Accuracy Assessment Of Global Human Settlement Layer(GHSL) Built-up Products And Urban Building Density Estimation

Building density map over large areas could provide essential information of land development intensity and human settlement environment.It is an important data set for supporting human settlement environment planning and studies.The Global Human Settlement Layer(GHSL)is a mapping data set of human settlement at global scale.It was developed by the Joint Research Centre(JRC),European Commission.The built-up area product is an important part of GHSL.Currently,the validation of the GHSL built-up area products was preliminarily conducted in the United States and European countries.However,as a typical East Asian region,China has significant architectural form and urban layout with these the United States,European countries.Therefore,it is necessary to perform accuracy assessment of GHSL in Asian countries like China.With individual building footprint data of 20 typical cities in China,this study presents our effort of validating the GHSL built-up area products.The aggregation mean and neighborhood search based algorithms are adopted for matching building footprint data and the GHSL products,through the regression analysis at perpixel level,the building density map in raster format are generated as validation data set.The accuracy index of GHSL built-up area was calculated for the study areas.The results show that built-up products aggregated by the building footprint have the best correlation with GHSL built-up products in the case of low resolution,but GHSL has certain error in the indicator of building density.This study suggest that GHSL builtup area products in typical cities in China can provide quantitative information about built-up areas,but the product accuracy still need to be improved in the regions with heterogeneous formations of human settlements like China.The presented study also provides a training dataset from high resolution images for generating products of built-up area density in China.Through the analysis of the validation results of the GHSL built-up areas products,the conclusion is that we need to produce a building density map products in China with improved accuracy.Therefore,this study carry out the development of urban building density estimation model based on google earth engine.Three representative cities in China,i.e.Beijing,Shanghai and Tianjin are selected as study sites.Explore the relationship between building density and urban development level and geographic location,and construct an urban building density estimation model for China.It mainly includes following aspects:1)Using the Google Earth Engine(GEE)to extract the spectral and texture feature information of all Landsat8 images in 2016,and perform pixel-by-pixel matching with the raster data obtained from the architectural contour vector data to construct Beijing,Shanghai and Tianjin respectively.Six training sets for the summer and winter of 2016,and extracting the single-temporal remote sensing feature information for the summer and winter of 2016 and 2018 as a test set.2)The Bayesian Ridge regression and XGBoost machine learning algorithm are used to establish the urban building density estimation model.The building density image data in the study area is obtained according to the test set,and compared the predicted results with the real values.The results show that:(1)The prediction results of building density for different feature combinations show that better model estimation results can be obtained by using the spectral feature information of the image.And the prediction results using each city’s unique training set are better than the single training set for all city predictions.(2)Compared with Bayesian Ridge regression and XGBoost machine learning algorithm,XGBoost has a better correlation between the results of urban building density estimation and true value.(3)Compared the summer and winter datasets,based on the two algorithm models,the summer training set prediction results are better than the winter training set prediction results.This study not only provides basic accuracy validation information for the application of GHSL building density products in China,but also lays a good foundation for the development of urban building density mapping products throughout China.