| RAMMB CIRA 4th Quarter Report |
|---|
July August September 2008 |
Preliminary work has begun on the project entitled “Satellite Analysis of the Influence of the Gulf Stream on the Troposphere: Convective Response.” This project is a pilot study in which Advanced Microwave Sounder Unit (AMSU) temperature and wind retrievals, World Wide Lightning Locator Network (WWLLN), and GOES imagery will be used to document the influence of the Gulf Stream on the troposphere. A joint proposal was submitted by Dudley Chelton, CIOSS, Oregon State which has also been accepted. (A. Schumacher)
Using the temperature profiles from the COSMIC (Constellation Observing System for Meteorology, Ionosphere and Climate) satellites, plots of quasigeostrophic potential vorticity (q) can be made. Shown in Figure 1 is a plot of 300 hPa heights and q over the U.S. on 20 December 2006, as well as a cross section through the high q region over the four corners region. Both plots show air of stratospheric origin (q > 1.5) protruding into the troposphere. Such protrusions are often associated with midlatitude cyclones. With this particular system, Fort Collins, CO received 21” of snow. In order for the quasigeostropic potential vorticity to have the units of Ertel’s potential vorticity, the values were multiplied by a factor of –g(dθ/dp), where g is the acceleration of gravity, and θ is a reference-state potential temperature given by the U.S. Standard Atmosphere. Potential vorticity derived from COSMIC soundings will be combined with geostationary data to monitor upper-tropospheric structure, such as the stratospheric intrusion depicted here.
Figure 1.
Analysis has begun on data collected up through early September. RUC 00-hr analyses from summer 2008 are being compared with RAOBs at 26 locations across the central U.S., looking specifically at water vapor mixing ratios from the surface to 700 mb. The Figure 2 below shows the RUC mean mixing ratio errors for 00Z for all analyzed raob locations. A moist bias exists near 925 mb, while a dry bias can be seen at 700 mb. Multiple regression analysis shows that using information from surface dew point observations can correct for these biases and improve the overall absolute error. The next step is to determine how GOES data can further improve the boundary layer moisture analyses. (D. Lindsey)
Figure 2. RUC 00-hr analysis mean water vapor mixing ratio error (g/kg) from 26 raob locations across the central U.S. from the summer of 2008.
Imagery for the Okmok (Alaska Aleutian) volcano eruption from 12/13 July 2008 has been analyzed thru Principal Component Image (PCI) analysis. PCIs, which extract dominant image combinations from the available GOES bands, are enhanced with RGB (Red, Green, Blue: 3-color) analysis to better show the associated clouds and ash in the images. See Figures 3 and 4 for an explanation. Animated loops of these PCI and RGB products are also available. (D. Hillger)
Figure 3: PCI analysis presented in a 4-panel display: The upper-left panel is dominated by the LWIR composite (high clouds are white); the upper-right panel is dominated by the visible band; the lower-left panel is dominated by the SWIR band, showing low (white) vs. high (darker) cloud; and the lower-right panel is dominated by the split-window difference, showing mainly thin cloud and ash (dark areas). PCI-4, which is dominated by the WV band, is not shown.
Figure 4: PCIs from Figure 3 are combined in this image using RGB (3-color) analysis. The colors were chosen to enhance the ash cloud, with PCI-3, 2, and 5 as Red, Green, and Blue, respectively. Clear areas in the image are deep purple, high clouds are mainly green, lower clouds are yellow, and heavily-ash-dominated cloud is orange. Note the higher concentration of ash in the plume south of the volcano vs. the plume east of the volcano.
In monitoring recent forest fires from GOES imagery, it was noticed that one of the overnight band-2 (3.9 μm) images showed a great increase in fire signature from the previous image a half-hour earlier, and that the signature for same fire decreased as rapidly as it increased in the image a half-hour later. The fire flare-up occurred at 0800 UTC on 30 July 2008 for the “Rich Wildland Fire” in Northern California. The flare-up (and flare-down) is emphasized in the attached figure of the shortwave-albedo product, which is useful for enhancing imagery for the detection of fire hot spots. Note the large increase in the number of fire-affected pixels, covering an area about 15 pixels North-South, by about 31 pixels East-West. (The larger East-West than North-South extent is due to the approximate 2-to-1 oversampling of GOES data in the along-scan direction.) However, the otherwise large extent of the fire signature is unlikely to be due to a rapid increase (and decrease) in the fire size alone, but must be either caused by, or accompanied by, a large increase (and decrease) in the fire temperature as well, even though only a very few (2-4) center pixels reach saturation temperature (340 K) in the images. For this fire, the temperature increased (and again decreased) by an average of about 1 K for all the pixels in the larger fire-affected area. (D. Hillger)
The especially large area affected by the fire is speculated to be due to the broad GOES point spread function, in which any detected signal may affect pixels several fields-of-view away from the actual signal. This effect is undetectable when the entire field is homogenous, is minimal when the field is fairly flat, but will be more obvious when the field is highly variable. And, the effect will be most pronounced when the signal is a fire hot spot and the band in question is in the shortwave spectrum, for the same reason that fires are more easily detected in the shortwave spectrum than in the longwave spectrum. Further analysis of this case will be undertaken to try to resolve the cause of the rapid appearance and equally-rapid disappearance of the unusually large area that is affected by the fire. Images at 15-minute intervals will be added to the analysis, and model simulations will be made of the effect of extreme fire temperatures on GOES imagery, utilizing the known GOES point spread function. It’s possible that this far-field-of-view effect, although thought to be an undesired artifact, may lead to a way to estimate extreme fire temperatures that are otherwise undetectable, being far hotter than the saturation temperature of band-2 data. (D. Hillger, L. Grasso, M. Sengupta, R. Brummer)
Figure 5: An explosive fire flare-up as manifest in the shortwave-albedo product which enhances the band-2 (3.9 μm) vs. band-4 (10.7 μm) temperatures. In these half-hourly images, note the large area of dark pixels surrounding the few saturated (white) fire pixels in the center image of the top row. The pixels in the large fire-affected area average 1 K warmer in the band-2 images, an effect that disappeared as rapidly as it appeared.
Processing of the large sector U.S. climatologies continues. Products completed include monthly large sector composites for March. Processing has been slow due to illness. (C. Combs)
Processing of wind regime products continues. Monthly wind regime composites from both channel 1 and channel 4 for February and March 2008 have been completed. Combined monthly products have also been completed for February and March 2008. (C. Combs)
Preprocessing of GOES west data over the Eureka area continues. Processing and quality control for hours not normally processed but in our DVD archive has been completed for May-Sep 1999. Data is being pulled from the CIRA archive, which slows the processing down with additional steps. (C. Combs)
Work continues with Becca Mazur, Treena Hartley and Mel Nordquist from the Eureka, CA National Weather Service office. With help from Don Hillger, we have converted sample GOES 11 images (channels 1,2 and 4) in the Eureka sector to GeoTIFFs. We sent GeoTIFFs to Eureka to see if they can be read into their GIS system. (C. Combs)
A new collaboration with Declan Cannon, a forecaster from the Duluth, MN National Weather Service (NWS) office has been started. After email communications, three sample Wind Regime climatologies over Duluth for July 1998-2007 were sent. Due to continued interest, the next step will be to set up a website to display loops of wind regime composites. (C. Combs)
Powerpoint proposal presentations were updated, one each for GIMPAP and GOES-R, to review work during FY08. These were presented for GIMPAP on August 5th, and for GOES-R on September 3rd. (C. Combs)
MSG data for January 2007 were obtained from an archive and sectored out Europe. Developed and tested code that identifies snow pixels in an image, blending techniques from MODIS and SEVIRI papers using channels 0.64 μm, 0.8 μm, and 1.6 μm. Using this information, wrote and tested code to created monthly visible backgrounds and cloud cover % composites. Compared new code results with results from code currently used in the regional cloud climatologies over the US for the current GOES satellites. Preliminary results look favorable. (C. Combs)
Two new hourly employees were hired. The first one worked through the summer before heading for college, and the second one started end of August. Both worked on the Eureka data sets
.
Figure 6: Cloud cover percent for Jan 2007 at 1115 UTC, using method currently used in regional cloud climatologies over continental US. Note high cloud percent over Alps and northern Spain.
Figure 7: Cloud cover percent for Jan 2007 at 1115 UTC, using new method that identifies snow covered pixels using MSG channels 1.6 μm and 0.64 μm. Note the cloud percent is reduced over the Alps and northern Spain.