Bottom Two Panels: Clouds and Precipitation
Let<92>s start with the lower-left panel of the 6-hour WRF prog intialized at 12Z on February 11, 2009, which we<92>ve isolated in the figure below. For this panel, the designated level is 700 mb. Why 700-mb? This pressure level is approximately midway between 1000 mb and 500 mb, which places it near the epicenter of the "weather action.<94> Moreover, 700 mb marks the first mandatory pressure level that almost completely lies in the <93>free atmosphere.<94> In other words, 700 mb is high enough to be removed from the constraints of the earth<92>s surface, except for some of the highest peaks in the Rockies and the tops of other mountains in the western United States (the Cascades in Washington and Oregon, for example; here's a photograph of Mount Ranier, whose elevation is 14,410 feet).
To get your bearings, we point out that the thicker, solid contours represent 700-mb heights (labeled in dekameters). Without reservation, the predicted 700-mb heights provide forecasters with a more complete picture of the evolving weather pattern in the lower half of the troposphere. Perhaps the most useful field on the bottom-left panel is the color-coded forecast for relative humidity. For starters, contours of predicted 700-mb relative humidity (the thinner solid lines) are drawn for 10%, 30%, 50%, 70% and 90. Green shades represent regions with high 700-mb relative humidity (greater than 70%). The darkest green marks areas with 700-mb relative humidity greater than 90%, which is getting pretty close to saturation. Meanwhile, oranges indicate low 700-mb relative humidity, with the darkest orange indicating areas with 700-mb relative humidity less than 10%.
If you look at the top two panels of the WRF's 6-hour forecast initialized at 12Z on February 11, 2009, and focus your attention on the surface low over Illinois and the closely trailing 500-mb trough, you'll note that the 500-mb trough has a northwest-southeast tilt, indicating that this low-pressure system is a mature cyclone (revisit Chapter 13). As a result, we should expect the low to have a dry conveyor belt. And, indeed, it has one! You should be able to detect the developing dry slot over western Mississippi, western Tennessee, southeastern Missouri, and southwestern Kentucky on the 700-mb prog.
Obviously, mature mid-latitude cyclones are not always a fixture over the United States. So what other utility does the bottom-left panel serve? Based on empirical evidence, there is a strong statistical correlation between clouds and a predicted 700-mb relative humidity greater than 70%. Though this may seem at odds with your understanding of clouds forming when the air becomes saturated, keep in mind that we emphasize "statistical" here. Indeed, forecasters have observed the link between clouds and 700-mb relative humidity over a long time.
Want more concrete proof? Check out the visible satellite image above at 18Z on February 11, 2009 (same as the valid time for WRF's 6-hour forecast initialized at 12Z on February 11, 2009). Note how the green pattern of high relative humidity around the low centered in Illinois correlates closely with the observed cloud shield. The WRF made a pretty good forecast, wouldn't you agree? Compare the bottom-left panel to the satellite image around the time the prog was valid. It essentially predicted the shape of the cloud structure associated with the mature low, correctly capturing the position and shape of the comma head west and southwest of the low's center. The WRF also did a pretty good job on the dry slot, which is the clearing associated with the dry conveyor belt to the east of the cyclone's comma head.
The model wasn<92>t perfect (no surprises there). For example, note that the WRF predicted 70% relative humidity (a high probability of clouds) over parts of southern Illinois. But the 1815Z visible image indicated partly to mostly sunny conditions in this area. As a result of the premature clearing in this region, showers and a few thunderstorms started to develop as solar heating at the ground combined with lowering temperatures at 500 mb (associated with the trailing trough) to destabilize the troposphere. For confirmation, here's the 1815Z radar reflectivity from the radar at St. Louis, Missouri (KLSX).
Weather forecasters can also predict precipitation (especially stratiform precipitation) using the bottom-left panel. When the predicted 700-mb relative humidity is greater than 90% (dark green areas on progs from Penn State), precipitation is likely. To comprehend this assertion, let's continue to exploit information from the bottom-left panel on the WRF's 6-hour forecast initialized at 12Z on February 11, 2009. Focus your attention on the comma head of dark green (predicted 700-mb relative humidity greater than 90%) arcing westward from northern Illinois across Missouri. Much of this dark, green zone of high relative humidity is the handiwork of the mature low's cold conveyor belt. Now compare the shape of this area to the corresponding region on the composite of radar reflectivity at 18Z on February 11, 2009 (shown below). Here's a side-by-side comparison. Clearly, the region of 700-mb relative humidity greater than 90% associated with the cold-conveyor belt was a pretty good indicator of where stratiform precipitation would eventually fall.
Though the WRF 700-mb prog did a pretty good job of capturing the precipitation north and west of the low, you may have noticed the lack of 700-mb relative humidity greater than 90% with the lines of showers and thunderstorms ahead of the low's cold front in Alabama. To resolve this apparent contradiction, keep in mind that the spatial scale of a thunderstorm (cumulonimbus cloud) is typically less than the horizontal spacings of one grid box in most operational short-range computer models (see the schematic below ). Although the relative humidity inside a thunderstorm is about 100%, the relative humidity surrounding the thunderstorm is typically much lower. This has the effect of decreasing the overall average relative humidity in the grid box. Thus, in convective situations, the model-predicted 700-mb relative humidity can be misleading if you use it as the sole indicator of precipitation.
Speaking of precipitation, let<92>s turn our attention to the lower right-hand panel of the 6-hour WRF forecast initialized at 12Z on February 11, 2009 (see below). The first line at the bottom of the panel tells the story of this prog: The colored areas represent the total liquid precipitation that the WRF predicts will have fallen in the six-hour period ending at the time that the computer prog is valid (in this case, 18Z on February 11, 2009). So the colored-coded areas represent a cumulative forecast of the liquid precipitation that will have fallen during the entire six-hour period between 12Z and 18Z on February 11, 2009. For the record, liquid precipitation translates to rainfall or the equivalent amount of meltwater (assuming it snows). We want to emphasize here that the color-coded regions on the bottom-right panel do not constitute a snapshot of future radar reflectivity at the specific time the prog is valid.
You can use the color key along the bottom of the bottom-right panel as a guide to quantify the predicted liquid precipitation. The numbers corresponding to each color represent precipitation totals expressed in hundredths of an inch. So, for example, 10 means 0.10 inches, while 150 equals 1.50 inches. Purple areas (like the one in northwestern Pennsylvania and western New York) represent regions where the WRF predicted between 0.01 and 0.10 inches of liquid precipitation (1 to 10 on the scale) in the six-hour period ending at 18Z on February 11, 2009.
Here<92>s another example. Note the light blue off the Middle Atlantic Coast. Reading off the color scale along the bottom, light blue corresponds to a six-hour predicted rainfall between 0.10 inches and 0.25 inches. This range is consistent with the "23" near the center of the light-blue blob, which is the prog<92>s way of designating a relative maximum of 0.23 inches of rain. Lesson learned: Numbers appearing within a colored area on this panel represent a relative maximum in precipitation. Moreover, a snowflake-like icon marks the exact location where the model predicts the relative maximum.
We<92>re almost home. There<92>s one other 700-mb parameter plotted on the bottom-right panel. The thin, solid lines that are labeled with pluses indicate areas of predicted downward motion at 700 mb, while the thin, dashed isopleths labeled with minuses mark areas of predicted upward motion at 700 mb. The integers that mark some of the pluses and minuses represent the speeds of the air's ascent and descent (details will be forthcoming).
The use of "-" for upward air motions and "+" for downward motions may seem backwards to you. To reconcile this seemingly contrary sign convention, consider a parcel of rising air. Over time, the parcel's pressure decreases as it ascends higher and higher into the atmosphere. For subsiding parcels, pressure increases during descent. Thus, the sign convention corresponds to the pressure change of the parcel with time - negative for rising air and positive for sinking air. Please note that some Web sites reverse this convention, but you can usually tell which sign convention prevails because precipitation will generally correspond (or lie close to) areas of upward motion.
For the record, the units of vertical motion on the bottom-right panel are microbars per second (a microbar is one-thousandth of a millibar). A speed of 1 microbar per second is about 1 centimeter per second, which is less than half an inch per second. These tortoise-like up-and-down speeds are representative of synoptic-scale values because vertical pressure gradients are, on average, balanced by gravity. Such slow ascent and descent are the rule rather than the exception.
There you have it. A complete whirlwind tour of a standard four-panel prog. You now have the basics to navigate the models listed on the Penn State e-Wall. We should caution you, however, that there are many permutations of the four-panel prog on the World-Wide Web. These variations can range from simple color differences to wholesale substitutions of a prog from a different pressure level. For example, the image above shows a four-panel prog of the GFS model, initialized at 00Z on August 7, 2009, and valid 48 hours later at 00Z on August 9. This prog comes directly from the Web site of the National Centers for Environmental Prediction in Camp Springs, Maryland, where the computer model is run. Here, a 200-mb prog occupies the upper-left panel, showing heights and winds. The upper-right panel is the familiar 500-mb chart with predicted heights and absolute vorticity, but with predicted wind barbs added. The lower-left panel shows predicted sea-level pressure and 1000-500 mb thickness, but also liquid precipitation. Finally, the lower-right panel is an 850-mb prog, displaying predicted heights, temperature, and wind barbs.
Armed with the ability to interpret computer forecasts, how do meteorologists resolve the inherent imperfections in the WRF and other computer models?