Air temperature and relative humidity are the key variables in environmental health research. Both of them are difficult to map especially at national scale because of spatial heterogeneity. This paper presents a methodology for mapping relative humidity, average and extreme temperature in hot summer (June to August) over China. Several data as explanatory variables were applied to random forest regression models to predict relative humidity and temperatures, including surface reflectance, land cover, digital elevation model (DEM), enhanced vegetation index (EVI), latitude, nighttime lights (NLs), as well as buffer zones of road, railroad, river system and administration center. Results based on cross-validation reflect acceptable prediction errors in estimating relative humidity (RMSE=7.4%), average temperature (RMSE=2.4°C), average maximum temperature (RMSE=2.5°C), and extreme maximum temperature (RMSE=2.6°C). Despite the strong correlation between average and extreme temperatures, significant differences exist in their spatial distribution along the latitude direction, especially in the areas such as Hebei, Szechwan, Hubei, Henan, Shandong, and Inner Mongolia. Specifically, social economic activity, relative humidity and vegetation tend to affect extreme heat events, and both latitude and DEM (i.e., geographical position) determine the average level of temperature. Compared with interpolation technology and statistical methods, the proposed methodology demonstrates the ability to generate relative humidity and temperature maps with finer gradients in hot summer over China.