基于聚类和分类与回归树的地力等级评价研究

Assessment of Farmland Productivity with Cluster Analysis and Classification and Regression Tree(CART)

以黄淮海平原粮食主产区河南封丘县为研究区域,利用基于GIS的土壤空间和属性数据库,采用聚类分析和分类与回归树(CART)相结合的方法建立了耕地地力评价模型。研究结果表明,基于聚类分析和CART的地力评价模型准确度为93.56%,较单独使用决策树模型的准确度有明显提高;根据耕地地力分级规则,一等地至五等地分别占全县61 733.3 hm2耕地的28.167%、49.518%、9.389%、5.77%和7.156%;地力等级较高的耕地主要分布于封丘西北部,地力较低的区域主要在东南部,由西北向东南地力呈带状递减趋势。本文的研究结果可为当地中低产田及其障碍因子的解析和农田精准管理提供参考依据。

In the present study, a combination of classification and regression tree （CART） and cluster analysis method was applied in assessing the farmland productivity in Fengqiu County, Henan Province, based on county-level soil spatial and attributive databases. The results indicated that the prediction accuracy of the proposed combination model was considerably improved （to 93.56%） as compared to that by CART approach alone. According to the resulting grading rules, the first, second and third grade farmland accounts for 28.167%, 49.518% and 9.389% of the total area of farmland in this county, respectively; while the fourth and fifth grade farmland accounts for only 5.77% and 7.156% of the farmland, respectively. The higher grading land was mainly distributed in the northwest of Fengqiu County, while the lower grading land was mainly located in the southeast region. There was also an obvious banded decreasing trend of farmland productivity extending from the northwest to the southeast. The results of this paper may help to analyze the spatial distribution of the middle to low-yield fields and limiting factors for improving grain yield, and may also provide references for decision-making on regional farmland management.