Abstract: Based on the discrete in situ PM2.5 observed data, AOD data, as well as the meteorological data, in this study the temporal-spatial distribution of PM2.5 concentration at monthly scale in fifteen key tourism cities of China for years 2013, 2014, 2015 was firstly simulated by using the land use regression (LUR) method. Then, the seasonal and annual PM2.5 concentrations across the entire study period were consequently produced based on these monthly PM2.5 distribution maps with the aid of spatial overlay and raster calculation techniques in ArcGIS tool. Finally, the spatial-temporal heterogeneity of PM2.5 concentrations at scales of monthly, seasonal and annual in each key tourism city were statistically analyzed. Generally, the annual average PM2.5 concentrations of the fifteen key tourism cities showed a gradually declining trend from 2013 to 2015 with significantly higher values in winter and lower values in summer. Spring and autumn were usually with annual average PM2.5 concentrations between those of winter and summer. Meanwhile, the declining trend of PM2.5 concentrations could also be clearly observed at seasonal and monthly scale except a little abnormal increase in the summer of 2014. Moreover, comparisons of the PM2.5 concentrations at the three time scales among the fifteen key tourism cities in this study revealed that the deputy provincial cities were usually with higher PM2.5 concentrations, while the prefecture level cities were with the lowest values, and the values for municipalities were in the middle. Additionally, as for the suitable time for traveling in terms of air quality , the months from April to November were the good time window for travel for the fifteen tourism cities, especially during the period from July to September. However, the suitable travel time window obviously increased in spring and winter of 2015 with the air quality improvement. Spatially, the deputy provincial cities and prefecture-level cities located in southeast coastal areas and high forest coverage rate areas usually had longer suitable travel time, such as Kunming, Sanya, Xiamen and Guangzhou. Inversely, the suitable travel time was relatively short in the central and western regions as well as the Yangtze River Delta region such as Wuhan, Chengdu and Shanghai. In sum, as the first quantitative analysis of the suitable travel time of tourism city in China with spatial mapping method, this study would provide helpful suggestions for tourist decision in choosing the suitable travel time for specific tourism city. And this would be a significant extension of current research fields of urban tourism, and would consequently promote researches on the influences of inner driving mechanism of haze pollution on urban tourism, as well as on the tourism product structure and tourism operation.

Key words: tourism city, PM2.5, haze, tourism travel, land use regression method