Receptor Modelling using PMF2: influence on source fingerprints and their contributions using different input options and measurement uncertainties
Loading...
Date
2012-11-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Australian Nuclear Science and Technology Organisation
Abstract
Receptor modelling is used to determine source fingerprints, from measurements at the receptor site, and then apportion the measured mass to the identified possible sources. Mathematical techniques used to solve this problem include Principal Component Analysis and multivariate techniques such as Positive Matrix Factorisation and UNMIX.
In this report PMF analysis is carried out using the PMF2 program. When using PMF2 a number of options control the optimisation process and by varying the options the final solution might be impacted. In addition the uncertainties associated with the measurement have an impact on the solution. Here we systematically vary the PMF2 options and the specified uncertainties of each measurement and compare each solution to a base case solution which has recently been published for the selected dataset.
The main findings were that the uncertainties in the measurements and the number of factors chosen to be recovered were the key quantities for the PMF analysis of the considered dataset.
Description
Keywords
Receptors, Analytical solution, Factorization, Optimization, Mathematical models, Air pollution, Aerosol monitoring
Citation
Crawford, J., & Cohen, D. D. (2012) Receptor Modelling using PMF2: influence on source fingerprints and their contributions using different input options and measurement uncertainties (ANSTO/E-776). Lucas Heights, NSW: Australian Nuclear Science and Technology Organisation.