| 卫星面降水估计人工神经网络方法 |
| |
| 引用本文： | 师春香,卢乃锰,张文建.卫星面降水估计人工神经网络方法[J].气候与环境研究,2001,6(4):467-472. |
| |
| 作者姓名： | 师春香 卢乃锰 张文建 |
| |
| 作者单位： | 国家卫星气象中心,北京,100081 |
| |
| 摘 要： | 介绍了人工神经网络原理和卫星云图估计降水的原理.从GMS红外卫星云图资料中抽取12个降水云图特征量,构造了网络结构为12-98-7的降水估计人工神经网络模型,并用1993年的小时地面雨量资料和GMS数字云图资料对神经网络模型进行训练,用1992和1994年资料对该神经网络模型分别进行测试.在日面降水估计试验中,地面雨量计值和卫星估计降水之间的相关系数分别为0.94和0.97,相对误差分别为41%和32%.
|
| 关 键 词： | 卫星云图 降水估计 人工神经网络 |
| 修稿时间： | 2000年4月18日 |
Precipitation Estimation from Remotely Sensed Information Using Nueral Networks |
| |
| Shi Chunxiang,Lu Naimeng and Zhang Wenjian.Precipitation Estimation from Remotely Sensed Information Using Nueral Networks[J].Climatic and Environmental Research,2001,6(4):467-472. |
| |
| Authors: | Shi Chunxiang Lu Naimeng and Zhang Wenjian |
| |
| Abstract: | We designed a model with 12 inputs, one hidden layers and 7 outputs (12-98-7) for precip-itation estimation using GMS IR image. The neural network model is trained with hourly ground ob-servations and 12 satellite imagery features data from summer 1993 and tested with data from summer 1992 and 1994. Then it is applied to estimating daily area rainfall. The results reveal that their errors is31.8% and 41.0% for daily area rainfall estimation, and their correlation coefficients are 0.97 and 0.94. |
| |
| Keywords: | satellite image precipitation estimate artificial neural network |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《气候与环境研究》浏览原始摘要信息 |
|
点击此处可从《气候与环境研究》下载全文 |
|
