Daytime fog over snow in Wyoming on 21 November 2018

By Ed Szoke and Dan Bikos

On 21 November 2018, here is the GOES-16 visible (0.64 micron) imagery:

Visible loop:
click here or on the image for the animation

Do you see any fog in this imagery?

How about in this product, do you see any fog?

click here or on the image for the animation

This is the experimental CIRA Snow/Cloud Layers product. The purpose of the product is to distinguish clouds from snow cover on the ground during the daytime hours. In the product, low/water clouds generally appear as yellow or possibly greenish (as in this case) while high/ice clouds appear as magenta. Snow on the ground will appear white. This product uses a number of GOES-16 ABI bands (0.47, 0.64, 1.37, 1.6, 2.24, and 10.3 microns) as a RGB product with additional calculations. Note in our loop over central Wyoming we see a region that is green/brown in color over the white snow field that is decreasing in areal extent over time, indicating low cloud or fog. This product is currently on the RAMMB SLIDER page, however will be experimentally available on AWIPS via LDM from CIRA.

A product that is currently available in AWIPS is the Day Cloud Phase Distinction RGB which is shown here:

click here or on the image for the animation

in this product, snow on the ground appears green while low cloud or fog will appears lavender to cyan in color during the daytime. The reason is that we have relatively large contribution from the blue component with liquid cloud, small contribution from red component and large contribution from green component since it’s reflective in the visible band. The combination contributes to the lavender color for the low cloud / fog.

For this case, the METAR site at Riverton, WY reported fog as seen in this METAR plot at 1600 UTC:

A list of the observations from Riverton during the morning are shown here:

KRIW 211853Z AUTO 00000KT 6SM BR CLR M07/M09 A3016 RMK AO2 SLP279 I1001 T10721089
KRIW 211753Z AUTO 00000KT 6SM BR SCT002 M09/M10 A3018 RMK AO2 SLP291 I6001 T10941100 11094 21161 58001 $
KRIW 211727Z AUTO 00000KT 5SM BR FEW002 M11/M11 A3018 RMK AO2 T11061106 $
KRIW 211717Z AUTO 19004KT 4SM BR BKN002 M11/M11 A3019 RMK AO2 T11061111 $
KRIW 211707Z AUTO 00000KT 9SM SCT002 M10/M11 A3018 RMK AO2 T11001106 $
KRIW 211653Z AUTO 00000KT 8SM BKN001 M11/M11 A3019 RMK AO2 SLP293 I1001 T11061111 $
KRIW 211639Z AUTO 00000KT 3SM BR OVC001 M11/M12 A3019 RMK AO2 T11061122 $
KRIW 211633Z AUTO 00000KT 1SM BR OVC001 M11/M12 A3019 RMK AO2 T11111122 $
KRIW 211553Z AUTO 03003KT 1/4SM FZFG VV001 M13/M15 A3019 RMK AO2 SLP303 T11331150 $
KRIW 211543Z AUTO 00000KT 1/4SM FZFG VV001 M13/M14 A3019 RMK AO2 T11331144 $

Posted in GOES Low Cloud / Fog Imagery | Leave a comment

Dry…but not THAT dry!

The GOES-16 water vapor imagery for all 3 channels showed a narrow band of very warm brightness temperatures (implying sinking air and a dry atmosphere) on Friday morning (1502 UTC) 9 Feb 2018.

GOES-16 6.9 micron (“mid-level”) water vapor image at 1502 UTC/9 Feb 2018.

GOES-16 6.2 micron water vapor image (“high level”) at 1502 UTC/9 Feb 2018.

GOES-16 7.3 micron water vapor image (“low level”) at 1502 UTC/9 Feb 2018.

This narrow zone of sinking air is on the southern (anticyclonic) side of a very strong upper level jet draped across the CONUS, seen in the 1200 UTC 300 mb analysis below.  The dry slot stretches back into the Pacific south of the CONUS jet.

Notice how the brightness temperatures within the narrow zone are quite distinct, being at the very warm end of the scale (for this time of year).  So does seeing these warm brightness temperatures at all 3 water vapor channels indicate a lack of moisture through the column.  Well, not necessarily.  Check out the surface plots below at 1200 UTC and 1500 UTC on 9 Feb, with corresponding GeoColor imagery for these same times.  Bluish clouds in the GeoColor imagery indicate water clouds, which goes along with the low overcast conditions shown in the METAR plot.  By 1500 UTC the GeoColor nighttime imagery has morphed into visible imagery, verifying the low clouds streaming north from the Gulf of Mexico.  GeoColor imagery is not currently a baseline product, but CIRA can provide it to your WFO if you would like to try it.

Surface (METAR) plot at 12 UTC on 9 Feb 2018. Note the overcast conditions extending across northeast TX into southeast OK.


GOES-16 GeoColor image at 1202 UTC on 9 Feb. Lower level (water) clouds are bluish, while ice clouds are white. City lights are shown during nighttime imagery.


Surface (METAR) plot at 15 UTC on 9 Feb 2018.


GeoColor image at 1502 UTC 9 Feb.


The satellite imagery and METAR plots indicate relatively deep low level moisture that extends northwards across a portion of the very warm/dry slot shown in the water vapor imagery.  The 1200 UTC 9 Feb sounding from Dallas (DFW) in northeastern TX confirms the low level moisture (note that DFW at 1200 UTC reported scattered clouds at 1800 ft (AGL) and a 3000 ft solid cloud deck).

Dallas sounding at 1200 UTC on 9 Feb.


The moist layer extends from the surface to just above 900 mb.  Why is it that this moisture is not detected by the water vapor bands, even the lowest one?  The answer lies in what the weighting functions look like for the different water vapor channels for this sounding.  You can see what these look like on the real-time CIMSS site at

For the Dallas sounding shown above the weighting functions are shown in the plot below for the 3 water vapor channels.


The high level water vapor channel has a single weighting function peak around 400 mb, while the other two channels have a dual peak.  Even though the sounding showed that conditions were quite dry above the low level moist layer, there is just enough moisture to saturate the other two channels before reaching this low level moist layer aob 900 mb.  The low-level water vapor image (7.3 micron band) appears to have a small contribution nearly to the surface, and indeed, if you look closely at that image (above) you do see a slight difference in the color within the warm/dry band in northeastern Texas.  But for the most part the main message from the water vapor imagery would be a lack of moisture, yet clearly there is a low-level saturated layer.  This case indicates that interpreting water vapor imagery in regards to the amount of moisture present is not straightforward.  An earlier blog (go here to see the blog) showed an even more extreme case near San Diego using GOES-15 Sounder imagery (basically the same 3 water vapor channels that are on GOES-16, but at far worse horizontal resolution (10-km; and temporal resolution of course).  In the San Diego case all 3 channels suggested dry conditions, but in fact the San Diego sounding had a record level of Precipitable Water (PW) for the time of year, but the moisture, while deeper than in this case, was all present below where the 3 channels saturated.  This again could be seen using the weighting function profiles.

Bottom line – caution must be used when inferring a moisture profile using water vapor imagery, and the imagery does not indicate the actual value of moisture that may be present.  So is there a to determine how much moisture is present in the atmosphere (besides the raob)?  Indeed there is such a product, known as the Advected Layered Precipitable Water (ALPW) product.  The ALPW product uses data from Polar orbiting satellites that have instrumentation that detects the amount of moisture (PW) in the atmosphere, with the ability to also see through clouds (which saturate water vapor imagery, unless of course we have a case like this one where the channels saturate before reaching the level of the clouds).  The advantage of looking at the ALPW product versus looking at ABI water vapor bands is that it provides a quantitative value for the moisture in a given layer without the need to think about weighting function profiles or other complications (e.g., zenith angle) that affect your interpretation of assessing moisture from ABI water vapor bands alone.  The cursor readout function on AWIPS can be used with the ALPW imagery to get read-outs of the PW values for each layer.  Forecasters are familiar with the Total PW product on AWIPS, but more recently CIRA has developed the ALPW product, which is an “advected” version of the LPW product where the moisture present in 4 different layers is displayed.  The ALPW product for 1500 UTC on 9 Feb is shown below (at times there are missing swaths in the imagery and this was the case over the area of interest at 1200 UTC, so 1500 UTC imagery is used).  This image is from AWIPS2, using the 4-panel display to see each individual layer in one image.  The imagery is available at 3-h intervals, and real-time imagery can be seen at .   Training for this product is available at the VISIT site at

The ALPW image above depicts very dry conditions in the upper two layers (aob 700 mb), consistent with the sounding at Dallas and the water vapor imagery (which had the most contribution in these layers per the weighting functions).  But the ALPW does show a nice moisture plume extending north/north-northeast into northeastern Texas, southeastern Oklahoma and into Arkansas in the lowest layer (surface to 850 mb).  In the next higher layer most of this moisture is just making it towards the Dallas areas.  In practice, the most information can come from using the GOES-16 water vapor imagery, with its much higher time and space resolution, in concert with the ALPW imagery.  The ALPW imagery is not currently an AWIPS baseline product, but if you would like to try it at your WFO contact Dan Bikos or myself and it can be delivered to AWIPS over the LDM in real-time.



Posted in GeoColor Imagery | Tagged | Leave a comment

Lots of smoke moving south – a comparison of GeoColor with other bands for smoke visualization on 1 Aug 2017

The GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.  Users bear all responsibility for inspecting the data prior to use and for the manner in which the data are utilized.

Fires have been burning for some time in Montana as well as western Canada, as seen in the latest large fire incident map for 1 August 2017 below.



A broad upper-level ridge over the southwest to south-central CONUS has generally kept the smoke on an eastward trajectory.


500 mb plot and analysis from 1200 UTC on 26 July 2017.

A significant change to the pattern is occurring this week with a retrogression of the upper level ridge towards the West Coast (coincident with the very hot weather across the Pacific NW), allowing for shortwave troughs moving out of Canada to deepen into the central and eastern CONUS.  This has changed the flow across the High Plains southward through the eastern Rockies from west-southwest to north/northeast, as seen in the 500 mb analysis from this morning (1200 UTC/1 August, below).

500_170801_12 500 mb plot and analysis from 1200 UTC on 1 August 2017.

Smoke from fires can be a health hazard and if thick enough may negatively influence the development of convection, so tracking and monitoring the smoke is a concern for forecasters.   In the pre-GOES-16 era we know that smoke is typically most apparent on visible imagery near sunrise or sunset (as well as for an observer on the ground).  There are many bands and products available now with GOES-16, and here we take a look at a few of these from early (1317 UTC on 1 August) this morning to see the differences in the appearance of smoke.

First we look at a type of imagery developed at CIRA known as GeoColor.  Using a layering technique it combines 0.64 µm (Band 2) visible imagery with a “True Color” background during the daytime, and 10.35 µm (Band 10) IR imagery (along with 10.35-3.9 µm imagery to highlight fog and low clouds) with a static image of nighttime lights during the night.  This allows for a seamless transition from day to night when viewing a loop of the imagery.  Unique to GeoColor is the True Color background, which without a special algorithm developed using Himawari imagery would not be possible, since GOES-16 does not have a green band.  GeoColor creates a synthetic green band and by using this is able to make a very realistic looking image of the daytime surface, similar to what one would see if on the International Space Station.  GeoColor is available for display in AWIPS-2 through the LDM, currently at a reduced temporal resolution of 15 minutes.  If interested in the product send an email to myself ( or Dan Lindsey (  Dan presented a recent FDTD GOES-16 Applications Webinar (also known as a VISIT Satellite Chat) on GeoColor and some of its applications (the powerpoint can be downloaded and a recording viewed at ).  You can also view real-time GeoColor imagery, as well as all the GOES-16 bands and some RGBs, online at ).  The GeoColor image from earlier today at 1317 UTC (1 August) is shown below.



GOES-16 GeoColor image at 1317 UTC on 1 August 2017.

We see both the nighttime and daytime version of the GeoColor imagery in the image at 1317 UTC, with nightlights visible for some of the cities from Utah and Arizona westward.  In the daytime portion of the imagery the smoke is nicely seen extending from northeastern Colorado northwards to Montana and then east across the Northern Plains.  Now let’s contrast this image with the traditional visible imagery (Band 2, and 0.64 µm) from GOES-16 for the same time.



GOES-16 Band 2 (visible 0.64 µm) image at 1317 UTC on 1 August 2017.

Some of the thicker smoke is visible across the northern portion of the image from Montana to North Dakota, but most of the smoke is not easily seen.  Smoke can typically be seen better in the visible imagery at 0.47 µm (Band 1 on GOES-16) because it has higher reflectance in the presence of atmospheric aerosols compared to Band 2.


GOES-16 Band 1 (visible 0.47 µm) image at 1317 UTC on 1 August 2017.

For this case we do not see much difference between Band 2 and for both images the extent of the smoke is hard to determine compared to the GeoColor image.  An RGB image developed by CIMSS known as Natural Color (available in AWIPS-2) attempts to replicate True Color without using the complexities involved in the GeoColor imagery.  As seen in the Natural Color image below, while the background colors show many features of terrain and vegetation, they are not as “true” a representation as one gets with GeoColor.  However, often the colors of the background image make it easier to discern smoke and dust when compared to the visible imagery bands 1 and 2.


GOES-16 Natural Color RGB image at 1317 UTC on 1 August 2017.

For this case some of the smoke is easier to see across the northern portion of the image, but again we do not have the extent of smoke (especially thinner smoke layers) that were seen in the GeoColor image.

An interesting forecast product that is being run experimentally at NOAA/ESRL/GSD is a version of the HRRR with a chemistry formulation that is initialized for fires at this time using information from VIIRS.  The HRRR-Smoke model is run in non-operational mode four times per day out to 36-h and output can be found at .  A forecast from the 00z/1 August run valid at 13 UTC is shown next.

trc1_int_f13 HRRR-smoke 13-h forecast of vertically integrated smoke valid at 13 UTC on 1 August.

The southern extent of the thinner smoke layer across eastern Colorado in the forecast is in good agreement with the GeoColor imagery shown earlier.  The forecast from the same run out to 36-h, shown below, indicates the smoke is expected to continue moving southward into the Central and Southern Plains.


HRRR-smoke 36-h forecast of vertically integrated smoke valid at 12 UTC on 2 August.

Finally, a look at the extent of the fires using imagery that highlights the fire hot spots. First a look at the traditional 3.9 µm (Band 7) imagery from GOES-16, which of course has greater resolution (2 km) and can resolve higher temperatures than imagery from the previous GOES.  An image from this afternoon is shown next.


GOES-16 Band 7 (3.9 µm) image at 2232 UTC on 1 August 2017.

A simple grey color table is used in this imagery, so all fires appear as white dots, which can be seen at numerous spots from central Idaho through western Montana, with another fire in far northeastern Washington and another across the Canadian border.  Other color tables are available that discriminate some of the temperatures that can be seen with the GOES-16 39 micron imagery.  See, for example, the images posted by Bill Line in his blog on the Southern Plains fires on 6 March 2017 at   Also within that blog Bill shows how fire hot spots can also appear in the near-IR 2.25 and 1.61 µm bands (Bands 5 and 6).  CIRA has developed a Fire Temperature RGB that combines information from the three shortwave IR bands (ABI Bands 5, 6 and 7) to provide information on fire intensity.  In this RGB composite, the red component is Band 7 (3.9 µm), the green component is Band 6 (2.25 µm) and the blue component is Band 5 (1.6 µm).  As a general rule, the more intense a fire is burning, the more radiation it emits at shorter wavelengths.  Band 7 is capable of detecting most fires.  Less intense fires will only be detected in Band 7 and appear bright red.  Moderately intense fires will be detected by both Band 7 and Band 6 and appear orange to yellow, depending on intensity.  The most intense fires will be detected by all three bands and appear white.  Note that the range detected by this RGB will exceed the temperature range from the Band 7/3.9 µm imagery.  The Fire Temperature RGB image for the same time this afternoon is shown next.



GOES-16 Fire Temperature RGB image at 2232 UTC on 1 August 2017.

The same fires are seen in this imagery at various colors, indicating the varying intensities of these fires.  The Fire Temperature RGB should be available on your AWIPS-2.


The smoke from the Canadian and Montana (and other) fires also moved into the Pacific Northwest.  The link below is to a GeoColor loop over the Pacific Northwest that covers 1-2 August.

Posted in GeoColor Imagery | Leave a comment

Dust event in the El Paso vicinity on 4 April 2017

The GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.  Users bear all responsibility for inspecting the data prior to use and for the manner in which the data are utilized.

During the afternoon of 4 April 2017, strong westerly winds and low relative humidity was observed in west Texas, southern New Mexico and northern Mexico as seen in the surface observations at 2100 UTC:


The winds were in response to a deepening low near the Texas Panhandle.  Although none of the observations in the plot above show blowing dust, METAR sites such as ELP (El Paso, TX) included some periods of blowing dust in the late morning and afternoon hours.  Here we explore how well the areas of blowing dust were seen in GOES-16 satellite imagery, using 3 different loops during the afternoon hours of 4 April.

The first loop (below) goes from 1827 to 2157 UTC and displays GOES-16 visible imagery (0.64 microns/Channel 2).

Do you see any evidence of blowing dust in this loop?

Refer to the static image below which contains a yellow circle, refer to the visible loop above and look in the region of this yellow circle.



The blowing dust is extremely subtle in the static image and even in the animation, no other areas of dust are obvious in the loop.

Sometimes features such as blowing dust show quite well in the CIRA True Color product.  True Color imagery approximates the response of normal human vision, providing a depiction of the satellite-observed scene.  The natural color of the background often makes it easier to see certain features when compared to the standard 0.64 micron visible imagery. Since there is no green channel on GOES-16, CIRA creates a “synthetic” green band to make this product, more information is provided at:

The GOES-16 True Color product:

Does the True Color product improve our ability to see dust in this case?


If we return to the same region (yellow circle), this dust plume appears a little more obvious relative to the visible band only.  North of this plume you may be able to discern additional plumes of dust, however they are still fairly subtle.

Certain products have been developed that are designed to highlight features such as dust plumes.  One example is the band difference product between the 10.3 micron and 12.3 micron bands.  The difference product would show a negative value in the presence of dust, which is shown as purple in the color table shown in the loop below:

With this in mind, do you see any additional dust plumes in this product?


You should now be able to discern multiple east-west oriented dust plumes (purple color).  Some of the plumes have a distinct source region associated with them.  This product helps us to see dust plumes that was not obvious in the other imagery.  Even in the GOES-16 era with 0.5 km visible band, dust may not be discernible without the help of a product designed to highlight dust.  Examples of other dust products that are under development include one by CIRA known as the DEBRA dust product with real-time data available at:



Posted in Blowing Dust Detection (Split-window technique) | Leave a comment

22 March 2017 GOES-16 imagery in the northeast US into southeast Canada

The GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.  Users bear all responsibility for inspecting the data prior to use and for the manner in which the data are utilized.

By Ed Szoke and Dan Bikos

On 22 March 2017 the GOES-16 visible band at 0.64 micron over Maine and southeast Canada shows many different cloud motions across different scales:

We identify some of these on the following image with 1600 UTC METARs overlaid:


The mesolow can be seen in the animation entering the scene and moving northeastward.  This mesolow is likely along a cold front with the synoptic surface low to the north over New Brunswick.  The brown dashed line above corresponds to a line of convection seen in the animation, this line is along the cold front moving eastward.  Ahead of the cold front in the warm sector we observe wave clouds over Nova Scotia early in the loop that transition to unstable cloud streets followed by convection.   Over north central Maine, we observe clouds at different heights.  There are higher level clouds moving northward ahead of the upper-level trough, meanwhile underneath those clouds we see low-level banded clouds moving relatively quickly to the southeast (in low-level cold advection).

The cloud heights can be more easily identified by looking at multiple channels on GOES-16.  In this 4-panel:

We are looking at 0.64 microns (Band 2 / visible) in the upper-left, 1.6 microns (Band 5 / snow / ice) in the upper-right, 1.38 microns (Band 4 / cirrus) in the lower-left, and 0.87 micron (Band 3 / veggie) in the lower-right.

Note the discrimination between low (liquid) versus high (ice) clouds for the following bands:

1.6 microns: High (ice) clouds appear darker / gray and low (liquid) clouds appear lighter (white).

1.38 microns:   High (ice) clouds appear lighter (white), whereas low (liquid) clouds are much more subtle and may appear dark.

The 0.87 micron (veggie) band is useful here in that Nova Scotia shows up much more clearly compared to the 0.64 micron (visible) band since vegetation surfaces have high reflectance and water surfaces have low reflectance at this wavelength.

Posted in Uncategorized | Leave a comment

A look at water vapor imagery from GOES-16

The GOES-16 data posted on this page are preliminary, non-operational data and are undergoing testing.  Users bear all responsibility for inspecting the data prior to use and for the manner in which the data are utilized.

By Ed Szoke and Dan Bikos

Among the 16 channels on GOES-16 are 3 water vapor channels at 2 km resolution (at NADIR), compared to a single water vapor channel on GOES-15 or GOES-13 at 4 km resolution.  Here we take a look at the new water vapor imagery over and near Colorado from Monday 6 March 2017.  We’ll use the 4-panel display from AWIPS II at the Boulder NWS WFO, with the lower resolution GOES single water vapor image in the lower right panel.  A series of three 4-panel images is shown below at hourly intervals from 1702 UTC to 1902 UTC.

Ed4panelWV1702_screenCapture Ed4panelWV1802_screenCapture


Ed4panelWV1902_screenCaptureIn the 4-panels above, GOES-16 water vapor imagery is channel 10 (7.34 µm), in the upper left panel, channel 9 (6.95 µm) in the upper right panel, channel 8 (6.19 µm) in the lower left panel, and from GOES-15 the single water vapor band at 6.5 µm.  A strong trough is passing across Colorado at this time, as seen in the RAP analysis of 500 mb heights and temperature below.  Many wave features on various scales are seen in all the imagery, but more distinctly with the higher resolution GOES-16 imagery.


Notice the very cold air near and just behind the trough axis, which passes across the Front Range of Colorado at this time (1800 UTC).  The strong warming/drying signature in northeastern Colorado is just south of the center of a smaller scale circulation associated with the overall trough.  Meanwhile farther to the east very warm brightness temperatures are seen across the Texas Panhandle into southwestern Kansas behind the dryline that had pushed eastward.  Plunging dewpoints and strong SW winds in this area created very high fire danger and numerous grassland fires developed and rapidly spread, as detailed in other blogs accessible through the VISIT homepage.

A strong jet accompanied this trough, as seen in the 250 mb wind speed and height analysis below.


One interesting feature that was quite evident in the looping of the imagery was a small piece of cirrus that moved rapidly from the UT/AZ border at 17z to just WSW of the 4-corners point at 1802z to ESE of the 4-corners by the last image at 1902z.  We can see this feature also in this loop of the channel 10 (7.34 µm) lower-level water vapor imagery from GOES-16:

At first we were a bit unsure as to what this was since it was moving so much faster than all the other cloud features.  In the higher resolution imagery it was quite apparent as a small “blocky” feature.  Using the tracking feature on AWIPS II we estimated a motion of ~115 kts, which is in agreement with the wind speed analysis of the stronger winds with the leading edge of the main jet.

Some of the other bands from GOES-16 were used to confirm the feature was a patch of cirrus.  These are shown in the 4-panel below.


This 4-panel from AWIPS II has 4 different channels from GOES-16:  channel 2 V=visible imagery (0.64 µm), in the upper left panel, the channel 4 “Cirrus band” (1.38 µm) in the upper right panel, channel 14 IR (11.2 µm) in the lower left panel, and channel 5 “snow/ice” imagery (1.61 µm) in the lower right panel.  In this case the cirrus and IR channels most clearly define the patch of cloudiness as an ice cloud.








Posted in Uncategorized | Leave a comment

Potential challenges of interpreting water vapor imagery: A tropical moisture case over San Diego on 21 Sep 2016

While water vapor imagery is no doubt very useful for forecasting, it can at times be tricky to interpret.  This blog discusses a particularly challenging case over Southern California on 21 September 2016, when deep tropical moisture produced a rain event for San Diego.  Even though the environment was very moist, indeed the precipitable water (PW) was a record for the date, GOES-15 water vapor imagery (with a central wavelength of 6.7 microns) implied a “dry” atmosphere.  GOES-R (now GOES-16) has 3 water vapor channels, one that senses higher in the atmosphere and the other lower than what is currently on GOES.  The central wavelengths for the new water vapor channels are similar to what is currently available from the GOES-15 Sounder, and for this case we will examine all 3 channels to see how the 2 new channels compared to what we have now on GOES.  Finally, we will also examine the CIRA/NESDIS Blended Total PW imagery and a new product being developed at CIRA, Blended Layered PW (LPW) imagery, to see what they reveal for this case.

We were alerted to this case by Alex Tardy, the WCM at the San Diego WFO (SGX), who also supplied some of the graphics.  Alex noted that the 12 UTC 21 Sep SAN sounding (below) had a record TPW for the date (chart below).




All this moisture produced widespread rain across far southern California, as depicted in the radar image shown below.


IR imagery at 1130 UTC indicates the clouds producing this rain did not extend to very high levels.  Sampling of the IR temperature near SAN yields a cloud top temperature of +1C or 274K.  Using the 12z SAN sounding this corresponds to a cloud top of about 580 mb.


First let’s look at a water vapor image near the time of the above imagery (1145 UTC) from GOES-15, which has a single channel at 6.19 microns (μ).  The weighting function (discussed further below) for the imager channel is fairly broad, extending generally from 300 to around 400 mb.  Note that over San Diego and areas to the south this water vapor channel shows a large area of warm brightness temperatures, which forecasters often interpret as indicative of a dry atmosphere.  The color table being used is the “RAMSDIS water vapor” table available on AWIPS2.













We can use water vapor imagery derived from the GOES-15 Sounder to get an idea what the 2 new water vapor channels that are on GOES-16 (-R) would show.  While the central wavelengths of the Sounder-based water vapor imagery compare to what is on GOES-16, the resolution is of course far worse (~8 km vs 2 km), and of course not as good as what is currently on GOES (4 km).  We can see this resolution difference (between GOES and the Sounder) by comparing the above image to the equivalent mid-level channel from the Sounder shown below.  Besides the resolution the imagery otherwise are fairly similar.

Slide14The 7.0 μ water vapor image shown above would generally sense somewhat lower in the atmosphere compared to the GOES imager water vapor image shown earlier.  The next image is the upper level water vapor channel from the Sounder-based imagery, shown below for the same time (1201 UTC), which would sense vertically a little higher than the current GOES imager channel.  The area of warmer brightness temperatures is not as extensive as in the mid-level imagery, indicating contribution from moisture at higher levels (the average weighting function for the 6.5 μ imagery is centered close to 300 mb for the standard atmosphere).


Perhaps the most intriguing of the new water vapor channels is the lower level channel, which from the Sounder would sense near 600 mb or lower.  The image for 1201 UTC from this channel at 7.4 μ is shown below.


Even though this channel senses lower into the atmosphere, we still see warm brightness temperatures from the SAN area southward.  These warm temperatures and by inference perhaps dry conditions are found even above an area where rain is occurring AND the environment has a record TPW level.  How is this possible?  To answer this question we need to look in more detail at what levels the various channels are actually sensing. Remember, the weighting functions depend on the temperature and moisture distribution in the environment, and the satellite viewing angle, and this means they will be sensing different levels across a domain.  Below is what the weighting functions look like for the SAN sounding that was shown earlier.


In the above figure the channel number is given for each band; Sounder Channel 10 = 7.4 μ, 11 = 7.0 μ and 12 = 6.5 μ, and the GOES imager water vapor channel is #3.  The brightness temperatures that are listed in the above figure would be equivalent to what one would find using the sampling tool on AWIPS2 for the various channels when sampling over SAN.

The weighting functions above differ from what would be calculated for the “standard atmosphere”, shown in the figure below.


The actual weighting functions are generally narrower and more complex than those from the standard atmosphere.   One important caveat to note about the weighting functions calculated for the 1200 UTC sounding is that while the viewing angle has been considered, the profiles are calculated using a forward model for clear sky conditions.  In our case we know that there are overcast conditions with a cloud top temperature derived from the IR imagery of 274K or ~580 mb over the SAN area.  The cloudy conditions would quickly saturate a given channel if it were to sense into the cloud layer, so there will be no contribution much below the top of the cloud layer.  In our case we note that if it were clear two of the Sounder channels would penetrate below the 580 mb level, so for these two channels the presence of the cloud layer should shift the actual weighting function upwards (colder), moreso for the 7.4 μ channel compared to the 7.0 μ channel.  Indeed, when we sampled these two channels over the SAN area the brightness temperatures found were 261 K for 7.4 μ and 259 K for 7.0 μ.  For the lower 7.4 μ channel this is 3 K colder than the brightness temperature listed for this channel as calculated for clear sky conditions, so a shift to higher in the atmosphere, as expected.  For the 7.0 μ channel the difference is only one degree, as would be expected smaller since that weighting function under clear sky conditions did not penetrate below the level where the cloud top occurred (580 mb).  With this in mind, for this case there was just enough moisture in the environment above the deep moist layer to saturate even the lowest water vapor channel before it sensed down to the cloud top, so it looked relatively “dry” even though we had a record TPW present in the atmosphere.

This case shows that while the new water vapor channels will be powerful tools for forecasters, we must be careful to note that they do not necessarily tell us what the moisture profile looks like through the entire atmosphere.  In other words, we cannot determine from the water vapor imagery how much PW is present in the atmosphere.  There is, however, satellite imagery that does allow one to detect the actual TPW present.  Most forecasters are familiar with this imagery on AWIPS2 known as the NESDIS Blended TPW product, which is shown for this case below.


This imagery combines measurements from several Polar satellites into a single image, hence the “blended” title and the time stamp that covers a time period instead of a single time.  The Polar satellites used have instruments that can measure the TPW even in the presence of clouds.  What we see of course for this case is the high TPW area associated with the remnants of the tropical system just to the south of San Diego, with the higher TPW values extending to the north.  We certainly get a different picture of the water vapor present in the atmosphere from this imagery compared to what we saw with the water vapor imagery.  CIRA is experimenting with a variant of this product that shows the PW in different layers, with the blended product known as the Layered Precipitable Water product (LPW).  The LPW for this case is shown below.


In agreement with what we saw in the San Diego 1200 UTC sounding, the LPW product shows moisture through the lower 3 layers but dry conditions above 500 mb (this higher level dry air is what was implied by the water vapor imagery).  This is also a blended product and while it is labeled at 1200 UTC it actually covers several hours leading up to this time.  A future version of this product will use an advective scheme that will make it more visually appealing.  This product is currently being tested at some WFOs, and if you are interested contact us and we can make it available on AWIPS2 through the LDM.



Posted in Uncategorized | Leave a comment

Challenging forecast for Colorado mountains snow event of 5 December 2016

On 5 December 2016, a significant storm was approaching the mountains of Colorado with various NWS winter weather watches/warnings posted for 6 December.  This snow event on 5 December occurred ahead of the storm in what looked like drying conditions behind a fast moving shortwave.  The GOES satellite imagery seemed to support the idea of drying on 5 December in all 3 available GOES Sounder water vapor channels.  However, enough low-level moisture remained in the presence of low/mid level steep lapse rates to produce this storm before the storm  on 5 December (generally around 6 inches at most ski resorts in the central/northern Colorado Rockies) that almost matched the amount of snow that occurred with the “main” storm on 6 December.  In this blog entry, we’ll show how the low-level moisture, while not obvious in the GOES sounder/imager water vapor bands, was observed with the CIRA Layered Precipitable Water Product.

The GOES Sounder provides water vapor imagery that includes the GOES-R/16 three water vapor bands.  Of course the resolution of the GOES Sounder imagery is considerably coarser (nominally hourly at 10 km) relative to GOES-R/16.  In order to get accustomed to looking at three water vapor channels of GOES-R/16, we can make use of current GOES Sounder bands (at least for the western CONUS since GOES-East Sounder has failed).

Next, we bring up a loop of the GOES-West Sounder/Imager satellite imagery between 0602 UTC 5 December to 0602 UTC 6 December:

The water vapor imagery that forecasters are accustomed to viewing in AWIPS is the GOES Imager water vapor band at 6.5 microns as shown in the upper right panel. Keep in mind that there are notes at the bottom of each frame in the loop.  To summarize what we see, an initial shortwave moves across northern Colorado between 0600 – 0900 UTC 5 Dec. followed by another shortwave thereafter with a fairly distinct signal in all water vapor bands, all within a strong westerly jet.  The dominant signal after 1800 UTC 5 Dec. is an elongated region of warmer brightness temperatures associated with the westerly jet slowly drifting southward, most readily observed in the mid- (6.5 micron imager and 6.95 micron sounder) and upper-level (6.2 micron sounder) bands.  One thing to note is that amidst the general warming signal we see an area of cooler brightness temperatures lingering over the central mountains of Colorado; what do you think this feature is?  As will be shown in other imagery later, what we are seeing are the tops of a lower cloud layer that persists.

Next, we bring up a loop of multiple products, including the CIRA Layered Precipitable Water (LPW) product, along with GOES IR, visible as well as mosaic radar at night:

Forecasters are likely familiar with a satellite derived Total Precipitable Water (TPW) product, such as the NESDIS blended TPW product on AWIPS.  The LPW product is derived from microwave instruments onboard multiple polar-orbiting satellites.  The unique aspect of the LPW product is that the moisture is shown for 4 separate layers as indicated on the looping imagery.  Keep in mind, the resolution of the instruments is considerably less than GOES and varies between instruments which is why the products appears more “blocky” at some times compared to others.  Also, the satellite passes occur at irregular times, therefore post-processing displays what data is available at 3 hour intervals.  The product time shows the latest available satellite pass, however keep in mind the most recent pass at a point may be several hours old due to irregular satellite pass intervals.  For example, look at central Nevada between 0918 and 1218 UTC 5 December, the same data is displayed since a new pass does not occur until after 1218 UTC, which shows up in the 1517 UTC image.  This product is still experimental, however it is available from CIRA in real-time, as well as in AWIPS as a satellite proving ground product (contact us at  Additionally, CIRA is currently developing an advected version of the LPW product that will display at higher resolution and appear less blocky.

One of the main points to emphasize in this particular case is that we do not see a sharp north-south gradient over central/western Colorado in the 700-500 mb layer of the LPW (this is the most applicable layer for moisture flowing over the elevations of the Colorado mountains).  Let’s focus on the 0018 UTC 6 Dec. LPW imagery when we had a higher resolution pass over Colorado.  We may have concluded from the GOES water vapor imagery shown earlier that the region of warmer brightness temperatures was associated with “drying” through a deep layer upstream of the mountains (i.e., western Colorado).  The LPW product clearly shows that there is moisture present in the 700-500 mb layer at 0018 UTC across western Colorado that would still be available to later move over the mountains in the westerly flow.  We don’t see this moisture in the various water vapor bands since the weighting function profile is not low enough (even at 7.35 microns).  If moisture is your primary forecast issue, the LPW product can clarify moisture at different levels in a more definitive way than implying it from GOES water vapor imagery alone.  In this case, the moisture that remained produced a period of snow that lasted past 0600 UTC 6 Dec. in the mountains which was not predicted, perhaps owing to what looked like a drying signal in the water vapor imagery.  This shallow moisture was also way underdone by even the high resolution models which predicted almost no snowfall and as seen in some of the radar mosaic images, was not resolved by radar either (due to overshooting beam).  Here, the lower level moisture was sufficient to produce moderate snowfall because of the steep lapse rates that were present behind the initial shortwave, as seen in the Grand Junction sounding:


Also note the moisture between 600 and 700 mb which would be reflected in the LPW 700-500 mb layer imagery.

GOES IR imagery from 0000 to 0600 UTC 6 Dec. displayed above with the LPW product, shows the low clouds over the central mountains of Colorado, however they are subtle due to insufficient contrast between the low clouds and cold ground.  The clouds are less subtle in the loop below:

This is the CIRA GeoColor product which is an experimental GOES-R Proving Ground product.  The low clouds at night are peach colored because this product utilizes the 10.7 minus 3.9 micron product at night to highlight lower clouds / fog.  Note that the higher level clouds are white in this product.

Posted in GeoColor Imagery | Leave a comment

1-minute imagery of warm conveyor belt on 1-2 February 2016 winter storm


A winter storm that passed through Colorado on 1-2 February 2016 resulted in significant snowfall over northern / northeast Colorado:


One of the key aspects of this extra-tropical cyclone was the development of a warm conveyor belt (Harrold 1973).  The 4-km NSSL WRF-ARW synthetic water vapor imagery from the 0000 UTC 1 February run valid 1800 UTC 1 Feb to 0600 UTC 2 Feb:

forecasts what appears to be a warm conveyor belt (WCB) by around 0100 UTC 2 Feb, highlighted here:



Keep in mind, one the of the limitations of the synthetic water vapor imagery from the NSSL WRF is that that areal coverage of cold cloud tops tends to be underdone due to the microphysics scheme of this model run.  Interpretation should focus on the feature of interest (the WCB) rather than the specific cloud top temperatures to compare with GOES.

For this event, GOES-14 Super Rapid Scan Operations for GOES-R (SRSOR) was in effect, meaning that 1-minute imagery was collected (Schmit et al. 2015).  This is in preparation for GOES-R (scheduled for launch in October 2016) since 1-minute imagery will be available much more frequently for significant weather events.

In the 1-minute GOES-14 water vapor imagery between 1800 UTC 1 Feb to 0100 UTC 2 Feb:

We can see the development of the WCB over eastern Colorado and western Kansas, consolidating by the end of the loop:


Next, lets consider the perspective of the GOES-IR (10.7 um) 1-minute imagery:

Note the rapid expansion in areal coverage of colder cloud tops from the Texas panhandle / western Oklahoma into western Kansas then curving cyclonically westward into Colorado.  This is the WCB of interest.  Where is the surface low in this loop?  Note the highlighted region below:



In the animation near this time, we observe a circulation with colder brightness temperatures just after 2300 UTC followed by what appears to be a deformation zone that is quasi-stationary.  The surface low is slightly southeast of this zone that is quasi-stationary.

Later, between 0100 and 0600 UTC 2 Feb:

We see the continued expansion of colder cloud tops associated with the WCB that is impacting Colorado.  During this period, snowfall rates increased as a result of this WCB as heavy snow impacted much of northern / northeast Colorado.

We observe a number of interesting features in the 1-minute imagery that we normally would not see under routine GOES scanning at 15 minute intervals.  For example, note the highlighted region in the following two images at 0300 and 0418 UTC:




Closer inspection of the animated imagery during the time period shows the development of transverse bands along the western edge of the WCB (which has a well defined limiting streamline).

Moving on to the next time period of IR imagery:

We see a number of gravity waves along the edge of the WCB that appear and disappear over short-time scales, for example, note the highlighted region at 0645 UTC:


1-minute imagery from this winter storm illustrates one of the reasons why GOES-R will be a “game changer”.  The high temporal resolution imagery will show features that were not sampled adequately to be observable under current/past temporal resolution.  Since some of these features have not been seen before, there will be an opportunity for research into these new features to understand what we are observing, and more importantly, potential applications for use in operational meteorology.  Another consideration is the 1-minute imagery latency on AWIPS will be approximately 1-minute, much greater than currently available for GOES RSO (Rapid Scan Operations), this impacts how much more effectively the data could be used operationally.

For completeness, analyze the 1-minute water vapor imagery for the remainder of the event.  What additional features do you see?

How does the development of the WCB relate to the increase in snowfall rates as observed from the NESDIS Snowfall rate product retrieved from polar orbiting satellites?



Harrold, T.W. 1973: Mechanisms influencing the distribution of precipitation within baoclinic disturbances. Q.J.R. Meteorol. Soc., 99, 232-251.

Schmit, T.J., and Coauthors, 2015: Rapid Refresh Information of Significant Events: Preparing Users for the Next Generation of Geostationary Operational Satellites. Bull. Amer. Meteor. Soc.96, 561–576.

Posted in SRSOR 1-minute imagery, Synthetic NSSL WRF-ARW Imagery | Leave a comment

Himawari imagery of 924 mb low in the Pacific

At 0600 UTC 13 December 2015, the NOAA Ocean Prediction Center analyzed a 924 mb surface low in the north Pacific in the vicinity of the western Aleutian islands:


The analyzed minimum central pressure of 924 mb ties the record for lowest pressure in the north Pacific during the period of record (since the winter of 1969-1970).

Satellite imagery from the new Japanese satellite (Himawari-8) provided a spectacular perspective of this cyclogenesis event.  First, we will look at the larger scale by analyzing the RGB airmass product (please allow sufficient time for this loop to load):

Warmer air is displayed in green and red where the green regions have higher moisture content than the red regions. Mid-latitude air has a bluish color and areas of dark red show areas of subsidence and high ozone and potential vorticity.   Click here for more detailed information.

The first feature that catches your attention is typhoon Melor east of the Philippines.  As you may suspect for an intensifying tropical cyclone, green colors in the vicinity of the storm indicate higher moisture.

Focusing our attention to the extra-tropical cyclone further north.  Early in the loop, we see the system over Japan moving northeast with a trailing deformation zone on the northwest flank of the cyclone.  In time, this feature dissipates as we see the system intensify, it develops a comma cloud followed by a strong surge of high ozone air (orange/red colors) associated with the dry slot.  This corresponds to a stratospheric intrusion deep into the troposphere.  Soon after this feature we see a cusp that develops and eventually wraps cyclonically around the upper low until becoming vertically stacked by the end of the loop.  Note the red/orange colors during this process as well, corresponding to high potential vorticity that is tracked by the high ozone air caused by the stratospheric intrusion.

Next we’ll look at the longwave infrared (11 um) loop zoomed in over the north Pacific:

In this channel, we can see low-level cumulus clouds which correspond to low-level cold advection over the relatively warm sea surface.  The position of the cold front can readily be tracked by following the low-level cumulus.  When the cold front wraps around the low and intersects the warm front, the occlusion process begins.  The leading edge of the relatively colder clouds that wrap around the system appear to be associated with a sting jet.  Wind gusts over 100 knots were observed with this storm.

Finally, we analyze a zoomed in perspective of Himawari True Color / Geocolor imagery:

Alternately, you may view this loop:

True color imagery is shown during daylight hours, and Geocolor imagery is shown during nighttime hours.  The CIRA Hybrid Atmospherically Corrected (HAC) method is applied to produce this “true color” imagery.

The Hybrid Atmospherically Corrected (HAC) true color method uses the red, green, and blue Himawari bands, in addition to some information from bands 4 (0.86 micrometers) and 13 (10.4 micrometers).  A Rayleigh correction is performed at each band in order to correct for the effects of Rayleigh scattering.  The result is an image that is significantly more crisp and clear, and less milky, than without the correction.

Once the imagery transitions over to nighttime the CIRA Geocolor algorithm is applied to the imagery.  White colors are high level ice clouds, reddish colors represent lower level liquid water cloud and city lights (static) are shown in yellow.

For more detailed information on analyzing spiral rings around extra-tropical cyclones, see this article.

Real-time Himawari-8 imagery may be viewed at:


Posted in GeoColor Imagery | 1 Comment