13 Nov 2018
13 Nov 2018
Status: this preprint was under review for the journal ACP but the revision was not accepted.

Bias correction in assimilation of AOD observations with WRF-Chem

Anton Kliewer1, Milija Zupanski1, Qijing Bian2, Sam Atwood2, Yi Wang3,4,5, and Jun Wang3,4,5 Anton Kliewer et al.
  • 1Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, CO, USA
  • 2Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado USA
  • 3Department of Chemical and Biochemical Engineering, The University of Iowa, Iowa City, IA USA
  • 4Center of Global and Regional Environmental Research, The University of Iowa, Iowa City, IA USA
  • 5Interdisciplinary Graduate Program in Informatics, The University of Iowa, Iowa City, IA USA

Abstract. Accurate prediction and representation of three-dimensional aerosol distributions in the littoral (coastal) zone is both desired and difficult with many compounding factors contributing to this problem. To reduce uncertainty in forecasting in coastal regions, a coupled meteorological-aerosol data assimimilation (DA) system has been configured to include satellite observations of aerosol optical depth (AOD). These high-resolution observations are from newly-devised retrieval algorithms that utilize Moderate Resolution Imaging Spectroradiometer (MODIS) data to retrieve AOD over the coastal and turbid water surface. The Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) is combined with an ensemble-based DA system, the Maximum Likelihood Ensemble Filter (MLEF), to simulate a dust event over the Arabian Peninsula from 2016. The assimimilation of AOD observations required the development of a forward operator that converts model predictions into observation space. This operator, which incorporates hygroscopic growth of aerosol particles and determines extinction efficiency based via Mie theory, has a positive bias between the model guess and the retrieved AOD observations. In order to reduce this bias two different methods are proposed and evaluated. One is a moving average method employed throughout the case study while the other relies on a statistical re-sampling approach. The conclusion of these experiments, determined by a number of metrics including, but not limited to, root mean square (RMS) errors, an evaluation of the reduction in the cost function, and degrees of freedom for signal (DFS), indicate that the bias reduction scheme that accumulates bias information throughout the case study outperforms the method based on re-sampling. This conclusion is corroborated by inspection of the analysis increments from the DA process and by the innovations in observational space. An analysis of the non-Gaussian innovations resulting from the non-linear forward operator is also presented. This research is in support of a Multidisciplinary University Research Initiative (MURI) supported by the Office of Naval Research (ONR) with the primary goal of understanding aerosols in the littoral zone.

Anton Kliewer et al.

Status: closed
Status: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version Supplement - Supplement
Status: closed
Status: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version Supplement - Supplement

Anton Kliewer et al.

Anton Kliewer et al.


Total article views: 1,736 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
1,057 647 32 1,736 31 35
  • HTML: 1,057
  • PDF: 647
  • XML: 32
  • Total: 1,736
  • BibTeX: 31
  • EndNote: 35
Views and downloads (calculated since 13 Nov 2018)
Cumulative views and downloads (calculated since 13 Nov 2018)

Viewed (geographical distribution)

Total article views: 1,692 (including HTML, PDF, and XML) Thereof 1,689 with geography defined and 3 with unknown origin.
Country # Views %
  • 1
Latest update: 27 Jan 2023
Short summary
This research is focused on improving numerical weather prediction by including data regarding aerosols in the atmosphere. Using weather prediction models along with data assimilation (the process of marrying observations with a model prediction), a better representation of the atmosphere can be described. As no model or observational platform is ever perfect, the aerosol observations have to be de-biased (adjusting for systematic error). Here we look at two such methods.