Vahidi Mohammad Javad, Behdani Mohammad Ali, Servati Moslem, Naderi Mehdi
Department of Agronomy, Plant Breeding and Soil Science, University of Birjand, Birjand, Iran.
Shahid Bakeri High Education Center of Miandoab, Urmia University, Urmia, Iran.
Environ Monit Assess. 2023 Mar 20;195(4):488. doi: 10.1007/s10661-023-11109-9.
Using appropriate models in the land use planning process will help increase the accuracy and precision of decisions made by designers. The aim of this study was to investigate and compare fuzzy-based models (fuzzy set theory, fuzzy-AHP, and fuzzy-ANP) to evaluate the suitability of cotton cultivation in Sarayan region (located in eastern Iran). Twenty-eight land units were selected. Weighted arithmetic means of characteristics were performed in representative soil profiles of each unit. Landform-related characteristics were directly entered into the land suitability evaluation modeling. The land index was calculated using three selective qualitative land suitability model guidelines. Qualitative and quantitative land suitability was estimated. The validity of models was determined by r, RMSE, GMER, and MAPE indicators between predicted and actual production. Soil texture, pH, calcium carbonate equivalent, drainage, organic matter, salinity and sodicity, slope, and gypsum are the most important, respectively. Also, the fuzzy-ANP method is more efficient than other models due to its higher r (0.98) and lower RMSE (4.31) and MAPE (0.56) and GMER (0.99) closer to 1. The value of cotton production using fuzzy, fuzzy-AHP, and fuzzy-ANP methods was calculated as 1085 to 4235, 1235 to 4318, and 1391 to 4452 tons per hectare, respectively. The high efficiency of the fuzzy-ANP model is due to the characteristics of the lands used in the evaluation process that are not independent of each other and this model considers them. Examining these models in different weather conditions and combining with the other computational intelligence methods in future experiments are recommended.
在土地利用规划过程中使用适当的模型将有助于提高设计师决策的准确性和精确性。本研究的目的是调查和比较基于模糊的模型（模糊集理论、模糊层次分析法和模糊网络分析法），以评估伊朗东部萨拉扬地区棉花种植的适宜性。选取了28个土地单元。对每个单元代表性土壤剖面的特征进行加权算术平均。与地形相关的特征直接输入土地适宜性评价模型。使用三种选择性定性土地适宜性模型指南计算土地指数。估计了定性和定量土地适宜性。通过预测产量与实际产量之间的r、均方根误差（RMSE）、几何平均误差率（GMER）和平均绝对百分比误差（MAPE）指标来确定模型的有效性。土壤质地、pH值、碳酸钙当量、排水、有机质、盐分和碱度、坡度以及石膏分别是最重要的因素。此外，模糊网络分析法比其他模型更有效，因为其r值较高（0.98），RMSE较低（4.31），MAPE较低（0.56），GMER（0.99）更接近1。使用模糊法、模糊层次分析法和模糊网络分析法计算的棉花产量分别为每公顷1085至4235吨、1235至4318吨和1391至4452吨。模糊网络分析法模型的高效性归因于评价过程中所使用土地的特征并非相互独立，而该模型考虑到了这些因素。建议在未来的实验中在不同天气条件下检验这些模型，并与其他计算智能方法相结合。
