The Split Window Difference as a measurement of Atmospheric Moisture

April 7th, 2017 |

GOES-16 Split Window Difference (10.33 µm – 12.30 µm) with 850-mb Dewpoint Temperatures from the Rapid Refresh overlain (Click to enlarge)

GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.

GOES-16 includes both a clean infrared window (10.33 µm) and a so-called ‘dirty’ infrared window channel (12.30 µm). The clean infrared window is in a part of the electromagnetic spectrum where there is very little absorption of energy by water vapor; in the dirty infrared window, modest amounts of water vapor absorption occur. The brightness temperature difference, nicknamed the Split Window Difference (SWD for short), can highlight differences in moisture in clear skies.

The toggle above shows the SWD (10.33 µm – 12.30 µm) at 1430 UTC on 7 April 2017. A pronounced gradient stretches southeast to northwest from Louisiana to northeast Kansas and extreme southeastern Nebraska.  Values over Missouri, for example, are around 0.9-1.0 K vs. 1.7-2.2 K over Oklahoma.  The gradient in the brightness temperature difference aligns very neatly with the 850-mb dewpoint temperature from the Rapid Refresh. You can use this product to monitor moisture return from the Gulf of Mexico.

AWIPS Note: The Default enhancement in AWIPS for the Split Window Difference, shown above, does not include large enough negative values. The Split Window Difference value can exceed -5 K in regions of dust. See this link for a different enhancement for this case with a wider range of temperature differences. A similar image uses the mean 1000-700 mb dewpoint temperature rather than values from the single 850-mb level.

=============== Added ===============

An animation of this imagery (not shown) shows general increases in the SWD values with time.  A consistent signal of moisture will be present only if the temperature decreases with height in the moist layer (that is — if there is no inversion).  An increase in the SWD does not necessarily show an increase in moisture — it can, rather, signify an increase in near-surface temperature (for more information, consult this article by Lindsey et al.). The gradient in the field can remain, however, as in this example.

The Split Window Difference field does an exemplary job of detecting contrails over the southern Plains. The toggle below shows that the SWD signal of cirrus is more distinct than in the 1.378 µm Cirrus Channel! (Thanks to Matt Bunkers of WFO Rapid City for noting this!)

Cirrus Channel (1.378 µm ) and Split Window Difference (10.33 µm – 12.30 µm) at 1607 UTC on 7 April 2017 (Click to enlarge)

(Note that the SWD was something that was available from GOES-8 through GOES-11. Link)

NOAA/CIMSS ProbTor with a Severe Thunderstorm over Florida

April 6th, 2017 |

GOES-16 “Red” Band (0.64 µm) from 1222 through 1517 UTC on 6 April 2017 (Click to animate)

GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.

GOES-16 Visible Imagery early on 6 April 2017 showed a strong thunderstorm developing north of Lake Okeechobee and proceeding east towards the Atlantic Coast near Vero Beach (1-minute or 30-second imagery was not available over Florida because the GOES-16 Mesoscale Sectors were in their default locations; routine 5-minute CONUS imagery is shown above).

Clean Infrared  Window channel (10.33 µm) imagery from 1400-1500 UTC, taken from AWIPS, below, shows a well-developed albeit weakening storm that is moving off the coast. A pronounced Overshooting Top/Thermal Couplet is present at 1402 UTC; the brightness temperature of the overshoot is -77º C, and the downwind warm trench is -62º C.

GOES-16 Clean Infrared Window Band (10.3 µm) from 1402 through 1457 UTC on 6 April 2017 (Click to animate)

The NOAA/CIMSS ProbTor product was monitoring this storm as it moved eastward through central Florida. The animation below shows the cell strengthening rapidly after 1310 UTC, and maintaining large ProbTor values for about an hour, after which time values collapsed.

NOAA/CIMSS ProbTor product from 1256 through 1444 UTC on 6 April 2017 (Click to animate)

The time series below shows ProbTor as a function of time (1232-1444 UTC). The different parameters that are used in the computation of ProbTor are plotted as well. The times of the Severe Weather Warnings issued by the National Weather Service are drawn along the horizontal axis. The three wind events noted as vertical magenta lines (wind events taken from the Storm Prediction Center Storm Reports) occur within the envelope of highest ProbTor probabilities.

Plot of NOAA/CIMSS ProbTor values from 1232 through 1444 UTC on 6 April 2017. Vertical magenta lines are wind damage reports from SPC (Click to animate)

Note: NOAA/CIMSS ProbSevere Products — ProbHail, ProbWind, ProbTornado and the 2016 version of ProbSevere are all run using legacy GOES data (GOES-13 and GOES-15). GOES-16 data can be incorporated into this tool only after the statistical model has been trained on GOES-16 data, and that has not yet happened; A GOES-16 version is planned for the 2018 convective season.

With 16 Channels, the ABI on GOES-16 allows many different views of the same event. The animation below includes the 1/2-km resolution 0.64 µm “Red” Visible channel, the Snow/Ice channel (1.61 µm, 1-km resolution) that distinguished between clouds made of water droplets and clouds made of ice because ice strongly absorbs radiation at 1.61 µm and hence reflectances are smaller and the cloud appears less white, the Cirrus channel (1.378 µm, 2-km resolution) that highlights high clouds because radiation at the wavelength is very strongly absorbed by water vapor so reflectance from low-level clouds cannot escape to space before it is absorbed, and the Clean Infrared Window channel (10.33 µm, 2-km resolution) that affords a view that is least affected by water vapor absoption. In the animation below, the coastline does not become visible in the Cirrus channel because of strong absorption of radiation at that wavelength by water vapor. The Cirrus clouds in the 1.61 µm channel are considerably darker because of absorption by ice of radiation at that wavelength; a few water-based clouds do appear in the scene and are bright: 1.61 µm radiation is reflected quite well by water droplets.

GOES-16 “Red” Visible Band (0.64 µm), Snow/Ice Band (1.61 µm), Cirrus Band (1.378 µm) and Clean Infrared Window (10.33 µm) from 1222 through 1517 UTC on 6 April 2017 (Click to animate)

The GOES-R website has Quick Guide information on the Red Visible Band (0.64 µm), the Snow/Ice Band (1.61 µm), the Cirrus Band (1.378 µm) and the Clean Infrared Window Band  (10.33 µm) at this site.

Severe Weather over the southern United States

April 5th, 2017 |

GOES-16 “Red” Band (0.64 µm) from 1900 through 2100 UTC on 5 April 2017 (Click to animate)

GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.

On 5 April 2017, The Storm Prediction Center in Norman OK issued a Day 1 Convective Outlook that included Moderate and High Risk areas over much of the southeastern United States. Mesoscale discussion 442 and 448 on 5 April discuss the area shown above. One of two GOES-16 Mesoscale Sectors on 5 April viewed this scene, and a 2-hour animation spanning the outbreak of convection is shown above. It is difficult to predict which particular towering cumulus is going to grow into a cumulonimbus based on the visible imagery alone.

GOES-16 Low-Level Water Vapor Band (7.34 µm) from 2002 through 2127 UTC on 5 April 2017 (Click to animate)

Water Vapor Imagery from GOES-16 might help in that prediction. The animation above, from 2000 UTC to 2130 UTC, shows multiple subtle gradients in the low-level water vapor field that could be associated with impulses influencing convective initiation. Convection appears to form along those subtle gradients. The 16 channels on ABI offer far more information than legacy GOES.


=========================================================

CIMSS Scientists have been refining the NOAA/CIMSS ProbSevere product to account for individual threats (Hail, Wind, Tornado), and a screen capture of that product is shown below. Radar Objects are outlined in colors that relate to the Probability of a severe event. These outlines allow a forecaster to determine which cell is potentially most threatening based on the inputs into the determination of probability.

NOAA/CIMSS ProbSevere for All Hazards at 2130 UTC on 5 April (click to enlarge)

At 2130 UTC, shown above, a cell over Tennessee has the highest ProbTor probability; the readout (below) shows the variables that are used in computing all the hazard probabilities — ProbHail, ProbWind, and ProbTor (Not all variables are used for all products; for example, ProbTor does not use Satellite-derived growth rates; ‘PS‘ in the output is the value of the 2016 version of ProbSevere). ProbTor of 70% for the cell over Tennessee (compared to smaller values at nearby cells) suggests that particular cell deserves special attention from anyone monitoring the cell for development. The ProbTor for the warned cell in Alabama — a cell that produced Tennis Ball-sized hail near Heflin , showed a ProbTor value at this time of 43% — a larger value than for the storms on either side of it — again suggestive that it should be of most concern. Ten minutes later, at 2140 UTC, ProbTor for the Tennessee storm had dropped to 34%, and that of the Alabama storm had increased to 55%.

ProbTor will be one of the CIMSS products demonstrated at the Hazardous Weather Testbed in June and July this year.

NOAA/CIMSS ProbSevere for All Hazards at 2130 UTC on 5 April; The readout for the indicated cell is shown (Click to enlarge)

Note: ProbSevere products are computed using legacy GOES imagery only. GOES-16 data can be incorporated into this tool only after the statistical model has been trained on GOES-16 data, and that has not yet happened; A GOES-16 version is planned for the 2018 convective season.

GOES-16 Cirrus Channel and Dust

March 23rd, 2017 |

GOES-16 Visible (0.64 µm) images, 2132-2232 UTC on 23 March [click to play animated gif]

GOES-16 Visible (0.64 µm) images, 2132-2232 UTC on 23 March [click to play animated gif]

GOES-16 data posted on this page are preliminary, non-operational data that are undergoing testing.

The visible animation from late afternoon over west Texas, above, shows a characteristic signature of a shroud of dust around El Paso, TX behind a dryline associated with a developing cyclone in the lee of the Rocky Mountains. This pall of dust was visible in many of the 16 channels on the Advanced Baseline Imager (ABI) that sits on GOES-16. The toggle below cycles through the Red visible (0.64 µm), the Blue visible (0.47 µm), the Cirrus channel (1.38 µm), the Snow/ice channel (1.61 µm) and the Upper-Level and Lower-Level water vapor channels (6.19 µm and 7.34 µm, respectively) (Click here for a faster image toggle). In  addition, a 2-panel comparison of GOES-16 Visible and Cirrus band imagery is available here.

GOES-16 Visible (0.64 µm and 0.47 µm), Cirrus (1.38 µm), Snow/Ice (1.61 µm), Upper level Water Vapor (6.19 µm) and Lower Level Water Vapor (7.34 µm) images, 2132 UTC on 23 March [click to enlarge]

GOES-16 Visible (0.64 µm and 0.47 µm), Cirrus (1.38 µm), Snow/Ice (1.61 µm), Upper level Water Vapor (6.19 µm) and Lower Level Water Vapor (7.34 µm) images, 2132 UTC on 23 March [click to enlarge]

Several aspects of the toggle above bear comment. Note that the blue channel (0.47 µm) has in general a ‘hazier’ appearance than the 0.64 µm red channel. Atmospheric scattering is more important at shorter wavelengths, and that is picked up by the satellite. The 1.38 µm ‘Cirrus’ Channel generally does not see the surface because of water vapor absorption at that wavelength. However, the atmosphere behind the dry line is sufficiently parched (total Precipitable Water in the El Paso sounding on 0000 UTC 24 March is less than 6 mm; sounding from this site) that complete attenuation by water vapor is not occurring; dust is highly reflective at 1.38 µm and a signal becomes apparent in the dry air from west Texas southwestward into central Mexico.

Thin dust is very difficult to detect in the 1.61 µm snow/ice channel because solar energy at that wavelength reflected from the surface moves unimpeded through thin dust; thus you can generally see the surface in dusty regions in the 1.61 µm channel. On this date the 1.61 µm channel nimbly discriminated between water clouds (over central Mexico) and ice clouds (over much of the rest of the domain, as shown in this toggle between 0.64 µm and 1.61 µm : only the clouds composed of water are reflective (white) in both channels.

The atmosphere was sufficiently dry on this date that the lower-level (7.34 µm) water vapor channel detected surface features (horizontal convective rolls) associated with the blowing dust. (click here for the 6.19 µm image; surface features are not so apparent). Weighting functions computed at those wavelengths show a significant contribution from the surface at 7.4 µm (the red line), and also at 7.0 µm, (the green line), so the mid-level water vapor imagery from GOES-16 likely also shows surface influences); the 6.5 µm weighting function (the blue line) does not extend to the surface (These GOES-13 Sounder Weighting Functions that are similar to those from the GOES-16 ABI are from this site) so it’s unlikely that the 6.19 µm imagery shows surface features.

The GOES-R Website has fact sheets on the 0.47 µm, 0.64 µm, 1.38 µm, 1.61 µm, 6.19 µm and 7.34 µm channels.

Added: The RAMSDIS GOES-16 Loop of the Day from 23 March showed the Dust RGB product.