Mapping And Characterization Of Photovoltaic Power Stations Based On Intelligent Extraction From Satellite Remote Sensing

Solar energy,as a green,renewable,and clean energy source,has become an important solution to climate change and energy shortages worldwide within the context of sustainable development.Technological innovations and the expansion of photovoltaic(PV)power generation in recent decades have driven the rapid development of China’s PV industry,where the cost of photovoltaic kilowatt-hour electricity has been declining.Driven by the dual carbon strategy and the energy supply and demand gap,China’s coastal provinces and municipalities have accelerated the construction process of PV power stations.However,the rapid development of PV power generation in coastal provinces and urban areas has also aggravated land conflicts and may have potential negative impacts on the local ecology.Therefore,accurate PV spatial distribution data are necessary for the rational planning of the PV industry,light resource utilization assessment,and ecological evaluation.Compared with field survey methods,remote sensing has low cost,high efficiency,and wide coverage,making it advantageous for dynamic monitoring and environmental impact assessment of PV power stations.Although previous studies have focused on remote sensing extraction of PV power stations,most are small-scale identification methods based on deep learning.Limitations of high data acquisition and computation costs restrict the application of deep learning methods to large-scale PV spatial mapping.Furthermore,most existing large-scale PV spatial mapping studies use only single optical information,which has limited accuracy in the range of coastal zones that are severely affected by clouds.Therefore,to quickly and accurately determine the spatial distribution and development of PV in China’s coastal provinces and municipalities,this study constructs an automatic PV extraction method.The method uses multi-layer information such as optical,radar,and texture information as input parameters through the GEE cloud computing platform based on Sentinel-1 and Sentinel-2satellite image data.It effectively alleviates the problem of cloud obscuration and feature confusion in the coastal zone area using a single optical image.This method enables the fast thematic mapping of PV power plant distribution in coastal provinces and municipalities at the national scale.This study analyzed the characteristics and spatial distribution of PV stations based on the extraction results.It also analyzed and discussed the causes of their distribution from physical geography and socio-economic perspectives.The main conclusions of the study include:(1)PV power stations exhibit unique spectral waveform and signal characteristics on optical and radar images.However,the spacing,inclination,orientation,and height of PV panels vary greatly according to the environment and needs of different regions,resulting in strong spatial heterogeneity in remote sensing images.The random forest method,which combines spectral,radar,and texture features of PV stations,has demonstrated good accuracy in the extraction of PV stations in the coastal zone region of China.The short-wave infrared band ratio is an important feature for distinguishing between PV and non-PV targets.The overall accuracy of the classification reached 96.9%,with a Kappa coefficient of 0.91,after accuracy testing.(2)As of October 2022,the total PV area in China’s coastal provinces and municipalities reached 837.3 km~2,with the majority of it distributed in northern areas such as Hebei and Shandong,dominated by land-based PV stations.The top 5%of the area accounted for 50.2%of the total PV plant area in the study area.The number and scale of PV stations varied greatly from city to city,with northern regions having more PV projects and larger PV plant construction scales and densities.The provinces and cities with the highest construction density differed significantly from those with the lowest.With increasing administrative shoreline landward distance,the construction density of PV power stations and water-based PV power stations gradually decreased,with the construction density of PV power stations within a 1 km buffer zone being as high as 42.5 km~2/10~4 km~2,approximately seven times higher than the average construction density in China’s coastal provinces and urban areas.The rapid development of water-based photovoltaic power stations,represented by the"fishery-photovoltaic complementary"model,is the primary reason for the high PV construction density in near-shore areas.The main sources of land for PV plant construction in China’s coastal provinces and urban areas are arable land,water,and bare land.Woodland contributes about one-fifth of the land for PV plant construction in the southern region,while PV plant construction in the northern region relies more on cropland,bare land,and shrub growth areas.Water areas are the main source of land for PV plant construction in nearshore areas.(3)The construction of PV stations in China’s coastal provinces and municipalities generally matches the corresponding conditions on the production and supply side,although there is still a local mismatch.Overall,the development potential of the northern region is higher than that of the south,with Liaoning Province having more room for development.The degree of light resource utilization and the level of PV power supply in the south are both lower,especially in the Pearl River Delta city cluster,which needs to be the focus in achieving the double carbon goal in the future.(4)The distribution of PV power stations is influenced by various natural and human factors.Areas with higher light radiation intensity,flat topography,and abundant land supply tend to improve the power generation efficiency of PV power stations and reduce their construction costs.These factors or their combination make such areas more attractive to PV projects.Additionally,PV industry policies in different provincial and municipal areas influence the pattern of PV power plant distribution by reducing the construction costs of PV power stations to varying degrees.Despite the considerable variation in the construction of PV power stations across China’s coastal provinces and municipalities,the contribution of PV power generation to the country’s overall electricity consumption remains relatively small.Achieving carbon neutrality will require further efforts to increase this contribution.To this end,it is essential to allocate PV construction targets according to the specific conditions of each province and to take into account the various factors that affect power generation efficiency when building PV power stations,with a view to improving land use efficiency.This study presents a high-resolution(10 m)spatial distribution map of PV in China’s coastal provinces and municipalities,which can provide valuable data for assessing PV power generation efficiency,studying ecological and environmental impacts,and evaluating risks to food security.The accuracy of the map makes it a useful tool for policymakers and researchers alike,as they work to develop strategies for increasing the contribution of PV power generation to China’s energy mix and meeting its ambitious carbon neutrality targets.