Estimating The Spatial Distribution Of GDP Based On Nighttime Light Image And Analysis Of Correlation Between It And PM_2.5 Concentration

Since the invention of the lighting technology, human gradually get rid of the darkness of night life. With the social progress and economic development, the intensity of human activity is gradually increasing at night. The emergence and development of night lights remote sensing technology provides an effective tool for monitoring the intensity of human development and construction activity. There are two types of nighttime light image released from National Oceanic and Atmospheric Administration（NOAA）/National Geophysical Information Center（NGDC）. They are the nighttime light image derived from Defense Meteorological Satellite Program（DMSP）/Operational Linescan System（OLS） and the nighttime light image derived from Visible Infrared Imaging Radiometer Suite（VIIRS）. Based on their unique ability to monitor the lights at night, these two types of nighttime light image have been widely used in the socio-economic parameters estimation, urbanization monitoring and evaluation, important event evaluation and other research fields.Although they have been widely used in many fields, there are still some problems for the application of DMSP/OLS and VIIRS nighttime light images. Due to DMSP/OLS nighttime light image have not been calibrated onboard when they are produced, the images in the long time series DMSP/OLS nighttime light dataset lack comparability. In addition, this type of nighttime light image includes a lot of saturated pixels in the bright cores of urban areas. These problems not only lead the nighttime light image cannot be compared with each other when they are used in research（such as GDP spatialization estimation） but also will affect the accuracy of relative study. As compared with above-mentioned nighttime light image, VIIRS nighttime lights image has a higher temporal and spatial resolution, and this type of nighttime light image does not include saturated pixels. Even so, the drawbacks of later release time, lack of historical images and ephemeral lights source and background noise have not been removed also affect the accuracy of their applications. Additionally, the advantage of high temporal resolution of VIIRS nighttime light image is rarely reflected in study.Firstly, this paper proposes a correction method based on the invariant region method to correct the DMSP/OLS nighttime light image. Secondly, based on the county level GDP statistic data of Guangdong Province, the corrected DMSP/OLS nighttime light image and land use/cover data are used to stimulate the GDP spatial distribution of Guangdong Province and the estimated results are applied to spatio-temporal dynamic analysis. Thirdly, based on the characteristic of VIIRS nighttime light image that they are released monthly, the spatial distribution of Guangdong Province’s GDP are simulated for each quarter. Finally, in the basis of spatial contrast between the corrected DMSP/OLS nighttime light image and PM_2.5 spatial distribution image, the research of the effects of human activities on the PM_2.5 concentration spatial distribution are carried out.（1） In order to improve the comparability of DMSP/OLS stable nighttime light image data set and remove the saturated pixels in this type of image, this article developed a correction method based on the invariant region method. For the analysis of the saturation region and the non-saturation region in the nighttime light image, different invariant regions were selected. After radiance calibration nighttime light image in 2006 which does not include saturated pixels was determined as reference image, the correction model was built and the stable nighttime light images were corrected based on the correlation of the invariant regions between the stable nighttime light image and the reference image. By comparison between the pixels in the uncorrected nighttime light image and which in the corrected nighttime light image and verification based on socio-economic statistic data, the stable nighttime light images after correction could compare with each other and pixel saturation had also been improved.（2） The different GDP spatialization model of Guangdong were established based on corrected DMSP/OLS nighttime light images and based on corrected DMSP/OLS nighttime light images and land use/cover data. After calculating and comparing the two types of gridded GDP simulation results, the latter model showed higher simulation accuracy. In this method, the corrected DMSP/OLS nighttime light images were used to simulate second and third industry output and the land use/cover data was used to estimate the output value of primary industry in Guangdong Province. These two types of simulated results were combined to the final GDP spatialization results for Guangdong Province. A simple spatial and temporal variation analysis of Guangdong Province’s GDP spatial distribution was carried out by the estimated results of this GDP spatialization simulation approach. The results showed that in terms of both the Guangdong’s GDP value and GDP growth rate, the Pearl River delta was at the top, followed by eastern and western Guangdong, and the northern mountainous area of Guangdong came last.（3） In this article, the VIIRS nighttime light images were used to simulate the GDP spaitial distribution of Guangdong Province after the ephemeral light source and background noise in this type of image having been removed. By comparing with GDP spatialization simulation results based on DMSP/OLS nighttime light images before and after correction, simulation results based on VIIRS nighttime light image has the higher precision. In addition, based on Guangdong-quarter GDP statistic data, the GDP spatialization of each quarter from 2014 to 2015 were estimated using this type of nighttime light image. This study also tried to use 2014 annual VIIRS nighttime light image to simulate GDP spatialization for each quarter. These two types of quarterly results have a high degree of consistency. The results not only achieve the applications of VIIRS nighttime light image in higher time resolution, but also provide a fast simulation tool for spatial distribution of quarterly socio-economic statistic data.（4） A variety of average PM_2.5 concentrations images of Chinese mainland were used to analyze the spatial distribution and time-varying characteristics of Chinese PM_2.5 concentrations. Subsequently, by comparing the light intensity values of corrected DMSP/OLS stable nighttime light and PM_2.5 concentration values of the corresponding PM_2.5 concentration spatial distribution images, the tentative analysis of human activities on the PM_2.5 concentration spatial distribution was carried out. The results showed that there was a good spatial consistency between high intensity of human activities region in the corrected stable nighttime light image and the high PM_2.5 concentration region which had a greater human contribution. Finally, Chinese three major urban agglomerations and cities in them were analyzed for relative GDP quality, and the results showed that the Pearl River Delta region had a relatively high GDP quality, as its economic development model put relatively more emphasis on the coordinated development of economy and the environment.