Abstract
China has experienced successive waves of dengue epidemics over the past decade. Nationwide data on 95,339 dengue cases, 89 surveillance sites for mosquito density and population mobility between 337 cities during 2013-20 were extracted. Weekly dengue time series including time trends and harmonic terms were fitted using seasonal regression models, and the amplitude and peak timing of the annual and semiannual cycles were estimated. A data-driven model-inference approach was used to simulate the epidemic at city-scale and estimate time-evolving epidemiological parameters. We found that the geographical distribution of dengue cases was expanding, and the main imported areas as well as external sources of imported cases changed. Dengue cases were predominantly concentrated in southern China and it exhibited an annual peak of activity, typically peaking in September. The annual amplitude of dengue epidemic varied with latitude (F = 19.62, P = 0.0001), mainly characterizing by large in southern cities and small in northern cities. The effective reproduction number R(eff) across cities is commonly greater than 1 in several specific months from July to November, further confirming the seasonal fluctuations and spatial heterogeneity of dengue epidemics. The results of this national study help to better informing interventions for future dengue epidemics in China.