Data assimilation for air quality models

Research output: Book/anthology/dissertation/reportPh.D. thesis

  • Jeremy David Silver, Denmark
The chemical composition of the Earth’s atmosphere has major ramifications for not only human health, but also biodiversity and the climate; hence there are scientific, environmental and societal interests in accurate estimates of atmospheric chemical composition and in understanding the governing chemical and physical dynamics.
Concentrations of atmospheric trace gases such as ozone, carbon monoxide and nitrogen dioxide vary substantially in space and time, and this variation can be investigated by various methods including direct measurements, remote-sensing measurements and atmospheric chemistry-transport models (CTMs). Each of these methods has their limitations: direct measurements provide only data at point locations and may not be representative of a wider area, remotely-sensed data from polar-orbiting satellites cannot investigate diurnal variation, and
CTM simulations are often associated with higher uncertainties. It is possible, however, to combine information from measurements and models to more accurately estimate the state of the atmosphere using a statistically consistent framework known as “data assimilation”.
In this study, three data assimilation schemes are implemented and evaluated. The data assimilation schemes are coupled to the Danish Eulerian Hemispheric Model (DEHM), a large-scale three-dimensional off-line CTM, and the data ingested were retrievals of atmospheric composition from polar-orbiting satellites. The three assimilation techniques applied were: a three-dimensional optimal interpolation procedure (OI), an Ensemble Kalman Filter
(EnKF), and a three-dimensional variational scheme (3D-var). The three assimilation procedures are described and tested. A multi-faceted approach is taken for the verification, using independent measurements from surface air-quality monitoring stations, satellite retrievals of atmospheric chemical composition and comparison with idealised simulations.
The 3D-var and EnKF schemes are capable of performing multi-species adjustments, meaning that observations of different chemical components can be assimilated simultaneously.
Furthermore, observations of one chemical species can be used to adjust concentrations of other (unobserved) species. Most of the methodology used in data assimilation for CTMs is based on developments within the field of numerical weather prediction, where multiparameter assimilation schemes are the norm. The verification of the 3D-var and EnKF schemes are expanded to assess the potential benefits of joint multi-species adjustments (c.f.
adjusting individual species independently) or direct adjustment of unobserved species.
Original languageEnglish
Number of pages196
ISBN (Electronic)978-87-92936-86-8
Publication statusPublished - 31 Jan 2014

See relations at Aarhus University Citationformats

ID: 69848990