关键词: 花粉过敏, 百度指数, 城市群, 夜间灯光, 光合有效辐射吸收分量, 机器学习模型

Abstract: Studying the spatio-temporal characteristics and influencing factors of pollen allergy network attention is beneficial for mastering relevant information and serving people's livelihoods. This paper combines Baidu index,meteorological,and remote sensing data to analyze the spatiotemporal characteristics and influencing factors of pollen allergy network attention in eight urban agglomerations from 2017 to 2021. Random forest and back propagation neural network models are used for simulation. The results show that: ①In terms of spatiotemporal characteristics,the annual peak of attention is in spring (April and May). Cities with concentrated populations receive the highest attention. The city is spatially concentrated,with the Beijing-Tianjin-Hebei,Shandong Peninsula,Yangtze River Delta,and Pearl River Delta urban agglomerations as hot spots. ②In terms of relationship with influencing factors,within the effective range,with the increase of temperature,short term thunderstorm rainstorm,poor air quality,increased fraction of absorbed photosynthetically active radiation,and strong nighttime light,attention increases,and with high humidity,high wind speed,heavy rain,and prolonged rain,attention decreases. Attention has the highest correlation with nighttime light.③In terms of urban agglomerations,the temperature of the northern urban agglomerations,the humidity of the coastal urban agglomerations,the wind speed of the urban agglomerations with large undulating terrain,the precipitation of the southern urban agglomerations,the air quality index of the Pearl River Delta and the Yangtze River Delta and the fraction of absorbed photosynthetically active radiation of the Pearl River Delta and the Beijing-Tianjin-Hebei urban agglomerations are of great importance,and the nighttime light in each urban agglomeration is irreplaceable. ④Random forest and back propagation neural network are suitable for simulating attention,with R² ranging from 0.64 to 0.92 for each urban agglomeration,and RMSE and MAE below 1. Back propagation neural network has better simulation effects than random forest. The fit of both models in the Chengdu-Chongqing urban agglomeration is excellent,followed by the Beijing-Tianjin-Hebei,Pearl River Delta,and the middle reaches of the Yangtze River urban agglomeration. The methods and results of this paper can provide reference for work related to pollen allergy.

Key words: pollen allergy, Baidu index, urban agglomerations, nighttime light, fraction of absorbed photosynthetically active radiation, machine learning models