Global High-resolution Impervious Surface Extraction From Sentinel 1/2 Data

Accurate and timely impervious surface mapping is fundamental to support global environmental and socio-economic research.Most of the available global impervious surface products are limited to a spatial resolution of 30m.The fusion of optical and SAR data for global high-resolution impervious surface mapping is not yet been widely explored.In this study,we proposed an effective impervious surface extraction method using ascending/descending orbits of Sentinel-1A SAR data and Sentinel-2 MSI(MultiSpectral Instrument,Level 1C)optical data acquired in 2015.Finally,we produced global impervious surface product(10m)based on backscattering coefficient characteristics of SAR and optical phenological characteristics.Main contents as follows:(1)Potential impervious surface(PIS)was identified first through logical operations on yearly mean and standard deviation composites from time series of ascending/descending orbits of SAR data.Yearly Normalized Difference Vegetation Index(NDVI)maximum and modified Normalized Difference Water Index(MNDWI)mean composite were generated from Sentinel-2 imagery.The slope image derived from SRTM DEM data was used to mask mountain pixels and reduce the false positives in SAR data over these regions.We applied a region-specific threshold on PIS to extract the target impervious surface(TIS)and a global threshold on the MNDWI mean,and slope image to extract water bodies and high-slope regions.We took 29 megacities around the world as test areas and collected samples from these cities,and we estimate the parameter threshold to extract the impervious surface of the target city.These megacities were chosen as the testing region to validate the accuracy and robustness of our proposed method through validation points randomly selected.from high-resolution Google Earth imagery.Additionally,a total of 120 blocks with a size of 900 X 900m were randomly selected and used to compare our product's accuracy with the global human settlement layer(GHSL,2014),GlobeLand30(2010),and Liu(2015)products.Our method demonstrated the effectiveness of using a fusion of optical and SAR data for large area urban land extraction especially in areas where optical data fail to distinguish urban land from spectrally similar objects.Results show that the average overall and Kappa accuracies are 93.07%and 0.863,respectively.(2)Given the complexity of the distribution of features in large-scale extraction,it is difficult to meet the requirements of rapid production in accordance with the method of selecting empirical thresholds one by one.Therefore,based on the impervious surface extraction method of megacities,this paper optimizes the threshold selection method,and China was chosen as the test area to extract the impervious surface,and the main processing was carried out on the Google Earth Engine(GEE)platform.We set thresholds based on terrain and ecological stratification,then a majority filter with a 3 by 3 window was applied on previously extracted results to remove the 'salt-and-pepper noise'.According to the above verification method,a total of 735 blocks with a size of 900 × 900m were randomly selected and used to compare our product's accuracy with other products.Our method demonstrated the effectiveness of using a fusion of optical and SAR data for large area urban land extraction especially in areas where optical data fail to distinguish urban land from spectrally similar objects.Results show that the average overall,producer's and user's accuracies are 88.03%,94.50%and 82.22%,respectively.This provides technical support for the global impervious extraction.(3)This study divided the global ecological layer and established the "centroid point" of each ecological layer as the threshold reference of the layer.We designed a program based on the GEE platform for threshold optimization and impervious surface extraction.The grid with the latitude and longitude of 10°×10° was used as the production unit.The production unit extracts the impervious surface of each ecological layer and outputs the result according to the threshold selection principle.We conducted quality checks and assessments of the final results.Results show that the average overall,producer's and user's accuracies are 87.91%,82.83%and 95.03%,respectively,which satisfied the high-resolution global impervious surface production demand and provided products for relevant researchers.