基于多源地理数据和随机森林模型的土壤类型模拟预测研究——以宁洱县为例

Soil Type Simulation and Prediction Based on Multi-source Geographical Data and Random Forest Model:Taking Ning er County as an Example

以宁洱县为研究区域,在第二次全国土壤普查(简称二普)的基础上,利用第三次全国土壤普查(简称三普)的土壤调查资料和数字高程模型、遥感影像等多源地理数据,运用GIS工具提取一系列的土壤成土环境因素,在R语言中,采用随机森林模型建立土壤类型与多源地理数据之间的映射关系,从而预测未知地区的土壤类型。根据预测结果对土壤类型图进行制图与更新,继承和发展二普成果,形成土壤三普各级土壤类型图。结果显示,基于多源地理数据和随机森林模型的土壤类型预测方法具有较高的准确性和可靠性。预测结果与实际土壤类型分布较为吻合,验证了该方法的有效性和可行性。

Taking Ning er County as a research region,based on the second national soil survey,using the data from the third national soil survey and multi-source geographical data such as digital elevation models and remote sensing images,a series of soil forming environmental factors were extracted using GIS tools.In R language,a Random Forest model was used to establish a mapping relationship between soil types and multi-source geographical data,thereby predicting the soil types in unknown areas.Then,according to prediction results,soil type maps were created and updated,inheriting and developing the existing soil survey results,thus generating soil type maps at all levels of the third national soil survey.The results showed that the soil type prediction method based on multi-source geographical data and Random Forest model has high accuracy and reliability.The predicted results are in good agreement with the actual distribution of soil types,verifying the effectiveness and feasibility of this method.