UAV-based PM₂.₅ monitoring system for small scale urban areas

In urban areas, air particle pollution is of precise interest because of its impact on health. Air quality data collection near the ground surface is difficult, particularly in small complex regions, and the usage of satellites image may not suffice and do not achieve the required accuracy. A variety of Unmanned Aerial Vehicles (UAVs) based on remote sensing technology enables data collection in these particular regions and overcoming obstacles and the difficulties obtaining required data. Remote sensing can be considered the best significant tool to assist in data monitoring for estimating and predicting air quality parameters. The recent monitoring stations are fixed stations and are not designed to denote exposure on a small scale adequate. Most of the studies rely on satellite observations from Aerosol Optical Depth (AOD) and have used lower resolution AOD to estimate PM2.5 levels. In general, this used resolution of AOD products is often insufficient to define exposure estimations in urban areas. In this manner evaluation at different altitudes can offer extra information to assess air quality. The research aims to introduce a PM2.5 prediction algorithm based on PM2.5 measurements from a developed a system capable of measuring PM2.5 concentrations in small-scale areas and validate the model at specified low altitudes. Observations based on UAV-based PM2.5 monitoring sensors were applied around 1.6 km² area for collecting data at low altitude. Meteorological parameters including temperature and humidity were collected. This study uses an empirical method via applying amassed records of PM2.5 concentrations and meteorological par ameters to create a geographically weighted regression (GWR) model to estimate PM2.5 concentrations in a small-scale area. For the predicted model, an accuracy value is computed from the probability value given by the regression analysis model of each parameter. To validate our method, we have utilized two types of data, training, and testing. To evaluate and validate the suggested GWR model, we applied the model using testing measured points. Results showed a relatively good fit of the model to the observed data. Where the maximum accuracy obtained was set as 65% in July and 73% in August. Also, the obtained results showed that there is a good statistical correlation between the measured in situ data and testing data, the maximum accuracy was set as 93% in July and 94% in August. The developed tool can be considered as an independent method for sample collection demonstrated that the characteristics obtained by analysis are able to monitor and predict the concentrations of PM2.5 in small-scale areas with high accuracy. This suggested approach is useful to cover the area within a short amount of time, with low cost and limitless flexibility.

Preview Text FK 2018 152 - IR.pdf Download (660kB) | Preview

Actions (login required)

View Item