首页> 中文期刊> 《环境科学研究》 >基于神经网络的三江源区草地地上生物量估算

基于神经网络的三江源区草地地上生物量估算

         

摘要

三江源区位于青藏高原腹地,作为长江、黄河、澜沧江三大河流的发源地,是我国重要的水源涵养和生态功能保护区.为了及时准确地获取该区域草地生物量信息,根据三江源区高寒草甸、高寒草原采样点的地上生物量实测值,结合遥感植被指数、海拔、气象观测数据(光合有效辐射、年均气温、年降水量)构建BP神经网络模型,估算2001-2010年三江源区的草地地上生物量,并对其进行分县统计和年际变化分析.结果表明:①通过多次反复的训练与测验得到的BP神经网络模型,对高寒草甸、高寒草原的地上生物量模拟值与实测值的R2分别为0.73、0.79,表明BP神经网络模型具有较好的模拟效果.②2001-2010年三江源区草地地上生物量多年平均值为172.34 g/m2,其中高寒草甸为214.81 g/m2,高寒草原为130.07 g/m2.③三江源区草地地上生物量的空间分布具有明显的空间异质性,呈从东南向西北递减的趋势.其中,位于东部的河南县草地地上生物量最高,为413.46g/m2;而北部的曲麻莱最低,仅为69.04 g/m2.④2001-2010年三江源区草地地上生物量呈缓慢波动上升趋势,平均升幅为0.93g/(m2·a).研究显示,利用站点地上生物量实测数据构建BP神经网络模型并对地上生物量进行模拟,对于分析区域尺度的草地地上生物量分布格局和变化趋势行之有效.%The Three-River Headwaters Region,located at the Qinghai-Tibet plateau,is the source of the Yangtze River,Yellow River and Lantsang River,and an important water source and ecological function conservation area.Grassland is the most widely distributed ecosystem in the Three River Headwaters Region.Timely and accurate estimation of the grassland biomass is significant for protecting the grassland resources.Based on the aboveground biomass of alpine meadow and alpine steppe grasslands,the aboveground biomass model was constructed by training a BP neural network.Driven by remote sensing (EVI),elevation and meteorological data including photosynthetically active radiation,temperature and precipitation,the regional aboveground biomass of grasslands during 2001-2010 was simulated by the model.The results showed that:(1) The BP neural network model performed well in the aboveground biomass estimation,with the correlation coefficient of determination between the predicted and measured aboveground biomass of alpine meadow and alpine steppe being 0.73 and 0.79,respectively.(2) The annual average aboveground biomass of grasslands during 2001-2010 in the Three River Headwaters Region was 172.34 g/m2,with greater value (214.81 g/m2) in alpine meadow than in alpine steppe (130.07 g/m2).(3) The spatial pattern of grassland aboveground biomass in the Three River Headwaters Region was heterogeneous,lower in the northwest and higher in the southeast.The average grassland aboveground biomass of Henan county was the highest (413.46 g/m2),and that of Qumalai County was the lowest (69.04 g/m2).(4) There was a slightly increasing trend in grassland aboveground biomass from 2001 to 2010,with an increasing rate of 0.93 g/(m2 · a).These results indicated that the BP neural network model we constructed in this paper could not only effectively simulate the grassland aboveground biomass,but also provided an approach for the analysis of the spatial-temporal variation in the regional scale grassland aboveground biomass.

著录项

  • 来源
    《环境科学研究》 |2017年第1期|59-66|共8页
  • 作者单位

    中国科学院地理科学与资源研究所,生态系统网络观测与模拟重点实验室,北京100101;

    中国科学院大学,北京100049;

    中国科学院地理科学与资源研究所,生态系统网络观测与模拟重点实验室,北京100101;

    中国科学院地理科学与资源研究所,生态系统网络观测与模拟重点实验室,北京100101;

    中国科学院大学,北京100049;

    中国科学院地理科学与资源研究所,生态系统网络观测与模拟重点实验室,北京100101;

    中国科学院大学,北京100049;

    中国科学院大学,北京100049;

    中国科学院地球化学研究所,贵州贵阳550002;

    青海省生态环境遥感监测中心,青海西宁810007;

    中国环境科学研究院,北京 100012;

  • 原文格式 PDF
  • 正文语种 chi
  • 中图分类 生态系统与生态环境;
  • 关键词

    草地地上生物量; 三江源; BP神经网络; 时空变化;

相似文献

  • 中文文献
  • 外文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号