Rapid Refresh (RAP) and High-Resolution Rapid Refresh (HRRR) Forum
Re: Mid-level data missing?
Posted By: John Brown In Response To: Re: Mid-level data missing? (Scott Dennstaedt)
Date: 17-March-06 2010Z
In Response To: Re: Mid-level data missing? (Scott Dennstaedt)
Sorry, I must have inadvertently truncated my last post responding to Scott Dennstaedts queries...
I'll try to explain. The key is in the "hybrid" nature of the vertical coordinate. We assign each coordinate surface (from 1 to 50) a potential temperature (theta) value, starting at 224K (-49C) for the bottom level, and increasing at variable increments to 500K at the top (level number 50). Further, we don't let the coordinate surfaces go below ground, and we don't let them get too close to each other in the vertical. Given these 2 constraints, which we might call the above-ground and the separation constraints, the RUC model seeks to put the coordinate surface as close to its assigned potential temperature as possible.
The first level above the surface (assigned 224K) we keep always at 5 m above the ground surface, so it is always "terrain following". The second one, 232K (-41C) potential temperature, is also very cold, and almost always is colder than any potential temperature occurring in the model domain. Under these typical conditions, the code will put this coordinate surface as close to the lowest coordinate surface as the separation constraint allows (for these 2 levels, a separation of about 5mb over low surface elevations). Unless conditions at low levels are very cold, same goes for the 3rd level above the surface (240K), and so on, regardless of the temperature lapse rate. Only when the potential temperature at the altitude of the coordinate surface dictated by the above-ground and separation constraints is actually colder than the potential temperature assigned to that surface, does the coordinate surface get placed at a higher altitude, specifically, at the altitude of its assigned potential temperature.
Turning to the RUC forecast example soundings from SFO linked from Scott Dennstaedt's post, the "sparse data" 18Z 10 March 2006 sounding
shows the data close together at the low levels even though the lapse rate is approximately dry adiabatic. This is because the coldest potential temperature in the sounding is about 9C (near the surface), or 282K. This happens to be the potential temperature assigned to the 13th RUC coordinate surface above the ground, so the 12 surfaces assigned a colder potential temperature are as close to the ground surface as the separation constraint allows. Because the separation constraint assigns the 282K coordinate surface to an altitude where the potential temperature is warmer than 282K, this surface likewise is forced to be terrain following. The lowest coordinate surface to actually be placed at the altitude of its assigned potential temperature is 285K, placed near 700mb. The next coordinate surface, assigned 288K, is placed just above 600mb, readily located on the plot by a kink in the dew point curve and the presence of a wind barb. Above that, the 291K surface is placed just below 500mb, 294K near 450mb, and so on.
The "not-so-sparse" example
is markedly more stable between 900 and 600mb, and has a higher trop. Here it is apparent that the first coordinate surface to be placed at its theta level in the sounding is 288K, ~830mb. The greater stability contributes to the "not-so-sparse" character of the plot. In addition, the coordinate surfaces are spaced only 2K apart in the 294-322K interval, further contributing to more coordinate levels (this range of potential temperature was farther aloft in the colder "sparse data" sounding).
The algorithm to locate, or "remap" the vertical coordinate is applied to each grid column at each time step of the model forecast. This means that in areas that are cold at the surface, the vertical extent of the model domain where the coordinate surfaces are constrained to be terrain-following will be less than areas where surface temperatures are warmer. An example of this for a RUC20 forecast can be seen on the right panel of Slide 7 of an on-line powerpoint at
This is a N-S cross section from near the Gulf Coast on the left (warm) to north of Lake Superior on the right (cold). Although this is for the old RUC20, the number and distribution of vertical levels is the same in the RUC13.
Scott, give me an email or phone call if this is still unclear.
Messages In This Thread
Rapid Refresh (RAP) and High-Resolution Rapid Refresh (HRRR) Forum is maintained by RR/RAP Forum Administrators with WebBBS 5.12.