Critical Review of Methods to Estimate PM2.5 Concentrations within Specified Research Region

Abstract

Obtaining PM2.5 data for the entirety of a research region underlies the study of the relationship between PM2.5 and human spatiotemporal activity. A professional sampler with a filter membrane is used to measure accurate values of PM2.5 at single points in space. However, there are numerous PM2.5 sampling and monitoring facilities that rely on data from only representative points, and which cannot measure the data for the whole region of research interest. This provides the motivation for researching the methods of estimation of particulate matter in areas having fewer monitors at a special scale, an approach now attracting considerable academic interest. The aim of this study is to (1) reclassify and particularize the most frequently used approaches for estimating the PM2.5 concentrations covering an entire research region; (2) list improvements to and integrations of traditional methods and their applications; and (3) compare existing approaches to PM2.5 estimation on the basis of accuracy and applicability.

A Framework to Predict High-Resolution Spatiotemporal PM2.5 Distributions Using a Deep-Learning Model: A Case Study of Shijiazhuang, China

“…However, we can only obtain the accurate ground PM 2.5 values from the sparsely populated ground air quality station sites. If we want to get the spatially continuous distributions of PM 2.5 , we need to estimate the concentration based on the AOD and other key factors [13]. Thus, the purpose of this part is to build a regression model according to the ground truth PM 2.5 data based upon the air quality station site points, and then, the model is used to estimate the whole distribution in the whole city area.…”

Large-Scale, Fine-Grained, Spatial, and Temporal Analysis, and Prediction of Mobile Phone Users’ Distributions Based upon a Convolution Long Short-Term Model

“…In terms of population mapping, most previous studies concentrated on using simple areal-weighting methods (a technique for estimating the values of overlapping but incongruent polygon features) [29,30,31,32] and estimating the population data from the census data with respect to the census administrator units or regions [33]. Others used some ancillary data such as remote-sensing images, land-use data, e.g., urban or sub-urban boundaries, to estimate population counts within the census units.…”

“…Therefore, in this study, we use the mobile cell phone data which can directly reflect the phone users’ distribution to characterize people’s mobility based on the ubiquitous nature of mobile phones in modern society [14]. As machine learning and deep learning techniques are currently popular to predict or estimate output values in many fields [29], in this study, a deep learning approach is also proposed for estimating and predicting mobile users’ dynamic distribution that potentially contributes to mapping and predicting a population’s dynamics at a more fine-grained spatial scale and temporal scale, i.e., every two hours, compared to two traditional methods which can be used as a baseline, i.e., the autoregressive moving average (ARMA) model and a popular deep learning method that uses a long short-term memory (LSTM) model for predicting time-series data.…”

PM2.5 Pollutant in Asia—A Comparison of Metropolis Cities in Indonesia and Taiwan

“…However, it is a big challenge to access PM2.5 data, particularly in developing countries. During the last period, some methods have been established to tackle challenges related to air pollution, such as interpolation using kriging and IDW (Inverse Distance Weighting) [6,7], and LUR (Land Use Regression) [8][9][10]. The interpolation of pollutant concentrations is based on the monitoring sites with densely clustered stations, whereas it is difficult to monitor locations with few stations.…”