Abstract
The deterministic approach of sensitivity analysis is applied on the solution vector of an Air Quality Model. In particular, the photochemical CAMx code is augmented with derivatives utilising the automatic differentiation software ADIFOR. The enhanced with derivatives version of the model is then adopted in a study of the effect of perturbations at the boundary conditions on the predicted ozone concentrations. The calculated derivative matrix provides valuable information e.g., on the ordering of the infl uential factors or the localisation of highly affected regions. Two fundamentally different domains of the Auto-Oil II programme were used as test cases for the simulations, namely Athens and Milan. The results suggest that ozone concentration be highly affected by its own boundary conditions and subsequently, with an order of magnitude less, by the boundary conditions of NOX and VOC.
Highlights
A photochemical Air Quality Models (PAQM) utilises meteorology, air quality, terrain and emissions data and a chemical mechanism and simulates the concentration of chemical species in each cell of a 3D grid
Sensitivity analysis of 3D PAQM is a field of great importance
We applied a general approach that in only one execution of a PAQM can provide the sensitivities of all output parameters with respect to all its inputs
Summary
A photochemical Air Quality Models (PAQM) utilises meteorology, air quality, terrain and emissions data and a chemical mechanism and simulates the concentration of chemical species in each cell of a 3D grid. Comprehensive sensitivity analysis of 3-Dimensional (3D) PAQM does not exist, mainly because of the associated computational cost, human effort and data needs (Peters et al, 1995; Carmichael et al, 1997; Pielke, 1998). The enhancement of a computer code with derivatives can be made through four different techniques (Bischof, 1994) including «divided differences» or «symbolic differentiation». A new technique called Automatic Differentiation (AD) (Rall, 1981; Griewank, 1989; Griewank, 2000) became available which augments a computer code with derivatives in a completely mechanical way. The accuracy of the AD-calculated derivatives has been demonstrated in several studies (Bischof et al, 1992; Horwedel et al, 1992; Park and Droegemeier, 1999; Kioutsioukis and Skouloudis, 2001; among others).This study examines the robustness of a model’s predictions
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
