Study On The Scale Of Urban Spatial Feature Extraction Based On Nighttime Light Data

Since the nighttime light data was opened to the outside world,there have been many researches and applications based on data,and the spatial scale of data application is mostly concentrated at the national level,provincial level and municipal level,which indicates the social and economic conditions and the intensity and scope of human activities,while the research on the application of suitable spatial scale for the quantification of night light data of the system is still lacking.Therefore,this paper applies urban agglomerations and cities with different spatial scales in urban agglomerations to analyze the appropriate spatial scale of NPP/VIIRS data application.Firstly,this paper classifies the cities in the study area according to the standard of urban scale division in China,and then combines the ranking of urban commercial charm;Then,the natural breakpoint-mutation detection method is used to extract the urban built-up area,and the mutation detection method is improved to improve the speed of determining the optimal threshold;The accuracy of the extracted built-up area is verified according to the data of 30 meters impermeable water surface;Finally,using the extracted urban spatial features,the suitable scale of NPP/VIIRS data urban spatial feature extraction is studied.The study found that:(1)This paper studies the scale of night light data in urban information extraction and application by observing the extraction ability of different levels of urban spatial features on night light data.The NPP/VIIRS data can better indicate the urban built-up areas in cities with built-up areas larger than 80 square kilometers.The extraction accuracy is different in different cities,and the lower the city level,the worse the extraction effect.When NPP/VIIRS night lighting data is applied to urban agglomerations,first-tier cities,new first-tier cities and second-tier cities,the overall extraction accuracy of urban built-up areas is over 90%;In third-tier cities and fourth-tier cities,the overall extraction accuracy is between 80% and 90%.In fourth-tier cities,the extraction accuracy is most affected by the urban built-up area.The larger the built-up area,the higher the accuracy.In the fifth-tier cities,the extraction accuracy of built-up areas fluctuates greatly,and the extraction accuracy of built-up areas is between 20%and 90%.(2)In the scale analysis of NPP/VIIRS data in urban spatial feature extraction,it is found that the extraction accuracy of urban built-up areas is different at different scales,and the lower the city level,the worse the extraction effect;Compared with cities of the same scale,it is greatly influenced by area and slightly influenced by shape.The extraction accuracy of NPP/VIIRS night lighting data in the fifth-tier cities with built-up area less than 80 square kilometers is generally low and the error is large.Therefore,this paper thinks that NPP/VIIRS night lighting data is not suitable for the fifth-tier cities with built-up area less than 80 square kilometers.(3)In this paper,the method of extracting urban built-up area based on catastrophe detection method is improved.Firstly,the light value is reduced by dichotomy natural breakpoint method,and the minimum value iterated by natural breakpoint method is directly divided into non-built-up areas.Then,the light value larger than the minimum value is used to divide the threshold from small to large based on catastrophe detection method,to find catastrophe points and realize the extraction of built-up area.This method improves the recognition speed of mutation point by mutation detection method,simplifies the steps,and reduces the light threshold.