PROJECT 4

Title: Visualization of Uncertainty in Meteorological Forecast Models

Purpose: A prototype is designed to facilitate an understanding of meteorological forecast models by viewing an animation generated by sets of data and is able to tell the uncertainty between the results from the models and actual weather.

Models and Data:

(1) Three models: eta, NGM, and spectral, all of them are released as the NMC¡¦s (National Meteorological Center) short-term (2-day only) forecast product.

(2) Each of the three models is released as a series of static maps during 48 hours, and every six hours a map is produced (so there are 8 maps for each model). In each map, there are 30 points (latitudinal) by 40 points (longitudinal) grids (so there are 1200 grids in each map) representing the North America with 50Km between two points. In each grid, it is stored an array of meteorological variables, including surface and upper air winds, temperature, humidity and pressure. The data was released at 7:00am on February 2, 1995.

Visualization presented:

(1) Using surface pressure-Isobars to represent the weather conditions and using shaded area to represents the uncertainty between results from three models and actual weather data.

(2) Use interactive control of animation to represent the time-step changing of the weather. Click here to view the interactive control of the animation of this project. Ps. You might be asked to download ¡§shockwave¡¨ in order to view the animation, so go ahead and plug it in!!^_^

Uncertainty Information:

(1) In this project, they use data of three different models and averaging it, this generated disagreement between actual weather data and the result from these three model due to each model has it¡¦s own representing of the high/low pressure resolution.

(2) There existed ¡§temporal discrepancy¡¨(means not matched) among the models because of the time-period interval they collected data, i.e data from model was caught every 6 hours while actual pressure values were received every subsequent 12 hours.

Interface Control: We are able to stop/backward/forward the frames to catch the specific time-step corresponding frame.

Suggestion by author:

(1) The prototype needs to be implemented in a data-driven environment that the analysts can easily apply the tool to any data set of interest.

(2) More flexible interactive controls are needed to allow analysts to change symbolization used to describe each model.