Comparing GFS Medium Range Forecast Computer Model Products
There is a tendency for weather outlets to become ‘model dependent,’ or accept what various computer model runs forecast as gospel. Today I decided to share a few examples of GFS model runs side-by-side where you can observe their obvious and sometimes subtle changes for forecasts of specific regions, especially in the medium to long-range outlooks.
First we will begin with the forecast snow depth GFS product I mentioned previously in this post. Remember that this product essentially depicts that if snow were to fall in some area within a forecast cycle, how much snow would likely be left on the ground at the end of each interval within that forecast period? Today I will show 3 different GFS runs side-by-side from three different days, allowing you to easily view the changes for each region GFS has shown (please be sure to view closely the ‘valid’ timestamp(s) as the animations run on the bottom left-hand corner).
|GFS Snow Depth Product Comparison ( animations 800×600 click for full size )|
|Initialized September 30, 18z||Initialized October 3, 12z||Initialized today (Monday) October 6, 12z|
Notice I have shown two different time windows, the first and second animations above are from September 30 to October 3, (4 days) and the second window is only 3 days apart, yet since they are animations of a 7 day window plus (180 hrs), the time frames they represent overlap. Are their differences in total snow depth forecasts that GFS is representing over those time frames? Do the extended outlooks (later time frames) show the potential for differing amounts of snow depth? Of course. However, the information represented in all output fields available allow a deeper penetration into understanding atmospheric processes. Of course simple output fields such as above still represent important and pertinent information that we can use to forecast.
Almost forgot to mention, notice in the last animation above (far right) the large swath of snow GFS has forecast beginning around this Friday afternoon through next Tuesday. Also notice that in the last cycle (valid 180hrs) snow that was forecast to fall in the pacific Northwest and other ‘less obvious’ areas is missing in the last time frame. As a sample, I have included for your viewing pleasure the last valid time in the run, or the 180 hour forecast product. Remember this product is how much snow is forecast to be ‘left’ on the ground at the end of a forecast period. So if some snow ‘were’ on the ground at the end of the cycle, any new snowfall would be added to that which is existing. Exciting isn’t it! Let’s see if something happens! =)
|GFS Forecast Snow Depth Product ( static image 1024×768 click for full size )|
|Initialized 12z today (October 4, 2008), Valid time ending 180 hrs or 00z Tuesday, October 14, 2008|
Our second comparison today is of forecast liquid precipitation.
|GFS Forecast Liquid Precipitation Product Comparison ( animations 800×600 click for full size )|
|Initialized October 3, 12z||Initialized today (Monday) October 6, 12z|
The above graphics show forecast liquid precipitation at 6 hour intervals out to 180 hours, or just over 7 days. The graphic at above left was initialized for the October 3, 12z run, and the second for today at 12z. Both animations provide an example of interesting comparisons can occur, in this instance most notably in the Pacific Northwest, Great Plains, and southeastern United States. Does the intensity, apparent duration, and location of forecast liquid precipitation remain essentially the same for both runs, or do you see differences? Please remember to view the timestamps on the bottom left-hand corner when comparing! =)
It is important to remember that simple fields such as the above are only a few of many that must be examined to determine the type, intensity, frequency, location and so forth of various weather events that we forecast to occur. Once an understanding of atmospheric processes is obtained, from that one can utilized various tools such as computer model output to determine their forecasts. For example, last week I said that looking into my crystal ball I see the potential for tornadoes, based partially upon GFS output. Was I correct?
Be sure to stay tuned to your favorite weather outlet, stay informed, and stay safe!