Please use this identifier to cite or link to this item:
https://www.um.edu.mt/library/oar/handle/123456789/150272| Title: | Investigating the factors affecting the PM2.5/PM10 ratio with a view to improving PM10 measurements from particle deposition data |
| Authors: | Farrugia, Nicola (2026) |
| Keywords: | Air -- Pollution -- Malta Air -- Pollution -- Measurement Particulate matter -- Malta Environmental monitoring -- Malta Regression analysis |
| Issue Date: | 2026 |
| Citation: | Farrugia, N. (2026). Investigating the factors affecting the PM2.5/PM10 ratio with a view to improving PM10 measurements from particle deposition data (Bachelor's dissertation). |
| Abstract: | Monitoring of Airborne Particulate Matter (APM) in Malta is carried out at five fixed gravimetric stations, as is mandated by EU Directive 2024/288. Financial and logistical constraints limit expanding this network, prompting the use of low-cost passive deposition samplers, like the Sigma-2, for high-resolution spatial data. These samplers estimate PM10 concentrations by scaling naturally deposited coarse fractions using a PM2.5/PM10 ratio derived from nearby reference stations as a scaling factor. Consequently, the accuracy of this method depends on the predictive precision of the scale factor. This study evaluates how Saharan dust events, seasonality, and station type (spatial variability) influence the PM2.5/PM10 ratio, aiming to optimize scale factor determination and improve dry deposition measurement accuracy. Using daily PM10 and PM2.5 data from the Environment and Resources Authority (ERA), group comparisons and permutation-based ANOVA testing were carried out to quantify the impacts of these factors on particulate concentrations and their ratio. All factors tested were found to produce significant effects on the ratio, although station type was found to be the primary driver of variance. However, the selection of the optimal scaling approach is based on predictive performance rather than variance explanation alone. To identify the optimal regression technique to be utilised when determining the ratio, the predictive accuracies of different scale factors when applied to a test subset were evaluated by comparing scaled estimates to observed values. The scale factors utilised were equivalent to the gradient values obtained when a training dataset was fit to the following regression models: OLS, Theil-Sen, orthogonal and quantile. When applied to the full dataset, performance differences between the models were found to be marginal, however when applied to station-specific subsets Theil-Sen consistently produced the lowest error metrics. Thus, the incorporation of station-specific data and the application of robust regression techniques, such as Theil–Sen, is expected to improve the accuracy of PM₁₀ estimates derived from dry deposition methods. |
| Description: | B.Sc. (Hons)(Melit.) |
| URI: | https://www.um.edu.mt/library/oar/handle/123456789/150272 |
| Appears in Collections: | Dissertations - InsESEMP - 2026 |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 2608IESEMP302500015538_1.PDF Restricted Access | 2.17 MB | Adobe PDF | View/Open Request a copy |
Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.
