含沙量预测对流域泥沙治理、水沙调控以及水质与水环境管理等具有重要意义。长江上游地区幅员广阔,支流众多,水沙来源复杂,对准确预测三峡入库含沙量过程构成了挑战。针对长江上游区间降雨和干支流来水来沙对寸滩站含沙量产生不同程度...含沙量预测对流域泥沙治理、水沙调控以及水质与水环境管理等具有重要意义。长江上游地区幅员广阔,支流众多,水沙来源复杂,对准确预测三峡入库含沙量过程构成了挑战。针对长江上游区间降雨和干支流来水来沙对寸滩站含沙量产生不同程度的影响,提出了一种基于随机森林(Random Forest,RF)与长短时记忆(Long Short Term Memory,LSTM)神经网络结合的日含沙量预测深度学习模型RF-LSTM。首先,该模型利用RF算法筛选出与寸滩站日含沙量相关性强的水沙因子,然后将这些因子作为LSTM神经网络的输入变量,进一步识别出优选水沙因子与寸滩含沙量之间的映射关系,最后以长江上游向家坝至寸滩区间为研究区域,应用该模型对不同预见期下的寸滩站汛期日含沙量进行了预测,结果表明:与LSTM模型相比,RF-LSTM模型能较好地考虑预测因子对含沙量影响的滞后效应,且有效捕获与寸滩站日含沙量相关性强的特征,四种预见期下其在预测精度和性能方面均有较好表现,其中无预见期和预见期1 d时两种模型预测精度均较高,验证期的纳什效率系数均大于0.82,无预见期下RF-LSTM模型的纳什效率系数可达到0.91,相应的均方根误差和平均绝对误差分别较LSTM模型降低了13%和8%,且两种预见期下RF-LSTM模型可以较为准确捕获沙峰及峰现时间;当预见期增加至2 d和3 d时两种模型精度均有明显下降,但RF-LSTM模型计算精度仍优于LSTM模型。研究结果可为长江上游含沙量预测提供参考。展开更多
This research investigates the ecological importance,changes,and status of mangrove wetlands along China’s coastline.Visual interpretation,geological surveys,and ISO clustering unsupervised classification methods are...This research investigates the ecological importance,changes,and status of mangrove wetlands along China’s coastline.Visual interpretation,geological surveys,and ISO clustering unsupervised classification methods are employed to interpret mangrove distribution from remote sensing images from 2021,utilizing ArcGIS software platform.Furthermore,the carbon storage capacity of mangrove wetlands is quantified using the carbon storage module of InVEST model.Results show that the mangrove wetlands in China covered an area of 278.85 km2 in 2021,predominantly distributed in Hainan,Guangxi,Guangdong,Fujian,Zhejiang,Taiwan,Hong Kong,and Macao.The total carbon storage is assessed at 2.11×10^(6) t,with specific regional data provided.Trends since the 1950s reveal periods of increase,decrease,sharp decrease,and slight-steady increases in mangrove areas in China.An important finding is the predominant replacement of natural coastlines adjacent to mangrove wetlands by artificial ones,highlighting the need for creating suitable spaces for mangrove restoration.This study is poised to guide future mangroverelated investigations and conservation strategies.展开更多
文摘含沙量预测对流域泥沙治理、水沙调控以及水质与水环境管理等具有重要意义。长江上游地区幅员广阔,支流众多,水沙来源复杂,对准确预测三峡入库含沙量过程构成了挑战。针对长江上游区间降雨和干支流来水来沙对寸滩站含沙量产生不同程度的影响,提出了一种基于随机森林(Random Forest,RF)与长短时记忆(Long Short Term Memory,LSTM)神经网络结合的日含沙量预测深度学习模型RF-LSTM。首先,该模型利用RF算法筛选出与寸滩站日含沙量相关性强的水沙因子,然后将这些因子作为LSTM神经网络的输入变量,进一步识别出优选水沙因子与寸滩含沙量之间的映射关系,最后以长江上游向家坝至寸滩区间为研究区域,应用该模型对不同预见期下的寸滩站汛期日含沙量进行了预测,结果表明:与LSTM模型相比,RF-LSTM模型能较好地考虑预测因子对含沙量影响的滞后效应,且有效捕获与寸滩站日含沙量相关性强的特征,四种预见期下其在预测精度和性能方面均有较好表现,其中无预见期和预见期1 d时两种模型预测精度均较高,验证期的纳什效率系数均大于0.82,无预见期下RF-LSTM模型的纳什效率系数可达到0.91,相应的均方根误差和平均绝对误差分别较LSTM模型降低了13%和8%,且两种预见期下RF-LSTM模型可以较为准确捕获沙峰及峰现时间;当预见期增加至2 d和3 d时两种模型精度均有明显下降,但RF-LSTM模型计算精度仍优于LSTM模型。研究结果可为长江上游含沙量预测提供参考。
基金supported by China Geological Survey(DD20211301).
文摘This research investigates the ecological importance,changes,and status of mangrove wetlands along China’s coastline.Visual interpretation,geological surveys,and ISO clustering unsupervised classification methods are employed to interpret mangrove distribution from remote sensing images from 2021,utilizing ArcGIS software platform.Furthermore,the carbon storage capacity of mangrove wetlands is quantified using the carbon storage module of InVEST model.Results show that the mangrove wetlands in China covered an area of 278.85 km2 in 2021,predominantly distributed in Hainan,Guangxi,Guangdong,Fujian,Zhejiang,Taiwan,Hong Kong,and Macao.The total carbon storage is assessed at 2.11×10^(6) t,with specific regional data provided.Trends since the 1950s reveal periods of increase,decrease,sharp decrease,and slight-steady increases in mangrove areas in China.An important finding is the predominant replacement of natural coastlines adjacent to mangrove wetlands by artificial ones,highlighting the need for creating suitable spaces for mangrove restoration.This study is poised to guide future mangroverelated investigations and conservation strategies.
