By Guofeng, December 19, 2017
A new paper is published Improve ground-level PM 2.5 concentration mapping using a random forests-based geostatistical approach in Environmental Pollution. In this paper, we produced a gridded PM2.5 concentration datasets with 1 km spatial resolution by combing heterogeneous data sources via machine learning-based geostatistical methods. Using the monitoring measurements as a benchmark, our results outperform the most commonly used datasets. The resultant datasets can be downloaded at our GitHub repository.