for the growth of crops and plays critical roles in several physiological and biochemical processes of plants such as photosynthesis, protein synthesis, respiration, energy transfer, cell division and expansion, root growth, flowering and seed formation. Therefore, an adequate information about the buffering capacity of soil for P is crucial to increase crop growth and productivity and maintain food security for the growing world population. Buffering capacity is an important factor influencing the availability of phosphorus for plants, because it controls the release of phosphorus from its source to the soil solution. In this research, efficiency of intelligent models for predicting soil phosphorus absorption indices and determining the most important environmental and soil characteristics affecting phosphorus absorption indices were examined. Also, the effectiveness of visible to near-infrared spectrometry approach was eva‎luated in estimating these indices in 100 selected locations in Sanandaj region of Kurdistan province. In order to achieve the goals of this research, different data set were used included a set of indicators extracted from satellite images, digital elevation model, thematic maps and soil reflection spectra. In this study, some soil phosphorus buffering capacity were predicted using different machine learning models (Cubist (Cu), Random Forest (RF), Support Vector Machines (SVM) and Gaussian Process Regression (GPR)). The results of spectroscopic studies showed that among the seven pre-processing methods were applied on the raw spectral data, the best pre-processing selected for MBC and SPR index were BC and MSC, respectively, and SNV pre-processing was selected for PBC, SBC, Qmax. Also, the prediction determination coefficient (R2) of each of these pre-processing were 0.43, 0.33, 0.67, 0.49, and 0.48, respectively. The results of digital soil mapping showed that the use of topographic attributes as the only auxiliary variable is not enough to predict phosphorus buffering capacity indices. However, remote sensing data and its combination with soil properties were reliably used to predict phosphorus buffering capacity indices (R2 = 0.41 for MBC by the RF model, R2 = 0.49 for PBC by the Cu model, R2 = 0.37 for SPR with Cu model and R2 = 0.38 for SBC model by RF model). The lowest RMSE values were obtained for MBC by RF model, PBC by SVM model, SPR by Cubist model and SBC by RF model. The results also showed that remote sensing data as an easily available data set can reliably predict phosphorus buffering capacity indices in the study area. The results of variable significance analysis showed that among soil characteristics, cation exchange capacity (CEC) and clay content and among remote sensing indices B5/B7, middle index, coloration index, saturation index and OSAVI were the most essential factors for predicting phosphorus buffering capacity indices. Further studies are recommended to use other proximal sensing data to improve the prediction of phosphorus buffering capacity indices for accurate decision making across the landscape

شمس اله ايوبي

مهران شيرواني جوزداني

حسين شريعتمداري , احمد ميره اي