Modelling air quality to tackle pollution hazards on Canary Islands and Peruvian Andes

Numerical methods can provide the solid evidence required to tackle air pollution hazards. Thereby, mathematicians and engineers altogether have embarked upon a quest toward most sophisticated descriptions of how prevailing winds propagate hazardous emissions. In that domain, EScGD is contributing with new methodologies. Thus, we have developed mathematical methods that are able to offer richer information that is nevertheless easier to compute. Our work has yielded two main outcomes: 1) a methodology to estimate wind patterns in high mountainous areas with a complex orography and, consequently, establish air pollution patterns by large emitters; and 2) a method to characterize the polluting effects of large number of components that feature complex reactions whilst being transported by air. The first method has been applied to determining the prevailing wind patters in La Oroya, in Andean Peru, whilst the second one has been tested, among other cases, in the Canary Islands.

Wind patterns in La Oroya were quite difficult to estimate. Two valleys, the first running from Southeast to Northwest, and the second from Southwest to Northeast, intersect in an area that hosts a town as well as a metallurgical facility that emits large quantities to the air. Wind patterns not only vary along the time of the day, but also across seasons. To further compound the problem, winds are also shaped to a large extent by additional meteorological phenomena such as ‘El Niño’ and ‘La Niña’. In order to cope with such extreme uncertainty, we developed a method based on the Principal Component Analysis of the consecutive sequences – each of them at least one day long – assembled out of 9,431 measurements from nearby weather stations. The method is particularly helpful to determine the prevailing winds during cold dawns and thermal inversion periods, which are common across complex orographies. 

Similarly, the Canary Islands also possesses a fairly complex orography. Consequently, we introduced a three-dimensional mesh built upon discretized domains to model the interactions between pollution and complex terrain across the island of La Palma. As a result, the model is less demanding on the user as well as lower computational costs. It is also able to determine the concentrations of all pollutants across the whole three-dimensional domain. Inspired by our model for La Palma, we applied an analogue tridimensional mesh to 24-hour wind patters in La Oroya, Peru. 

Our third development is an adaptive numerical method to simulate the distinct evolution of coupled pollutant plumes that is suited to address photochemical reactions in pollutant description. The method is premised upon splitting the simultaneous processes of transport and reaction. Under that premise, the problem becomes less complex, since the non-reactive part for each component and the reactions in each node can be dealt with separately. To do so, the partial differential equations required to solve the non-linear problem are manipulated by using adaptive schemes. Additionally, a multiple mesh – in contrast with conventional single-mesh approaches – is also better suited to compute vertical interpolations of the presence of pollutants. All in all, our model significantly curtails the challenges associated to the size of the mesh, thus allowing to estimate pollution hazards over larger areas. Furthermore, the computational requirements are also lower, even when comparing with single-mesh solutions.

Our three developments promise to help closing the gap between local models of air quality micro-models – in the range of a few meters – and meso-models developed for larger scales – in the range of kilometers. In consequence, our work is now focused upon envisaging fresh models to increase precision of results of interest to fulfill normative criteria as well as to increase the accuracy of risk analysis. Moreover, coupling with meso-scale prognosis models, will allow local models based on Finite Element methods to downscale predictions accounting more precisely for the local influences on the results, whilst also increasing accuracy.

In addition to supporting environmental impact assessments of new large emitters, more precise models, when combined with Geographical Information Systems, are in a privileged position to offer the authorities the tools they desperately need to fine-tune their actions against pollution hazards. Dynamic management of large emitters premised upon real-time monitoring, as is the case in La Oroya, offers a promising way to go.