A Fine-grained Particular Matter Concentration Distribution Based On Crowd Sensing

The frequent hazy weather has brought enormous threaten to human health,especially in developing countries.Urbanization has brought a great part of social activities taking place in buildings rather than on the ground.Modeling a fine-grained 3D distribution of PM2.5(Particulate matter with diameters less than 2.5?m)concentrations is of great importance for government and citizens.For on-ground,the contradiction between the need of fine-grained distribution and the approach of coarse-grained collection has to be addressed immediately.Meanwhile,for the near-surface layer,there is no monitoring system to study PM2.5 concentration with acceptable coverage and granularity.To this end,this paper presents BlueAer,the first three-dimensional(3D)spatial-temporal fine particulate matter(PM2.5)monitoring systems which is designed to understand urban PM2.5 concentration distribution in a fine-grained level.(1)For cost-efficient data collection,inspired by crowding sensing,we propose a mobile collected strategy based on crowding sensing.Vast amount of 3D samples are collected by limited mobile vehicles with build-in low-cost sensors.Constrained by existed road network,we design suitable collective routes for testing cars based on path programming.We also employ a drone with a build-in sensor to acquire PM2.5 concentration in near surface layer.(2)The consistency of the data processing is responsible for dealing with chaotic data and deduces the new information which can represent the regional characteristic.It mainly includes improving the quality of original data,defining suitable spatial and temporal resolution and spatial data fusion.According to the continuity and region of sensor data,a preprocess algorithm is designed to improve its quality by filtering.On the basis of a large number of experimental statistics,the reasonable 3D resolution is determined.Data fusion is employed to calculate PM2.5 concentration with high quality for each spatial grid in the use of redundant data under the certain resolution.(3)For undetected area,the fine particle transport is simulated in a probabilistic manner rather than based on an amount of specific influent factors such as traffic volumes,street geometry,temperature,wind direction and the emission volumes for each vehicles.If those probabilities can be calculated,then the undetected concentration can be deduced.Based on the main idea and inspired by random walk,a 3D probabilistic concentration estimated method(3D-PCEM)is proposed to infer PM2.5 concentration.The experimental result shows that the inference accuracy of 3D-PCEM is enhanced by 15.4%and 41.0%.comparing to Gaussian Process(GP)and Artificial Network(ANN)respectively.(4)Sensor selection,calibration,installing,data transition and visualization in terminal equipment are all considered in this paper when we developing the prototype system of BlueAer.It has been implemented and worked throughout a year in a 64km2 testing area with a population of 400,000 in Hangzhou,China.Experimental data has verified that BlueAer can achieve good performance in terms of stability as well as a fine grained 3D distribution of PM2.5 concentration.BlueAer can easily be extended for a larger scale and applied city-wise.Besides,our findings can help ordinary citizens better understand their immediate air quality and serve as a framework towards detailed national-wise real-time pollution management.