Talking points for the Cloud Height/Sounder Retrievals VISIT session
1.
This session consists of two independent topics. They are combined into this
teletraining session since they are both short topics.
2.
The first topic we will deal with is the interactive cloud height algorithm in AWIPS.
We want to achieve a basic understanding of how to use this product, knowing how
it works and knowing when it may be in error.
3.
In order to compute a cloud top height, two pieces of information are needed. The
first is the 10.7 um brightness temperature of the cloud and a sounding. The
sounding may come from either a radiosonde, model output or climatology. The
one assumption of the algorithm is that the cloud is thick, in other words analysis of
thin cirrus will likely show cloud tops that are too low because warmer
temperatures from the ground are getting through causing a warmer brightness
temperature. To bring up the cloud top height first call up a 10.7 um IR imagery on
AWIPS, then hold down the right mouse button, the graphic shown appears. Click
on sampling then hold the mouse over “Sample Cloud Heights”, on this sub-menu
you pick the source for the sounding. Also be sure to highlight “SKEW-T”, this
tool is critical in determining the accuracy of the cloud top height estimate. The
reason you do not use the cloud height algorithm with the 3.9 um imagery is
because of its sensitivity to reflected solar radiation.
4.
The graphic shows what AWIPS will display when roaming on a 10.7 um IR image
with the “sample cloud heights” tool turned on (as described in previous slide).
Each part of the system output is described on the slide. Keep in mind that the
status cannot be changed by the user. The only way to see the status change is to
roam around the satellite image until you see the status change to a different
method.
5.
When you highlighted “SKEW-T” in the “Sample Cloud Heights” sub-menu a
skew-t image will appear off to the side of your AWIPS display window. It is
critical to always have this window up when analyzing cloud heights so that you
will see what the algorithm is doing. As you roam across the IR 10.7 um image you
will notice the sounding changes, it is matching the closest observed sounding in
space and time. If you selected model as your source it looks for the nearest gridpoint and closest time to the model output. You may experience a lag response as
you roam the image since computations are being made on the fly so be sure to
roam back and forth to make sure you are using the closest sounding and a status
method that seems reasonable. The small yellow circle is plotted at the cloud top
height corresponding to the cursor on the 10.7 um IR image. This is the best way to
assess if the cloud top height is a reasonable estimate or not.
6.
If you roam a 10.7 um IR image in a region that is outside of the model domain you
will notice “CLIMO” as the source. This is a very crude estimate for a sounding,
therefore you should choose GFS as the source since this will most likely be a more
representative sounding.
7.
With the S status method, the yellow circle will be plotted on the temperature trace
(red line) at the level corresponding to the observed IR brightness temperature. The
cloud top height accuracy is dependent on the type of cloud observed, for example,
if you are looking at thunderstorm cloud tops then the S status would most likely
yield an estimate with more error compared to using the C status method.
8.
Notice the yellow circle is plotted at the ground, corresponding to clear skies.
9.
When analyzing thunderstorm cloud tops, you should roaming the IR image until
the C status is shown. Remember to only use model output when looking at the C
status method, there is a bug in using raob as the source. A fix is “in the pipeline”
but probably won’t appear in AWIPS until 2006.
10. Remember with the C method the yellow circle corresponding to the cloud top
height will be plotted on the parcel trajectory (purple line on the skew-t).
11. This algorithm is highly sensitive to small (even 1 degree) changes in temperature.
The threshold of 4.5 °C is arbitrary, there is a .sup file that controls the thresholds
values. This status method should perform well with low stratus with no clouds
above it. The yellow circle may not necessarily be plotted along the temperature
trace, it will be plotted wherever the dewpoint drying occurs. The sounding used
here is the observed sounding from KVBG which adequately represents the lowlevel inversion. The cloud top is found to be at the top of the inversion below the
drier air, as you would expect.
12. M is useful when the observed IR brightness temperature occurs at more than one
level. Since this is usually a situation with an inversion, this method will be found
much more easily with a sounding in close time/space proximity since models
generally struggle with sharp low-level inversions.
13. 2 frames in this slide. Which of these two sources (raob and Eta) is likely to give a
more accurate sounding to determine the cloud top height? The key is to keep in
mind the time. Here, the time is 23:15 UTC so it is using the closest available
sounding time (12:00 UTC) which is quite old. The model at this time would
probably give the better estimate. If we were to analyze this situation 2 hours later,
with the 00:00 sounding then the raob would likely give a better estimate. Always
keep in mind proximity in time/space to soundings in determining whether to use
model or raob as the source.
14. Let’s compare two status method (D and S) in a situation with low stratus off the
California coast.
15. 2 frames in this slide. Keep in mind that the user cannot select the status; the only
way to change the status is to scroll on the image until you see the status change.
The estimate using the D method takes into account the inversion and the drier air
above it, making for a more reasonable estimate than S which appears too high into
the drier air. M may have done well here as well but it never appeared when
scrolling around the image. Always scroll around the image and watch the skew-t
to assess how reliable the cloud top height estimate is.
16. Here is a situation with low stratus in eastern Colorado. Using the RUC model as
the source shows the lack of resolution in the sharp low-level inversion. Since the
inversion is poorly represented in the RUC, this will lead to an erroneous cloud top
height estimate.
17. Low-level stratus over eastern Colorado in close space/time proximity to the
Denver, CO sounding, therefore raob is used as the source. Here, the inversion is
shown nicely using the observed raob, however the accuracy of the cloud top height
estimate varies considerably depending on what status appears as you roam the
satellite image. Using the S method, it finds a level where the observed brightness
temperature exists, but a quick look at the skew-t shows it is clearly above the
inversion in the drier air and therefore is an erroneous estimate. Moving the mouse
cursor until you find M as the status leads you to a much more accurate cloud top
height since it uses the temperature found at the lower level within the inversion.
18. Another product available in AWIPS is the GOES Sounder cloud top height
product. Here, you cannot choose a source, the GFS is used in generating this
product. This product is at a coarser spatial resolution since it is using the GOES
sounder rather than the GOES imager. This product only comes out hourly,
whereas the imager comes out 4 times per hour, or 8 if the satellite is on a RSO
schedule. The GOES sounder cloud top height product will do better in the upper
half of the troposphere since it uses a different method than the interactive cloud
height algorithm. The GOES sounder cloud top height product uses the radiances
and the GFS model in a physical relationship to compute cloud top height. The
sounder doesn’t suffer from the thin cirrus problem that the interactive cloud top
height product does (remember it must assume a thick cloud). In the bottom half of
the atmosphere the interactive cloud top height should do better since it can account
for multiple cloud heights, low-level inversions etc. which the Sounder might
struggle with since it just uses a simple blackbody. The methodology of coming up
with the height is different as well, it doesn’t make use of the radiances and a
physical relationship, it just takes an IR cloud top temperature and tries to find the
corresponding height on a sounding.
19. When analyzing convection in a tropical environment, be aware that the cloud top
temperature may be much colder than overshooting tops (OST) in mid-latitudes but
it may not be THAT much higher in terms of cloud top heights because of the
colder tropopause in the tropics. This can be useful in analyzing warm rain
processes with flash flooding in the warm season.
20. Sounder point retrievals have been available on AWIPS since the OB2 release, but
few forecasters know they exist, or how to plot them. This short session shows how
to plot the retrievals and discusses the limitations associated with them
21. 19 total channels results is relatively poor vertical resolution for temperature and
moisture retrievals. It's important to remember that the sounder and imager are
separate instruments, totally independent of one another, even though they both ride
on the GOES satellites.
22. The CONUS sector receives hourly scans, while the other sectors' scans are less
frequent. Note that South America receives no regular sounder scans. Rapid scan
operation (RSO) is possible for the sounder, but is rarely used.
23. Here is a loop from 29-30 April 2004 showing all 19 sounder channels. The
channels are arranged so that larger channel numbers are shorter wavelengths,
opposite that of the imager.
24. Sounder derived product imagery (DPI) has been available on AWIPS for several
years now. A VISIT teletraining session has already been developed which focuses
on the DPIs (see the last slide in this session for details). DPIs, in general, are much
better than the point retrievals because they display bulk quantities (like total
precipitable water), which is what the sounder can most accurately retrieve. There
is overlap among its bands' weighting functions, so retrieving moisture quantities at
a single level is difficult. DPIs include TPW, lifted index, skin temperature, and
cloud top pressure. They are best used for identifying gradients and trends on
hourly time scales.
25. The moisture adjustment from the first guess is much larger than the temperature
adjustment (which is at most 1-2 C from the first guess). For this reason, we'll
focus on the retrieved dew point profiles.
26. The points must be made "editable" in AWIPS before their location can be adjusted.
27. Use the volume browser to plot the sounding retrievals, choosing the "GOES Bufr"
option.
28. Here is an example sounder point retrieval. The surface parcel is the dotted line.
Remember that observed surface temperature and dew point are the surface points
for the soundings, and are used for the parcel.
29. The bottom line here is that actual dew point values from the sounder point
retrievals shouldn't be taken at face value. They tend to be more accurate in the
upper troposphere, and less accurate around 850mb. The best option is looking at
the sounder point retrieval along with other model forecasts...if there's a consensus,
it should increase your confidence. Give more weight to the sounder data in the
upper troposphere.
30. Here is an example of where the sounder retrieval improves the GFS first guess at
almost every level. There are some low clouds, so it's surprising a sounder retrieval
was performed, but apparently the cloud-clearing algorithm didn't pick them up.
The retrieval was pretty good nonetheless, probably since they were so low and had
very little vertical extent. The GFS first-guess was too dry in the lower troposphere
and too wet in the upper trop. The sounder retrieval correctly moistened the lower
part and dried the upper part. Remember, this is a single example.
31. This example was around 18Z, when DDC released a special sounding due to the
threat of severe weather. In this case, the sounder incorrectly moistened the lower
levels and dried the upper levels.
32. This example shows how using the point retrievals to evaluate moisture trends
provides generally accurate information.
33. There are a few hours missing due to clouds in the area, but you can easily see
moistening at the low- and mid-levels during the morning hours.
34. Summary
35. The GOES-R series of satellites will have much improved sounders and imagers.
The multispectral sounder will allow for much better vertical resolution and
therefore better accuracy in both temperature and moisture retrievals.
36. Links/References