Eta model had been used. The CIMSS single FOV (SFOV) retrievals
and DPI, started last spring (2002) for the
IHOP experiment, were originally, and continue to be, made
with an AVN first-guess. At 18 UT on 17 April 2003, the
switch over occurred for the 3x3 DPI. The effect of the changing
first-guess was evident at that time in the total precipitable water (PW)
DPI, from very dry conditions in SE Canada to more tropical moisture
settings across the Gulf of Mexico and around the West Indies, and
reaching into the SE US.
Now, use of a common first-guess (AVN) permits more readily straightforward
comparison of the difference between the 3x3 FOV and single FOV DPI.
Tests
were performed by
G. Gray and
J. Daniels at
NOAA/NESDIS/ORA/ARAD/FPDT that showed statistically improved
GOES retrievals were made using the AVN model, versus the Eta model.
It appears that the temperature forecast from the AVN model is
slightly better (the moisture forecast, slightly poorer) than that
from the Eta model. Thus, as the physical GOES retrieval algorithm
allows much larger changes [also read as "improvements"] to the
moisture profile (versus the temperature profile),
the better first-guess temperature data (i.e. from the AVN)
are preferred. Operational GOES retrieval processing at
NOAA/NESDIS/OSDPD/SSD also uses the AVN model as the first-guess.
[N.B. The AVN model is currently undergoing a name change
to the GFS (Global Forecast System).]
01 Apr 2003:
GOES-12 replaced GOES-8 as the operational GOES-East satellite,
beginning with the 18:20 UT (Atl) and 1846 (EUS) Sounder sectors.
As initially demonstrated during the
GOES-12 Science Test , GOES-12 remains noticeably much
cleaner than GOES-8 and more
consistent with GOES-10 (in the west), particularly in non-window channels.
01 Apr 2003: Regression based Sea Surface Temperature (SST) images,
derived from GOES-8 Imager radiances, were generated routinely
at CIMSS for many years and shown on the web since 1997. However, this
research demonstration ended on 1 April 2003, when GOES-12 replaced
GOES-8 as the operational GOES-East satellite. The regression relationship for
GOES-8 was based on use of the 11.5-12.5 micrometer "split window" channel
(band 5). GOES-12 does not have such a channel, but rather one at 13.3 micrometer,
invalidating continued demonstration of the "split window" application.
The
operational production of SST imagery from current GOES satellites
is now provided by NOAA/NESDIS/ORA/ORAD within the CoastWatch program.
26 Feb 2003: From both GOES-East and -West Imager observations, a cloud detection algorithm is used hourly to generate the Clear Sky Brightness Temperatures (CSBT) for all the Imager spectral bands. These cloud-free "clear sky" values, at about 50 km horizontal resolution, are provided to NCEP and ECMWF for assimilation into global numerical prediction models. A subset of this CSBT data is presented on the web in an image format, specifically showing the hemispheric coverage every three hours for the upper tropospheric moisture channel (#3) on the GOES Imager (at 6.7 um).
12 Aug 2002: Following the completion of support for two recent field campaigns (over the Southern US Plains during late spring 2002 for IHOP and over Florida during early summer 2002 for CRYSTAL), the focus for a more detailed, more mesoscale oriented display of GOES SFOV DPI was shifted to the "local" Wisconsin region. Java applets provide for easy animation and toggling of the DPI as well as overlays of first-guess fields and radiosonde values.
25 Apr 2002: In support of the IHOP 2002 spring field campaign in the south central US, CIMSS has implemented and continues to assess the generation of GOES Sounder retrievals and DPI at full spatial, or, also known as single field-of-view (SFOV), resolution. On the CIMSS Realtime GOES IHOP Page , SFOV DPI are shown over the last six hours, focussing on the Oklamoma, Kansas, and Texas Panhandle region.
In addition, simultaneous animation loops of the SFOV DPI are presented for larger scale views in the routine portion of the GOES DPI displays. The new SFOV retrievals are generated on the "vern" computer, using an AVN model first-guess; the 3x3 FOV retrievals are generated on the "suomi" computer, using an Eta model first-guess. (Currently, gif file displays of the SFOV DPI include an "A", for the guess type, as a distinguishing marker). In late March 2002, some of the SFOV DPI were beginning to be first posted on these web pages.
04 Dec 2001: A technique to identify snow during the daytime has been developed which uses the visible band and radiance information from the Geostationary Operational Environmental Satellite (GOES) Sounder shortwave bands. It was tested over the 2000/2001 winter and 2001 summer and, as of the nominal 19:00 UTC Cloud Product data set on 03 December 2001, has been incorporated into the current GOES Sounder cloud mask at the Cooperative Institute for Meteorological Satellite Studies (CIMSS).
An example of the cloud product using the routine (RTN) cloud masking algorithm versus the new (SNW) algorithm is shown for a late 2000/2001 winter case (at 19:46 UTC on 23 April 2001). The skies were indeed clear over the band extending from northwestern Nebraska through the eastern two-thirds of South Dakota and the extreme southeastern portion of North Dakota into western Minnesota. The improperly assigned cloud heights found in the RTN version of the Cloud Top Pressure (CTP) are in fact clear, snow-covered ground areas resulting from passage of the 22 April 2001 storm.
28 Aug 2001: Plots of recent severe weather reports over GOES Sounder Lifted Index (LI) or Precipitable Water (PW) DPI from the previous afternoon or evening have been added to provide a simple, quick look at the correlation that exists between the GOES derived atmospheric stability or moisture fields and later ensuing severe weather.
[The initial work on this display was done by NASA SHARP student C. Nosal (Pensacola, FL) who was sponsored by CIMSS staff (G.M. Bayler; G.S. Wade) during the summer of 2001.]
15 May 2001: Plotted differences between the GOES Sounder retrievals and their first-guess values have been expanded to include the 850 mb dewpoint, 850 mb temperature, and the 500 mb temperature values. These sample parameters and levels were selected due to their being primary ingredients of traditional thermodynamic atmospheric instability (that is, warm, moist air at low levels with cooler air aloft provides unstable atmospheric profiles). Note that the GOES temperature adjustments are generally rather minimal, especially over the US land mass where the forecast models do reasonably well with temperature; larger GOES temperature adjustments (1-2 C) are more frequent over oceanic regions. The relatively more robust moisture signal measured with the GOES Sounder is evident by the considerably larger dewpoint adjustments.
12 Apr 2001: The real-time display of cloud top pressures (CTP) from the GOES Sounder has been expanded to include added definition to the lowest layers, between 950 and 700 mb, as well as a second loop with more emphasis of the 48 conterminous US states. Confidence in the distinction of three layers of CTP in the lower portions of the troposphere is justified by the previous implementation, listed below, for handling maritime inversion situations.
19 Dec 2000: A new version of the cloud height subroutine (gopco2__plod.for) compensates for clouds in maritime inversion regimes. The height calculation requires a series of tests to determine if the cloud is high or low. This procedure compares the 11.0, 12.0, and 4.0 micrometer windows at night and the 11.0 and 12.0 micrometer windows during the day. Surface elevation and type are also considered.
This modified subroutine further requires the AVN model as a first guess (versus the Eta model). This new product is by no means perfect, but it certainly provides more realistic heights over the Eastern Pacific than in the past (950-850 hPa rather than 750-650 hPa).
This new technique was successfully implemented on an experimental basis, starting back in late September 2000, running in parallel to the old version (using the Eta as a first guess with no additional logic to determine cloud type). This new version was installed in the CIMSS routine processing on 19 December 2000 (and has been running smoothly ever since).
20 Nov 2000: Access to the text file (ascii) listings for the precipitable water and cloud top values from the last available hour has been added within the real-time PW and CTP sections.
02 Nov 2000: The display of three layers of, plus the total of, precipitable water (PW) from the GOES Sounder at the last available hour has now been included in the section on PW coverage of the US. Determination of the vertical distribution of moisture in otherwise similar areas of equal total PW amounts is now possible.
24 Oct 2000: The displays and animation of all 19 spectral bands of the GOES Sounder have now been expanded to include an option for GOES-10, covering the western US. Due to the geometric aspect ratio of the western US ("conus") sector for GOES-10 (being "tall" in latitude but "narrow" in longitude), the GOES-10 display shows imagery at 10 km horizontal resolution (versus 20 km for the more extensive coverage from GOES-8 in the east).
06 Oct 2000: Band 5 (13.4 um) has been added to the selections available as individual Sounder band composites of GOES-8 and -10. The capability to interactively toggle between any pair or more of the individual composites from the latest hour was also added.
05 Sep 2000: Animated image loops on the CIMSS Realtime GOES Page are now solely available as "java" animations. No longer are "fli" files being made. (Most of the products had already been provided in the more universal "java" format for over a year.)
30 Jun - 14 Aug 2000: During the NOAA Science Test period of the GOES-11 (L) check-out, Sounder imagery (all bands in 20 panel format) and products (PW, LI, CTP) from GOES-11 were displayed in near real-time. Besides typical animation loops of the last 24 hours, the GOES-11 data from the last hour were also shown compared to the same from the GOES-8 and -10 composite. GOES-11 was stationed at 104 West longitude. The last day remains as a sample of the CIMSS Realtime GOES-11 Page.
28 Jun 2000: Difference plots of the GOES Sounder retrieved values for PW and LI compared to their first-guess values are available for the last six hours,framed within the background of the appropriate DPI. Three sectors are available (North Central, SouthEast, and SouthWest). Java controls provide animation and overlay capability.
25 Apr 2000: The Total Column Ozone DPI is now displayed in a Mercator projection and uses ozone estimates from all four Sounder sectors. The format of the ozone "psuedo" DPI has been altered such that it is now identical to a true DPI produced with a retrieval program.
19 Apr 2000: The color enhancement table (GDCE505) now used for the CAPE DPI employs a smaller number of major distinctive colors (brown/beige>blue>yellow>red>magenta/purple), primarily switching at each 1000 J/kg level. Within each major color, shading on the order of about five levels is readily apparent. The previous enhancement table (GDCE500) had a few too many different colors, some of which seemed to "repeat", or were too similar, especially considering the typically large variations often seen in the CAPE DPI pixels.
7 Apr 2000: The localized "Wisconsin Cloud Product" display is no longer being produced. Large views of the cloud top pressure DPI remain.
25 Jan 2000: An interactive tool is now available for selecting a location where one wishes to see a plotted graph of the time series trend for the last 24 hours worth of GOES Sounder DPI values of total precipitable water (PW [mm]), lifted index (LI [degC]), and convectively available potential energy (CAPE [J/kg]).
The locations are chosen within a display of the most recent DPI imagery. Thin lines in the plot indicate periods of unavailable retrievals. This "java" capability was developed by T. Whittaker at CIMSS as part of the VISIT program.
7 Oct 1999: As of 17Z on October 7th Total Column Ozone (TCO) retrievals are using a new set of regression coefficients. Quality seems to have improved (with respect to TOMS TCO data).
31 Aug 1999: Total Column Ozone images are now "pseudo" DPIs instead of numbers plotted in color. "Java" animation loops have also been added for ozone.
27 Aug 1999: "Java" animation loops are now available for the DPI. (This should be helpful to those who have difficulty displaying the "fli" loop files.)
13 Aug 1999: As of 14 UTC, the cloud product enhancement has been changed. The low cloud category now includes 900-800-700 mb cloud top pressures as one color. The former version included three color levels for this range, the new version now consists of one.
The reason for the change is due to the limited vertical resolution of the first guess and the inability of the Window technique of the cloud top pressure algorithm to provide accurate information. An example of this shows that our cloud masking step confidently identifies whether a fov is cloudy or clear based on the visible imagery (upper left) and the SST analysis. The 11 micron window image (upper right) shows that the clouds are all very low (i.e. between the surface and 750mb).
The temperature/moisture profile (lower right) from the Vandenburg Air Force base at 12UTC shows that the cloud top is near 950mb. This is verified by PIREP observations for the same time and location. The cloud derived image (lower left) shows the cloud top pressure to be about 750mb.
08 Jul 1999: Following the 28 June 1999 loss of the machine ("oldthunder") responsible for much of the generation of displays for the realtime GOES product stream, transfer of all functions to another machine ("suomi") was completed. No longer are any "mpeg" animations being made (as overhead on their production is too costly); animations remain available in "fli" format.
10 Feb 1999: An initial display of CAPE (Convective Available Potential Energy) as a psuedo DPI has been added to the page, following the Lifted Index stability section. This CAPE DPI is generated, after the fact, from the 3x3 retrieval values. A preliminary enhancement table is being used.
Note that no "mpeg" animation of CAPE is being provided, as the overhead on mpeg generation is substantial and the phasing out of such files is being considered. Looping remains available in "fli" format, and more "java" animations will be implemented in the future.
28 Jan 1999: The precipitable water DPI enhancement table has been modified to the new proposed "consensus default" table. This table (GDPW1.ET) shows each 10mm block distinctly, with finer resolution to near 2mm by more subtle shading, while avoiding use of conflicting colors (as perceived by color-challenged users). For more information, comparisons, and details, see the GOES DPI ET Page. The change is effective as of 12 UTC.
05 Nov 1998: A Great Lakes Sector 12-hour animation of Sounder Bands 10/11 has been added; this product was requested by the NWS forecast office at Grand Rapids, MI, for evaluation during the 1998-1999 Lake Effect Snow season.
28 Oct 1998: The previous central/east US views of PW and LI DPI
(solely from GOES-8) have been replaced by a view of the "entire" conterminous
US (from composites of all GOES-8 and -10 sectors).
27 Oct 1998: For current status of the GOES satellites, a link was added to the NOAASIS page.
17 Sep 1998: Some text was added in the page (disclaimer, more GOES links, support acknowledgement).
04 Sep 1998: Band 15 (4.45um shortwave CO2) was added to suite of single band composites, for purpose of helping to monitor the radiometric quality of GOES-8 and GOES-10 Sounder data.
21 Jul 1998: Western coverage for GOES Sounder products was switched from GOES-9 to GOES-10. GOES-10 continues to drift westward towards its final destination of 135 W.
24 Jun 1998: The GOES Sounder Total Ozone Product was added; animations will soon follow...
18 May 1998: As of 18 UTC, the cloud product is available in two full resolution independent formats. The first
is cloud top pressure and the second is effective cloud amount.
In addition, an error in calculating effective cloud amount was corrected.
The old way (OW) versus new way (NW) computed too much overcast as opposed to broken or scattered.
The result for
effective cloud amount showed a significant increase in broken and scatter conditions.
As expected, the
cloud top pressure product was not affected.
24 Apr 1998: Hourly displays of all 19 spectral channels from the GOES-8 Sounder (over the central and eastern US) have been added.
20 Mar 1998: The CIMSS Realtime GOES Page resumed nominal operations, following the switching earlier in the week of the web serving machine (from "oldthunder" to "maple", with the alias remaining as "cimss"). Upon resumption of realtime displays, spatial composites from GOES-8 and GOES-9 of individual Sounder bands were added. The select bands were 3, 8, 11, 12, and 17.
08 Feb 1998: As of 03 UTC, the AVHRR SST which is used for the skin temperature value
over the ocean is back to its daily updating. Previously an old (97353) area was used.
The derived total precipitable water image shows that with the
new SST
has slightly less cloud than the
old SST.
Both versions correlate qualitatively
well with the corresponding observed sounder
IR window
and
visible image.
Overall, the TPW difference was less than a RMSE of 0.2 mm.
30 Jan 1998: As of 15 UTC, the GOES-8/9 derived cloud top pressure algorithm for the Sounder replaced the
RTTOVS transmittance model with the PLOD model. The changes between the new version (PLOD) and the old version (RTTOVS)
are small. On average the PLOD version is lower in pressure (higher in altitude) by about 15mb and 2% lower in
effective cloud amount than the old version for all levels of clouds. Examples of the
old version (RTTOVS)
and the
new version (PLOD) are shown.
13 Jan 1998: As of 21 UTC, the GOES-9 instrument noise is used for the GOES-9 soundings (not the GOES-8 noise).
Only very slight changes were noted.
The error was only in the CIMSS experimental version 1.75 retrievals.
05 Jan 1998: A script was enhanced to create the 23 UTC combined sounder water vapor channel-11 gif as well as the other 23 hours. (Basically, one needed to be subtracted from the day-of-year count, if the processing time was after 00 UTC.)
16 Dec 1997: One of the cloud-clearing tests was relaxed.
The test uses the 13.3 micron band.
11 Dec 1997: As of 15 UTC, the size of the processing box was reduced from a 5x5 field of view (fov) array to a 3x3 fov array.
This increases the product resolution from approximately 50km spacing to 30km spacing. Examples from GOES-8 for
total precipitable water and
cloud top pressures are given.
Radiosonde co-location stats for GOES-8 show the 3x3 retrievals are
slightly improved compared to the 5x5 retrievals.
21 Nov 1997: Additional checks were incorporated to limit any possible cloud contamination.
These include using the SST in the retrieval,
rejecting the retrieval if the cloud-cleared observed channel 8 (IR window) value is 1.5C colder than the calculated value from the first guess,
and flagging retrievals if the
SST background is not available. These changes affect less than 2% of the total number of retrievals.
For more information on these changes.
03 Nov 1997: The cloud clearing subroutine has been enhanced.
Ignoring the visible check over elevated mountainous terrain in the cloud detection algorithm has alleviated the problem of spurious clouds.
The cause is a relatively high surface albedo observed by the GOES sounder visible band.
For an example from GOES-9 before and after this correction.
18 Sep 1997: Switched to a full retrieval (version 1.75) for both the GOES-8 and -9 sounder products of Total Precipitable Water and Lifted Index. The image resolution is processed at a 5x5 field-of-view spacing. Version 1.75 is the PLOD transmittance model, plus: the zenith angle cut-off is increased from 60 to 62.5 degrees; we have turned off one of the cloud-clearing test to increase night-time coverage; and there's now a 2 degree guess temperature change limit due to the skin temperature. For more information.
23 Jul 1997: Changed ocean emissivity to 0.99 for cloud derived image to be consistent with temperature/moisture retrieval algorithm.
27 Jun 1997: PW and LI enhancements adjusted for moist unstable
summer conditions (GDPW15.ET, GDLI8.ET).
26 Jun 1997: ETA model replaced NGM model for use as first-guess.
09 Jun 1997: Adjust PW enhancement for summer (GDPW106.ET).
26 May -
03 Jun 1997: Sounder calibration was not fully correct.
19May 1997: Add displays of GOES Imgr Regression SST DPI.
15 May 1997: Add plot of radiosonde values to LI DPI.
05 May 1997: Add plot of radiosonde values to PW DPI at 00 and 12
UTC.
17 Apr 1997: Lake Effect coverage replaced by central US look.
09 Jan 1997: Adjust PW enhancement for dry winter (GDPW332.ET).
