基于夜间灯光影像的中国电力消耗量估算及时空动态

提出夜间灯光降饱和指数模型,以中国大陆为研究对象,基于DMSP/OLS夜间灯光数据、MODIS NDVI产品、基础地理信息数据及社会经济统计数据,构建电力消耗估算模型,定量估算了2000-2012年电力消耗量,并采用空间统计分析方法,从不同时间、空间角度对省级、地级和县级单元的电力消耗量变化趋势和空间集聚程度进行分析。结果表明：夜间灯光降饱和指数模型能较好地降低夜间灯光的数据饱和和溢出,其中MDNVI模型的效果最好。从县级尺度电力消耗变化趋势的显著性来看,无明显变化区域主要出现在青藏高原,迅猛增长型多数分布在京津冀、长三角、珠三角和中东部省会城市。

Estimate and spatio-temporal dynamics of electricity consumption in China based on DMSP/OLS images

Nighttime light (NTL) data from the Defense Meteorological Satellite Program/Operational Linescan System (DMSP/OLS) are able to provide information on nighttime luminosity, a correlation of the built environment and energy consumption. NVI (Nighttime-light Vegetation Index), RNVI (Ratio Nighttime-light Vegetation Index), DNVI (Difference Nighttime-light Vegetation Index), NDNVI (Normalized Difference Nighttime-light Vegetation Index), SANVI (Soil-adjusted Nighttime-light Vegetation Index) and MDNVI (Modified Difference Nighttime-light Vegetation Index) were used to compensate for the shortage of DMSP/OLS data. DMSP/OLS night lights data, MODIS NDVI products, China GIS database and socio-economic statistical data are also taken into consideration. An electricity consumption estimation model is used to obtain a figure for electricity consumption from 2000 to 2012. Lastly, we divide electricity consumption into four ratings and analyze the spatio-temporal patterns by using ESDA method. Results are as follows: All of the indexes can compensate the shortage of DMSP/OLS data, among which MDNVI is the best model. We reduced spatial overflow effect of night lights data by using MDNVI model. Then we built a linear regression model of electricity consumption by regression analysis, and we used it for DMSP/OLS data to retrieve China's electricity consumption spatial layout. We compared the MRE (mean relative error) between the result and related research, which proves that our result has a lower MRE and a higher accuracy. Finally we find a way to obtain China electricity consumption data from 2000 to 2012 quickly and effectively. Electric consumption grew quickly in China from 2000 to 2012; on the whole, maximum electric consumption increased from 6.79 M kW\begin{array}{c}?\end{array}h to14.82 M kW\begin{array}{c}?\end{array}h. We discovered, using downscaling analysis, electricity consumption showed significant differences within regions. We analyzed electricity consumption levels, Moran's I and LISA cluster in the study area from 2000 to 2012 by using statistical data. Results showed that generating capacity and electricity consumption of 31 provinces have a strong spatial correlation. Gradually formed four "HH" cluster areas, namely Langfang-Tianjin, the Pearl River Delta, Shanghai-Hangzhou-Nanjing and the West Bank of the Taiwan Straits from 2000 to 2012. Spatial agglomeration of electricity consumption at county scale is significant, with "HH" clusters mainly located in Beijing-Tianjin, the Yangtze River Delta, Pearl River Delta, Shandong Peninsula, Changchun-Harbin-Dalian area, and North Tianshan Mountains area. The "LL" pattern shows a gradual trend moving from the south-eastern edge of the Qinghai-Tibet Plateau to the Tibetan Plateau. The index proposed here combines information from both DMSP/OLS NTL data and MODIS NDVI data for more detailed characterizations in nighttime luminosity. Our assessments confirm its ability to reduce the NTL saturation. Moreover, its simplicity enables rapid characterization and monitoring of electricity consumption.