Spatial-temporal Pattern Evolution Of Impervious Surface And Thermal Environment Effect In Jinan City

The growth of impervious surface caused by urbanization is an important indicator for exploring urban expansion and measuring ecological environment.The rapid and accurate identification of impervious surface distribution information is important for urban management and monitoring,regional planning and sustainable development.In recent years,Jinan is in the stage of rapid urban development,and the impervious surface in the built-up area is rapidly increasing,resulting in significant ecological and environmental problems such as heat island effect.Therefore,this thesis develops an efficient and accurate impervious surface mapping method based on multi-source remote sensing data,deep learning and image segmentation.This method is used to monitor the spatial and temporal changes of impervious surfaces and the thermal environmental effects of impervious surfaces,aiming to provide a reference for the fine urban planning and management of Jinan city.Due to the complexity of the urban landscape and the variety of types and materials of impervious surfaces,the precise extraction of impervious surfaces remains a challenge.Deep learning has the advantage of automatic feature extraction,thus we introduce the semantic segmentation method based on deep learning to improve the extraction accuracy of impervious surface information.So as to realize the extraction of impervious surface distribution data with10m resolution in central city of Jinan from 2015 to 2021.Finally,it passed the accuracy verification and the accuracy check of details.On the basis of obtaining year-by-year impervious surface distribution data,the spatial and temporal evolution characteristics are analyzed.In addition,the thermal environmental effects generated by impervious surfaces are also quantitatively analyzed.The main contents and conclusions are as follows.（1）The impervious surface is labeled based on Sentinel-2 images,and the impervious surface sample set is created after data enhancement,data cropping.The impervious surface extraction model is constructed using Deeplabv3+model combined with attention mechanism.Meanwhile,in order to improve the problem of imbalance between positive and negative sample and difficult samples of impervious surface,the focal loss function is used for the calculation of model loss.The experimental results show that the Precision,Recall,Io U and F1 values of the impervious surface information extraction model with improved Deeplabv3+network reach 91.82%,86.22%,80.07%and 0.89,respectively.The high extraction accuracy and good application results prove the effectiveness and feasibility of this model.The validity and feasibility of the model can realize the dynamic monitoring of impermeable surface with high resolution over large scale and long time series.（2）The model was applied on the 2015-2021 images of the central city of Jinan to achieve efficient and accurate extraction of time-series impervious surfaces.The impervious surface area fluctuates and increases from 2015 to 2021,and the impervious surface area increases rapidly to 496.87 km~2 in 2021,with an area ratio of 48.93%.The distribution of impervious surface has obvious directionality,with a northeast-southwest direction.Both the long and short axes of the standard deviation ellipse fluctuate and grow,indicating that the expansion direction is mainly east-west,and the expansion in the north-south direction is also increasing.The results of the landscape pattern index show that the number of impervious surface patches decreases and the maximum patch index increases,indicating that the impervious surface increases in a filling pattern,and small areas are connected into large impervious surface patches,which are gradually connected from a broken state,and the degree of connectivity and aggregation increases.（3）There was a significant correlation between impervious surface and surface temperature.The distribution of surface temperature has a high consistency with the spatial distribution of impervious surface.The abundance of impervious surface was significantly and positively correlated with the surface temperature;the impervious surface patch density,landscape shape index,and aggregation index all had a high correlation with the surface temperature.Reducing the density of impervious surface,lowering the aggregation degree and irregular shape of impervious surface are all beneficial to reduce the surface temperature.Therefore,reasonable planning of impervious surface construction has scientific and practical significance for alleviating urban heat island effect.