期刊：
International Journal of Environmental Research and Public Health,2018年15(5):1032- ISSN：1661-7827
通讯作者：
Zhang, Qiuwen;Zhang, Gui
作者机构：
[Zhang, Xike; Zhang, Qiuwen] Huazhong Univ Sci & Technol, Sch Hydropower & Informat Engn, Wuhan 430074, Hubei, Peoples R China.;[Zhang, Xike; Nie, Zhiping] Hunan City Univ, Sch Municipal & Mapping Engn, Yiyang 413000, Peoples R China.;[Zhang, Gui; Que, Huafei] Cent South Univ Forestry & Technol, Key Lab Digital Dongting Lake Basin Hunan Prov, Changsha 410004, Hunan, Peoples R China.;[Gui, Zifan] Shenzhen Garden Management Ctr, Shenzhen 518000, Peoples R China.
通讯机构：
[Zhang, Qiuwen] H;[Zhang, Gui] C;Huazhong Univ Sci & Technol, Sch Hydropower & Informat Engn, Wuhan 430074, Hubei, Peoples R China.;Cent South Univ Forestry & Technol, Key Lab Digital Dongting Lake Basin Hunan Prov, Changsha 410004, Hunan, Peoples R China.
关键词：
daily land surface temperature;forecasting;data-driven;hybrid model;Ensemble Empirical Mode Decomposition (EEMD);Long Short-Term Memory (LSTM);Neural Network (NN);Dongting Lake basin
摘要：
Daily land surface temperature (LST) forecasting is of great significance for application in climate-related, agricultural, eco-environmental, or industrial studies. Hybrid data-driven prediction models using Ensemble Empirical Mode Composition (EEMD) coupled with Machine Learning (ML) algorithms are useful for achieving these purposes because they can reduce the difficulty of modeling, require less history data, are easy to develop, and are less complex than physical models. In this article, a computationally simple, less data-intensive, fast and efficient novel hybrid data-driven model called the EEMD Long Short-Term Memory (LSTM) neural network, namely EEMD-LSTM, is proposed to reduce the difficulty of modeling and to improve prediction accuracy. The daily LST data series from the Mapoling and Zhijaing stations in the Dongting Lake basin, central south China, from 1 January 2014 to 31 December 2016 is used as a case study. The EEMD is firstly employed to decompose the original daily LST data series into many Intrinsic Mode Functions (IMFs) and a single residue item. Then, the Partial Autocorrelation Function (PACF) is used to obtain the number of input data sample points for LSTM models. Next, the LSTM models are constructed to predict the decompositions. All the predicted results of the decompositions are aggregated as the final daily LST. Finally, the prediction performance of the hybrid EEMD-LSTM model is assessed in terms of the Mean Square Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), Pearson Correlation Coefficient (CC) and Nash-Sutcliffe Coefficient of Efficiency (NSCE). To validate the hybrid data-driven model, the hybrid EEMD-LSTM model is compared with the Recurrent Neural Network (RNN), LSTM and Empirical Mode Decomposition (EMD) coupled with RNN, EMD-LSTM and EEMD-RNN models, and their comparison results demonstrate that the hybrid EEMD-LSTM model performs better than the other five models. The scatterplots of the predicted results of the six models versus the original daily LST data series show that the hybrid EEMD-LSTM model is superior to the other five models. It is concluded that the proposed hybrid EEMD-LSTM model in this study is a suitable tool for temperature forecasting.
关键词：
Land cover change;Dongting Lake;support vector machine;Landsat TM/OLI
摘要：
Land cover in Dongting Lake region has been faced high variations during recent decades. Therefore , there has a strong need to investigate and understand the land cover changes in Dongting Lake region between land cover types. In this study, support vector machine (SVM) classification method was employed to detect changes in land cover dynamic in Dongting Lake region using Landsat images for the year 1995, 2006 and 2015. Land cover information was classified to five categories: waterbody, wetland, built-up, cropland and forestland. Quantitative analysis , change detection matrix and land cover dynamic degree were utilized for investigating and assessing the land cover changes in Dongting Lake region. The overall accuracy (OA) and kappa coefficient of the land cover classification results were over 96% and 0.9, respectively. The results indicated that in 1995 about 11.32% of study area was covered by water-body together with 13.31% of wetland. Nearly 50% of the area was covered by cropland and remaining 2.59% was covered by built-up. During the period 1995-2015, the change rate of the waterbody was evaluated at-0.29%, at-0.67% for the wetland and at-2.47% for the built-up. On the contrary, the for-estland and cropland increased by 0.72% and 0.03%, respectively. In addition, the results of this study can provide scientific information for government to formulate policy for sustainable land use management in Dongting Lake region. Land cover change, Dongting Lake, support vector machine , Landsat TM/OLI Land cover changes affect global climate, species diversity and ecosystem balance, which can accelerate land degradation and reduce ecosystem services [1, 2]. It has become a serious environmental problem. Over the past few decades, land cover in Dongting Lake region experienced tremendous changes by natural processes, as well as anthropo-genic activities [3]. In particular, anthropogenic activities , such as reclaiming cropland from lakes and returning cropland to lakes, have become a major concern of land cover changes in Dongting Lake region [4]. Therefore, a clear understanding of the spatial and temporal changes of land cover types in the Dongting Lake region in recent two decades is important. Remote sensing has been monitoring and capturing the earth land's surface every day and night by providing spatial and temporal images over large and inaccessible area for more than six decades [5]. Therefore, remote sensing became an acknowledged technology for monitoring the land cover changes. Some optical remote sensing products, such as Moderate Images Spectrometer (MODIS), Advanced Very High-Resolution Radiometer (AVHRR), and Satellite Pour 1'Obervation de la Terre (SPOT) with resolution at 250 m to 1 km, are the very suitable data resources for studying information of earth surface [1, 6-11]. Despite short revisiting cycle and large swath width, these low-resolution products are mainly available on the detecting of large scale coarse land cover changes, but the transformation details of land cover types and its ratio remains unknown which usually occurs at a small scale. In order to settle these problems and detail monitoring earth's land cover changes, medium remote sensing satellite data, such as Landsat Thematic Mapper (TM) [12, 13], Landsat Enhanced Thematic Mapper Plus (ETM+) [7, 14] and Landsat Operational Land Im-ager (OLI) [12, 15], with resolution of 30 m but re-visiting cycle of 16 day, have been widely utilized for mapping land cover and monitoring its changes. Numerous researches have been conducted and various algorithms have been developed for detecting land cover changes especially over Dongting Lake region using remote sensing satellite technologies. Li et al. [16] employed the Geographical Information System (GIS) and Remote Sensing (RS) technologies to study the characterized long-term land cover changes in Dongting Lake region using the Landsat images from 1978, 1989, 1998. Their results indicated that land cover patterns in Dongting Lake region had been greatly altered by empoldering. Three land type had changed remarkably. The cultivated land decreased,
关键词：
Surface water area variation;flood inundation frequency;Dongting Lake;The Three Gorges Reservoir;remote sensing monitoring;Landsat
摘要：
Dongting Lake is the second largest freshwater lake in China, with rapid seasonal surface water area fluctuations in the middle reach of the Yangtze River and downstream from the Three Gorges Reservoir (TGR). The marked variation of the lake's surface water area is considered to have been affected by the TGR over the past decades. In this study, Landsat TM/ETM+/OLI time-series imagery data were employed to estimate the wet season total surface water area variation in South Dongting Lake and East Dongting Lake from 1988 to 2016. The surface water area was extracted from Landsat data using Modified Normalized Difference Water Index (MNDWI). The results indicated that the surface water area variation was accordant with the variation of precipitation and runoff of Xiangtan, Taojiang, Taoyuan, Shimen, Shadaoguan, Mituosi, Ouchikou and Chenglingji five hydrological stations. Most of the large surface water areas were observed during the pre-TGR period, whereas the small surface water areas were observed during the post-TGR period. Surface water area data, precipitation and runoff from the four hydrological stations (except Shimen station) all indicated downward trends, with reduction rates of 5.866 km(2).year(-1), 0.802 mm.year(-1), 3.950 10(8)m(3).year(-1), 2.834 10(8)m(3).year(-1), 0.377 10(8)m(3).year(-1), 1.282 10(8)m(3).year(-1), 2.715 10(8)m(3).year(-1) , 0.318 10(8)m(3).year(-1) and 16.114 10(8)m(3).year(-1), respectively. The results of correlation analyses indicated precipitation and water from the Yangtze river may affect the fluctuation in the surface water area to a large extent. The results of flood inundation probability analysis indicated that approximately 502.9 km(2) of the study area was in the high flood hazard zone. In addition, the results of this study can provide scientific information to understand the effect of the TGR on downstream lakes and achieve better water resources and flood hazard management in this region.
作者机构：
[郭际明; 黄长军; 喻小东] School of Geodesy and Geomatics, Wuhan University, Wuhan 430079, China;[聂智平; 黄长军] School of Municipal and Surveying Engineering, Hunan City University, Yiyang 413000, China
通讯机构：
School of Geodesy and Geomatics, Wuhan University, China
