• ▶ 2008-2024年被中国情报信息研究所评价中心评为“中国科技核心期刊”
  • ▶ 2019-2024年连续三届被中国科学院文献情报中心中国科学引文数据库CSCD(核心库)收录
  • ▶ 2021、2023年入编北京大学图书馆《中文核心期刊要目总览》
  • ▶ 2020-2024连续四年入选《科技期刊世界影响力指数(WJCI)报告》
王 晓, 刘海新, 孙振宇, 王佳璘, 朱 研.融合多源数据与机器学习算法的综合干旱监测模型构建——以河北省为例水资源与水工程学报[J].,2024,35(6):207-219
融合多源数据与机器学习算法的综合干旱监测模型构建——以河北省为例
Integrated modeling for drought monitoring based on multi-source data and machine learning algorithm: a case study of Hebei Province
  
DOI:10.11705/j.issn.1672-643X.2024.06.21
中文关键词:  多源数据  土壤相对湿度  机器学习  干旱监测  河北省
英文关键词:multi-source data  soil relative humidity  machine learning  drought monitoring  Hebei Province
基金项目:国家自然科学基金项目(42071246、42171212);河北省自然科学基金项目(E2020402006、E2020402086、D2022402030)
作者单位
王 晓, 刘海新, 孙振宇, 王佳璘, 朱 研 (河北工程大学 矿业与测绘工程学院河北 邯郸 056038) 
摘要点击次数: 10
全文下载次数: 1
中文摘要:
      基于遥感数据、再分析数据以及土壤墒情站点数据,采用支持向量机(SVM)、随机森林(RF)、径向基函数神经网络(RBFNN)和极限学习机(ELM)4种经典的机器学习方法,构建了2001—2022年河北省不同生态分区植被生长季月际尺度的综合干旱监测组合模型,并结合Sen趋势揭示了河北省干旱空间变化情况。结果表明:RF、RBFNN和SVM的拟合效果在生长季不同月份以及不同生态分区下均表现出较强的稳定性,ELM拟合效果相对较差,因此,基于不同生态分区选取RF-RBF-SVM组合模型来构建河北省综合干旱监测模型;该模型的拟合性能在湿润环境下监测能力表现更优,时间范围上以植被生长季中期监测能力表现更优;模型的总体平均精度为85.15%,模拟值与站点实测值的年变化趋势一致性较高;研究时段内SM10的Sen趋势度的空间变化均值为0.174/a,土壤相对湿度呈增大趋势的面积占研究区总面积的77.21%,干旱情况有所缓解。
英文摘要:
      Based on remote sensing, reanalysis and soil moisture site data, an integrated drought monitoring model on monthly scale was constructed for vegetation of growing seasons in different ecological zones of Hebei Province from 2001 to 2022. Four classic machine learning methods, namely, support vector machine (SVM), random forest (RF), radial basis function neural network (RBFNN) and extreme learning machine (ELM) were adopted for the modeling. Then combined with Sen’s slope analysis, the spatial variation of drought in Hebei Province was revealed. The results showed that the fitting effect of RF, RBFNN and SVM presented strong stability in different months of the growing season, as well as in different ecological zones, while that of ELM was relatively poor. Therefore, the integrated RF-RBFNN-SVM model was selected based on different ecological zones for the modeling of comprehensive drought monitoring in Hebei Province. The application of the model indicated that its fitting performance was better in the wet environment, and in the middle of the vegetation growing season. The average accuracy of the model was 85.15%, and the predicted value was in good agreement with the measured value. The average spatial variation of SM10 was 0.174/a in the study period, and soil relative humidity in 77.21% of the area showed an upward trend, and the drought situation was alleviated.
查看全文  查看/发表评论  下载PDF阅读器
关闭