Fun Weather Facts: Where do Forecasts Come From? (Part Two)

In my last blog I wrote about how the atmosphere worked to produce weather. We learned that the weather this second determines the weather the next second and, so on, forever and ever, as the atmosphere moves itself along and weather happens. 

Plotting weather maps is fine for showing what the weather is now, but that approach is a disaster for predicting what will happen in the future.  The best (and only) way is to mimic how the atmosphere works on a computer. 

The first step is to figure out what the atmosphere is like right now.  Since it is too expensive to do this very often, measurements are only taken around the world at two times. This is at (no holidays off) 4 a.m. and 4 p.m. PST.  Data is taken both above the ground (balloons, aircraft, and satellites) and on the ground (land stations, ships). This data is not evenly distributed around the globe, but it is wherever data collecting stations can be located. Our local radiosonde stations (instrumented balloons) are at Quillayute on the Olympic Peninsula and Spokane.

The next step is to move the data onto a three-dimensional grid that the computer model will use for mimicking the atmosphere.  This can be tricky since the data is not evenly spaced and needs to be moved onto the grid preserving as well as possible all the weather.  There are always errors in measurements so these need to be found and eliminated.  Sometimes data that seems to be an error is really showing an important weather feature, and this data to be left in.

After this is done the computer model starts to move the atmosphere forward and create weather.  Because forecasts need to be done in a timely manner and the limitations on computer capacity, most forecast models move the atmosphere forward in individual steps of several minutes.

Since the computer grid is spaced apart, the mountains, valleys, hills are not perfectly formed in the models.  The larger the area of the globe the models forecast for, the cruder the surface of the earth (its topography) is depicted.  This can cause errors where the topography is important to the weather (think Puget Sound with the Olympics and the Cascades).

The computer models contain all the important things that happen in the atmosphere including such details as snow cover, and sea ice.  The models even mimic convective activity (thunderstorms, etc.) but not as detailed as in the real atmosphere.

Since the human brain is limited in computational power no forecaster can out-think the computer models, so every forecaster goes to the same well to drink.  Forecasters might slightly modify the computer forecast, but most forecasts are the same and it makes all forecasters pretty equal (except for their entertainment value).

The UW Atmospheric Sciences Department runs its own computer models that are especially designed for the Pacific Northwest.  Since these local models cover a smaller portion of the globe they can do a better job of mimicking the tricky NW topography.  If you are interested in learning more about the computer models you can stop by Young’s Enumclaw Flowers and Gifts and I can show you how to find and read the local models. 

Next time, I will write about why the computer models can fail and how satellite pictures and radar are used to improve forecasts.