ABSTRACT Predicting soil infiltration at the field scale with high contents in calcareous materials is important for a better understanding of land management. The objectives of this study were to develop a GIS-based digital modeling of soil infiltration in calcareous soils using environmental data in Iran. The soil infiltration data with three replications were measured at 92 points at the regional scale. At each site, the soil readily available properties were determined. Furthermore, remote-sensing and digital elevation model data were applied as auxiliary data. The artificial neural networks (ANN) model was applied for the prediction of soil sorptivity (S-parameter) and soil steady infiltration rate (A-parameter). Input data in this study were classified into two groups (i) based on the soil readily available properties and (ii) based on the soil readily available properties and principal components (PCs) obtained by using the auxiliary data. The results indicated a better performance of ANN models derived according to the soil plus PCs data than derived models that used only soil data for predicting both S- and A-parameters. The R 2 evaluation criteria increased from 0.39 to 0.57 in predicting S-parameter and from 0.44 to 0.59 in predicting A-parameter. It was concluded that the applying environmental data, i.e., data derived from topography factors and remotely sensed information, could be potential data for improving S- and A-parameter prediction and developing high quality of infiltration parameter maps.

中文翻译：

---

基于GIS的钙质土壤入渗数字建模

摘要 在田间尺度上预测钙质材料含量高的土壤入渗对于更好地了解土地管理非常重要。本研究的目的是利用伊朗的环境数据开发基于 GIS 的钙质土壤土壤入渗数字模型。在区域尺度的 92 个点上测量了具有 3 次重复的土壤入渗数据。在每个地点，确定了土壤容易获得的特性。此外，遥感和数字高程模型数据被用作辅助数据。人工神经网络 (ANN) 模型用于预测土壤吸附性 (S 参数) 和土壤稳定入渗速率 (A 参数)。本研究中的输入数据分为两组（i）基于土壤易利用特性和（ii）基于土壤易利用特性和使用辅助数据获得的主成分（PC）。结果表明，与仅使用土壤数据预测 S 和 A 参数的派生模型相比，根据土壤加 PCs 数据派生的 ANN 模型的性能更好。R 2 评价标准在预测S参数时从0.39增加到0.57，在预测A参数时从0.44增加到0.59。得出的结论是，应用环境数据，即来自地形因素和遥感信息的数据，可能是改进 S 和 A 参数预测和开发高质量下渗参数图的潜在数据。