COMET Module: "Ten Common NWP Misconceptions" (2002-2003)

Ten common misconceptions about the way NWP models work and hence how they can be interpreted, plus explanatory corrections of those misconceptions, plus quite a bit of more or less related material.

Appropriate level: Advanced undergraduates and above.

Summary comments: This module is heavily illustrated, has audio narration, and is very light on text. (A print version substitutes text for the audio narration; animation is lost but static color graphics remain.) A "Test Your Knowledge" section, with feedback, ends the presentation of each misconception.

Each misconception is a kind of Trojan horse; the misconception is addressed directly, but it is also used as an excuse to present quite a bit of less narrowly (but still at least broadly) relevant material.

The misconceptions vary in their degree of obscurity, but at least some of them look potentially useful for use in a basic NWP course that doesn't focus exclusively on theory. However, to appreciate the misconceptions and the corrections to them, students need already to know the basics about how numerical models are formulated and used, and in some cases more than that--some of the misconceptions are quite specific, as the titles below probably suggest.

At least one of the misconceptions (I didn't examine all of them closely), "A 20 km Grid Accurately Depicts 40 km Features" (#3 below), makes important points about model resolution, but its examples cite values relevant to 2002-2003 era models and so are not as directly relevant today as they were then. Several "misconceptions" refer to the eta model, which few students henceforth will recognize. The general concepts remain highly relevant, however.

Misconceptions addressed include:

1. The Analysis Should Match Observations
   (Presents a summary of observational platforms in a nice conceptual diagram. Raises the concept of assimilation cycling.)
   
   
2. High Resolution Fixes Everything
   (Makes the point that model components work synergistically; improving resulution alone won't guarantee a better forecast.)
   
   
3. A 20 km Grid Accurately Depicts 40 km Features
   (In addition to discussion of spatial resolution, this section includes an animated graphic showing effects of finite differencing on phase speeds of sine waves of various wavelengths and speeds.)
   
   
4. Surface Conditions are Accurately Depicted
   (Contains a long table summarizing the eta and Canadian GEM model surface fields; it would be nice if this could be updated to the WRF-NMM (the current NAM). Another section describes the effects of vegetation in a single-column model.)
   
   
5. CP Schemes 1: Convective Precipitation is Directly Parameterized
   (Explanation of a convective sequence in nature and one in a non-convection-resolving model without a convective parameterization, using schematic soundings superimposed on cloud drawings; a comparison of adjustment and mass flux convective parameterization schemes (Betts-Miller-Janic and Kain-Frisch in particular).
   
   
6. CP Schemes 2: A Good Synoptic Forecast Implies a Good Convective Forecast
   (Brief discussion of resolution, illustrated with a couple of diagrams; fine-tuning convective parameterization [CP] schemes; over- and under-active CP schemes; different schemes in the same model)
   
   
7. Radiation Effects are Well-Handled in the Absence of Clouds
   (Discussion of the complexity of representing radiative processes in a model, including the effects of clouds and clear-sky biases in the eta model--hope this gets updated! Brief summary of how models address radiation procesess in general.)
   
   
8. NWP Models Directly Forecast Near-Surface Variables
   (Adjustment of temperature from the lowest model level to 2 meters in the GFS model, illustrated; how this is done in other models, mentioned in very general terms; problems that can arise with this adjustment process; effect of vertical coordinate on the adjustment, including as examples the eta model, which is now out of date, and the Canadian GEM model, which uses a terrain-following sigma coordinate and is therefore still relevant to the WRF-NNM [the current NAM] below about 400 mb; effect of terrain representation--envelope, silhouette, mean--on the adjustment process.)
   
   
9. MOS Forecasts Improve with Model Improvements
   (An introduction to MOS, including its development and implementation. Issues with rarer types of events; smoothing; advantages and disadvantages of MOS schemes; situations when MOS might produce a poor forecast.)
   
   
   
10. Full-Resolution Model Data are Always Required
    (Comparisons of fields produced by AWIPS at 22, 40, and 80 km resolutions from eta model output; resulution vs. scale of atmospheric features, with animated graphic of sine waves; the issue of smoothing, illustrated with plots of specific humidity overlaid by temperature contours at 40 km resolution [unsmoothed] and 80 km resolution [smoothed]; vertical resolution, illustrated with tephigrams, which take some work to understand.)