Construction of Ultraviolet Radiation Fitting Model and Analysis of Correlation Factors in Guangzhou Based on Gradient Boosting Decision Tree
Based on data from ultraviolet radiation observations,traditional meteorological observations,and ambient air quality observations in Guangzhou from 2019 to 2021,we analyse the variation characteristics of ultraviolet radiant intensity in Guangzhou,and the correlations between ultraviolet radiation intensity and meteorological factors as well as ambient air quality factors.We select seven meteorological and ambient air quality factors significantly linked to the intensity of ultraviolet radiation in Guangzhou as characteristic elements for the input ultraviolet radiation intensity fitting model.We further use the Gradient Boosting Decision Tree(GBDT)algorithm to establish a fitting model for the city's ultraviolet radiation intensity.The results demonstrate that:(1)The ultraviolet radiant intensity in Guangzhou has obvious seasonal and daily variation characteristics.The seasonal variation of ultraviolet radiation intensity in Guangzhou is characterized by high levels in summer and autumn,and low levels in winter and spring.In 2020 and 2021,the maximum value of ultraviolet radiation intensity in Guangzhou occurs in July,while in 2019,the maximum value of ultraviolet radiation intensity in Guangzhou appeared in September.The minimum ultraviolet radiation intensity in Guangzhou between 2019 and 2021 occurred in March,with the lowest intensity 15.9 W·m-2in 2020.Daily variations in ultraviolet radiation intensity in Guangzhou typically show low in the morning and evening levels and a high at noon level,with ultraviolet radiation intensity reaching its daily maximum around 12 o'clock.(2)The meteorological and ambient air quality factors significantly correlated with the intensity of ultraviolet radiation in Guangzhou are air temperature,visibility,total cloud cover,relative humidity,solar altitude angle,ozone concentration,and nitrogen dioxide concentration.(3)The ultraviolet radiation intensity model in Guangzhou has a good fitting effect.The coefficient of determination for the training data set and the test data set of the ultraviolet radiation intensity fitting model in Guangzhou are 0.93 and 0.80,respectively.Moreover,the RMSE of the training and testing set data for the ultraviolet radiation intensity model in Guangzhou is 2.7 and 4.9 W·m-2 respectively.The estimate accuracy of ultraviolet radiation intensity levels in Guangzhou is 75％,one level difference in ultraviolet radiant levels accounts for 21％,and two level difference in ultraviolet radiant levels accounts for 4％.
ultraviolet radiation intensity characteristicultraviolet radiation fitting modelgradient boosting decision tree
NETL
NSTL
万方数据
