4.4 Graphical and Analytical Software Visualization Tools for Evaluating Meteorological and Air Quality Model Performance

Tuesday, 8 January 2013: 2:15 PM
Room 11AB (Austin Convention Center)
Irene Lee, Exponent, Natick, MA
Manuscript (908.6 kB)

Handout (2.6 MB)

A critical part of meteorological and air quality modeling is evaluating the output modeled data in order to accurately assess model performance. The ability to see the data in a visual format can be an indispensable tool. Toward this goal, a software package has been developed to allow users the ability to quickly and seamlessly visualize and assess model output data including data from numerical weather prediction models such as WRF, MM5, RUC, ETA, NAM, and RAMS, data from the CALPUFF dispersion modeling system including CALMET and CALPUFF, as well as observed data such as ISHWO and FSL.

The software package currently includes a tool to quickly animate multiple layers including wind vector fields, temperature, precipitation, and mixing height isopleths, as well as concentration contours. The selected layers can be superimposed over terrain, land use, or other user-selected custom base maps, such as aerial photos. The user-defined animations can then be saved and imported into PowerPoint, allowing multiple animations to be viewed in step with each other. This gives the user the ability to view multiple model simulations simultaneously and quickly assess, for example, the resulting impacts of various model settings. Another included software tool allows users to interactively plot multiple variables against each other as either time-series or scatter plots. This can be used to assess internal model performance, such as modeled temperatures during the day versus at night, or overall model performance, such as model output pollutant concentrations against monitored observations. Real-world applications of the software package will be demonstrated.

Additional software is currently in development including profile plots and computation of statistical measures such as bias, gross error, and residuals.

The entire software package will be freely distributed. In addition, all of the direct inputs to the software package are in a non-proprietary, readable ASCII format. This enables the software to be universally used with any type of data with just a straightforward conversion to the software ASCII format.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner