Year

2019

Degree Name

BEnviSci Hons

Department

School of Earth, Atmospheric & Life Sciences

Advisor(s)

Stephen Wilson

Abstract

Radon-222 is increasingly being used in air quality models as a tracer for understanding atmospheric dynamics and transport to improve the simulation of other key trace species such as CO2, CH4, NOx and O3. Applications such as this have driven efforts to improve the accuracy of radon emissions (also referred to as fluxes) in these models. Soil moisture is one of the primary factors which drives variations in radon emissions, and hence the concentration of radon in the atmosphere, however our understanding of the degree of its influence is currently lacking. This study aimed to gain a greater understanding of the influence of soil moisture on radon-222 emissions in the Sydney Basin over the period from January to December 2016. The focus of these efforts was aimed at determining the extent to which complexity of modelled soil moisture data influences the accuracy of modelled radon concentrations. This was done by comparing modelled radon concentrations derived from 4 distinct radon emissions maps with increasing levels of soil moisture complexity across varied climatic conditions and in response to a major rainfall event.

All modelled radon concentrations differed from observed values, with the best agreement shown by the most complex ‘daily’ emissions scenario with a correlation co-efficient of 0.68 for the whole year. In all instances, the two time-dependent, soil moisture driven scenarios performed very similarly to each other and exhibited greater precision than the two non-time-dependent scenarios, despite underestimating concentrations. A distinct diurnal cycle was observed for all scenarios, with peaks in night-time radon concentrations being poorly reflected by all of emissions scenarios, likely due to poor simulation of the nocturnal boundary layer. Following rainfall events all modelled radon concentrations more accurately reflected these observed night-time values. Both time-dependent scenarios exhibited decreased in emissions following rainfall, however the normalised mean bias remained relatively consistent throughout. This reflects the effective response of these more complex soil moisture driven, time-dependent scenarios, despite their underestimation of concentrations.

It is clear that implementing emissions estimates based on complex time-varying soil moisture inputs provides some improvement over more simple non-time varying methods. It is likely that further increasing complexity of the soil moisture models would provide little improvement for modelled radon concentrations. Rather improvements in the model itself, particularly its estimation of the nocturnal boundary layer, would assist in improving the accuracy of modelled radon concentrations. Thus, making radon a powerful diagnostic of model mixing and transport, in spite of the temporal variations in surface emissions.

Share

COinS
 

Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.