多源数据协同的陕西秦岭高精度土地覆盖分类研究

打开文本图片集

中图分类号：TP751;P237

文献标志码：A

doi：10.3969/j.issn.1000-1379.2026.05.020

引用格式： ， ， ， 等. 多源数据协同的陕西秦岭高精度土地覆盖分类研究 [J]. 人民黄河， 2026， 48（5）： 134-139， 149.

Research on High-Precision Land Cover Classification in Shaanxi Qinling Mountains Based on Multi-Source Data Collaboration

WANG Xinshuang 1 ， LIU Jiange 1 ， BAI Mu 1 ， WANG Bo 2

（1. Shaanxi Geomatics Center of Ministry of Natural Resources， Xi'an 710054， China;

2. Shaanxi Bureau of Surveying， Mapping and Geoinformation， Xi'an 710054， China）

Abstract： In complex terrain areas， there is a severe issue of spectral characteristics confusion among ground objects in remote sensing imagery. Taking the Qinling Mountains region in Shaanxi as the study area， a land cover classification method combining Support Vector Machine （SVM） and hierarchical segmentation was developed. First， various ground-measured sample data were integrated to construct training and validation sample sets with high spectral consistency and accuracy. SVM was used to obtain coarse classification results （of cropland， forestland， grassland， artificial surfaces， bare land， and water bodies）. The detailed expression characteristics of large-scale classification product data were fully utilized， and the classification of cropland， forestland， and grassland types was further refined by Fractional Vegetation Cover （FVC） estimates. Comparison with the classification results of the Random Forest method shows that the overall accuracy of this method reaches 88.74%， an improvement of 14.06 percentage points compared to the Random Forest method， with a Kappa coefficient of 0.82. The classification results exhibit complete patches and clear boundaries， effectively representing the continuous spatial distribution characteristics of actual land classes in the study area and revealing the “ecology-dominated， agricultural infiltration， controllable construction” human-land system features of the region.

Key words： land cover; classification; hierarchical segmentation; SVM; Qinling Mountains in Shaanxi

0 引言

土地覆盖是众多研究领域重要的基础数据源，在自然资源管理、生态环境监测、气候变化研究及相关政策制定方面具有重要作用[1-2]。（剩余10280字）