Downscaling Research And Application Of NPP/VIIRS Nighttime Light Remote Sensing Data

Because of its ability to visualise differences in human activity at night,Nighttime Light(NTL)remote sensing imagery has been successfully used on a global and regional scale in a variety of applications,including demographic,political,economic,and environmental protection areas.Among the vast amount of nighttime light remote sensing images,the Visible Infrared Imaging Radiometer Suit(VIIRS)on board the Suomi NPP satellite provides nighttime light remote sensing images(Day/Night Band,DNB band)as one of the important data sources for studying human activities.However,the spatial resolution of NPP/VIIRS NTL imagery is no longer sufficient to meet the needs of more detailed urban studies.In this paper,a downscaling study of NPP/VIIRS DNB nighttime light remote sensing image data based on multi-source spatial variables and MGWR(Mulite-scale geographic weighted regression)method is carried out in order to provide better quality data products for the subsequent study of nighttime light.Firstly,a statistical regression model was used to construct a spatially non-stationary multi-scale relationship between NPP/VIIRS DNB data and the auxiliary variables at a low spatial resolution(500 m)scale;secondly,it was used at a high spatial resolution(120 m)scale based on the assumption of"constant relationship scale",so as to achieve Secondly,it was used at a high spatial resolution(120 m)scale based on the assumption of"constant relational scale"to achieve spatial downscaling of NPP/VIIRS DNB nighttime light remote sensing data.In this study,several accuracy evaluation metrics are used and the nighttime light remote sensing image LuoJia-01 is used as the actual reference to validate the MGWR-based nighttime light downscaling model;meanwhile,the RF(Random Forest)algorithm and GWR(Geographic weighted regression)algorithm are chosen to further compare quantitatively based on the same impact factors.regression)algorithm to evaluate the downscaling accuracy of the MGWR model.Finally,this study extracted urban built-up areas using the downscaled NTL data,and verified the value of the new data in urban studies by comparing it with the high-precision urban built-up area data.Through the downscaling study of the NPP/VIIRS data,the important findings of this paper are as follows.(1)The multivariate-based MGWR method better describes the spatial variation of NTL than the GWR and RF methods.The downscaled NTL images showed higher data resolution than the original NTL in terms of more detailed information,sharper boundaries.(2)Due to the pronounced spatial heterogeneity scale differences of various influencing factors on the distribution of NTLs,it is difficult for global statistical regression and classical GWR models to reveal the spatial heterogeneity scale effects between the NTLs and the auxiliary variables.The proposed MGWR downscaling model improves on the classical GWR by allowing individual auxiliary variables to have different bandwidth set-tings according to the range of influence scales.Therefore,it can provide a more realistic and effective description of the spatial process,better explaining the effects of different auxiliary variables on the spatial variation of NTL.The results show that the downscaled NTL data quality is significantly improved,with the MGWR algorithm having a higherR~2(0.91 vs.0.90 vs.0.87)and lower RMSE(29.00 vs.17.57 vs.16.87 n W/cm~2/sr)compared to the GWR and RF algorithms.This indicates an improvement in data quality after downscaling and confirms the feasibility and applicability of the MGWR algorithm in performing spatial downscaling of nighttime lights.(3)The MGWR proposed in this paper mainly uses eight auxiliary variables for NDVI,NDBI,road density,POI density,LUCC,LST,and the latitude and longitude,which strongly correlate with the NTL.For the spatial downscaling of the NTLs,the construction of the relational model and the selection of the auxiliary variables are directly related to whether accurate and reliable high-spatial-resolution NTL data can be obtained.By capturing the different auxiliary variables of various auxiliary variables on the NTL distribution,the MGWR avoids the introduction of excessive noise and bias in the process of constructing NTL conversion function and provides technical support for the accurate realization of NTL space downscaling.(4)The downscaled NPP/VIIRS NTL is applied to the two urban built-up area extraction methods described in this paper,and its accuracy is higher than that of the original NPP/VIIRS data,which not only improves the overall accuracy of the extraction but also shows richer spatial details,effectively compensating for the lack of spatial resolution of NPP/VIIRS.The NTL spatial downscaling method proposed in this paper,based on the MGWR model,can improve the texture information of low spatial resolution NTL data and guarantee the accuracy and spatial consistency of NTL downscaling results.Our research is very meaningful for improving the data quality of NPP/VIIRS NTL and can provide reliable NTL datasets with high spatial and temporal resolution.