NASA SPoRT’s SST Composite Maps Capture Upwelling in the Wakes of Hurricanes Dorian and Humberto

NASA SPoRT’s SST Composite Maps Capture Upwelling in the Wakes of Hurricanes Dorian and Humberto

Written by Patrick Duran, Frank LaFontaine, and Erika Duran

Category 5 Hurricane Dorian passed over the Bahamas between September 1 and 3 2019, producing catastrophic destruction and causing at least 60 direct fatalities in the island nation. In addition to the impacts on human life, strong, slow-moving hurricanes like Dorian can leave lasting effects on the ocean over which they travel. Through a process known as upwelling, hurricanes bring colder water from below the surface up to the top layer of the ocean.  As a result, a trail of cooler sea surface temperatures (SSTs), also referred to as a “cold wake,” is often visible behind a passing storm. Meteorologists and oceanographers can monitor changes in SST and identify a cold wake following tropical cyclones using satellite data.

NASA SPoRT produces composite maps of SST twice daily using data from the VIIRS-NPP, MODIS-Aqua, and MODIS-Terra instruments, along with OSTIA-UKMO data obtained from the GHRSST archive at NASA’s Jet Propulsion Laboratory and the NESDIS GOES-POES SST product. The input data are weighted by latency and resolution to produce the composite, which is available at 2 km resolution.

Figure 1 shows a loop of the SPoRT SST composite from August 31 – September 23, 2019 over a region that includes the Bahamas and the Southeast United States. Two rounds of SST cooling are observed as Hurricanes Dorian and Humberto move through the region.

Figure 1: Animation of NASA SPoRT SST Composite Maps from August 21 through September 23, 2019. Daily images are displayed at 1800 UTC.

On August 31, very warm SSTs of around 29-30 deg Celsius (84-86 deg Fahrenheit) overspread the waters surrounding the Bahamas (Fig. 2).

Figure 2: SST Composite Map at 1800 UTC on August 31, 2019.

After Hurricane Dorian tracked through the region and made landfall in North Carolina on September 6, the waters north of the Bahamas were considerably cooler – in the 26–29 deg Celsius (79–84 deg Fahrenheit) range (Fig. 3).

Figure 3: As in Fig. 2, but for September 6, 2019.

Over the next week, the surface waters warmed a degree or two (Fig. 4), but did not fully recover to the same temperature observed on 31 August.

As in Figs. 2-3, but for September 13, 2019.

On September 13, Tropical Storm Humberto formed 210 km (130 miles) ESE of Great Abaco Island. As the storm tracked northeast past the Bahamas, it encountered the cold wake left by Hurricane Dorian the previous week. These cooler waters, combined with the influence of some dry air and vertical wind shear, inhibited the storm’s intensification as it passed by the Bahamas. On September 15, Humberto moved over the warmer waters of the Gulf Stream off the coast of North Florida (Fig. 5) intensified to hurricane strength.

Figure 5: As in Figs 2-4, but for September 15, 2019.

Humberto continued to strengthen, and attained a maximum sustained wind speed of 125 MPH as it passed by Bermuda on September 19. Its strong winds and associated waves overturned the same region of ocean that was previously affected by Hurricane Dorian, decreasing sea surface temperatures to as low as 25 deg Celsius (77 deg Fahrenheit) in some areas (Fig. 6).

Fig. 6: As in Figs. 2-5, but for September 19, 2019.

These images highlight the effect that tropical cyclones can have on SST, and how a hurricane can make it more difficult for any subsequent storms to intensify over the same region. Satellite analyses of SSTs (such as those produced by NASA SPoRT) allow forecasters to monitor SST across the globe, helping them to produce better forecasts of tropical cyclone intensity in all ocean basins.

Snow Cover Blankets Northeastern New Mexico

A potent winter storm system impacted portions of New Mexico on March 26, 2016, ending an extended stretch of very dry weather. Snowfall amounts of 3 to 9 inches were reported from the Sangre de Cristo Mountains eastward across the northeast plains. The MODIS and VIIRS satellite products proved useful for illustrating the extent of snow cover in both the daytime and nighttime scenes. The images below are graphical briefings posted to the NWS Albuquerque web page and shared via Twitter after this much needed snowfall event.

Graphical briefing showing the extent of snow cover during the nighttime and daytime periods on March 27, 2016.

Graphical briefing (part one) showing the extent of snow cover during the nighttime and daytime periods on March 27, 2016.

Graphical briefing showing the extent of snow cover through RGBs on March 27, 2016.

Graphical briefing (part two) showing the extent of snow cover through RGBs on March 27, 2016.

LEO Perspective of River-Effect Snow in North Alabama

The cold air outbreak over the eastern United States had impacts far and wide, including the development of snow showers all the way into northern Alabama.  However, between unseasonably low 850 mb temperatures and northwesterly flow, the outbreak also caused a semi-persistent band of snow to develop along the Tennessee River (downwind of a reservoir known as “Lake Wheeler”).

While most of the river-effect monitoring occurred with radar, the late-morning MODIS overpass captured one of the narrow river-effect bands (and did so more effectively than the lower-resolution GOES-East Imagery).


Figure 1. MODIS visible image, valid 1644 UTC 9 February 2016.  Larger lakes are outlined in blue, and the river-effect band is circled in yellow.

Snowfall reports from underneath the band have indicated 2 to 3 inches of snow, compared to the 1-2 inches reported with heavy or persistent snow showers elsewhere.  Unfortunately, orbit timing and cloud cover have not allowed us to view the snow swath using the Snow-Cloud RGB.  However, the Snow-Cloud RGB from the edge of this morning’s MODIS pass still illustrated the river-effect band persistence.


Figure 2. MODIS Snow-Cloud RGB image, valid 1549 UTC 10 February 2016.  The Tennessee River is the dark blue feature in the center of the image; the river effect band is circled in red.

Cyclone Observed by MODIS Air Mass RGB

MODIS Air Mass RGB (left) and 11 um image (right) from 08 February 2016 at 1427 UTC.

MODIS Air Mass RGB (left) and 11 um image (right) from 08 February 2016 at 1427 UTC.

An image captured this morning by the MODIS Terra instrument shows an impressive cyclone off the eastern coast of the US. The image on the left shows the cyclone in SPoRT’s Air Mass RGB and the image on the right shows the 11.0 µm from Terra (from 8 February 2016 at 1427 UTC). The deep red color on the RGB shows the intrusion of ozone-rich stratospheric air, which is an indication of deformation zones, jet streaks, and potential vorticity anomalies associated with rapid cyclogenesis, which itself indicates strong winds at the surface. This RGB is also limb-corrected for cooling at the edges of the swath, so we can assume the cyclone in this imagery is every bit as intense as it looks.

The new generation of geostationary satellites being deployed globally, such as Himawari, MTG, and GOES-R, will allow us to observe imagery like the Air Mass RGB several times an hour, enabling us to watch the cyclogenesis as it happens.

New Mexico Snow Event – SFR Assessment

An upper level low tracking over Arizona bought widespread rain and snow to New Mexico on January 13 and 14, 2015.  For this post, we will focus on the area from west central to central New Mexico.  A winter storm warning was issued for the higher terrain of western and central New Mexico with advisories posted for lower elevations. Originally, the Albuquerque Metro area zone was not included in the advisory, but on the afternoon of January 13th, the advisory was expanded to include the Albuquerque Metro area zone as snow was expected to impact the morning commute.

The image below shows the  6-h precipitation forecast for 06Z-12Z on 14 January from the 00Z run of the NAM12. Precipitation was forecast to focus on the area along and north of Interstate 40 from west of the Arizona border near Gallup, past Grants and just into Bernalillo County but west of Albuquerque, with another active area south of Albuquerque.


Several interesting features are illustrated in the next image – a comparison of the NESDIS Snow Fall Rate Product and 8-bit 0.5 Reflectivity from KABX near 09Z. The projections are slightly different, but the dashed line shows an area of somewhat enhanced snow fall rates from the NESDIS product and both Window Rock (KRQE) and Gallup (KGUP) are reporting snow. Note how the radar returns decrease in the same area, which is only covered by KABX.  Extreme western New Mexico, including the Four Corners area, has poor radar coverage, with the 0.5 scan over 11,000 feet AGL at KGUP. To the east along I-40, Grants (KGNT) is not reporting precipitation, while Albuquerque is reporting rain. Thus, the SFR product appears to accurately define the area of snow, though reports for the area north of I-40 are difficult to obtain, especially during the overnight hours.  By the afternoon, totals for the area along and north of I-40 ranged from 2 in to 12 in.

SFR/Z Comparison 09Z

The MODIS-VIIRS Nighttime Microphysics image depicts both the enhanced mid level clouds associated with snow (orange to red) and the low clouds (light pink/light green) impacting other areas of western and central New Mexico.


Moving two and one half hours forward, another SFR/0.5Z comparison is shown below.  Stations along Interstate 40 (KGUP and KGNT) are still reporting snow as are stations south of Albuquerque (KE80 and KONM), though snowfall coverage in the SFR product was reduced in both intensity and area. The radar image is focused on the greater Albuquerque area, or the locations within the white box on the SFR product. At this time, KABQ was reporting east winds gusting to 23kt. Gap winds through Tijeras Canyon (east of Albuquerque) have a downslope component of 1200 ft, often resulting in a snow-free zone around the Albuquerque metro area. The pink dashed line depicts the approximate edge of the precipitation-free area,  with the beam blocked area also evident further to the east. For this area, the SFR product indicates snow to the west and south of Albuquerque, verifying the observations.

SFR/Z Comparisons1130Z

The east gap winds relaxed and snow soon moved into the Albuquerque Metro area, with snow reported at KABQ at 1209Z. By 1313Z, conditions deteriorated with visibility of one-quarter mile in snow, and the airport picked up a quick inch of snow in just over an hour, with 1-2 inch amounts elsewhere. Thus, the winter weather advisory for the Albuquerque zone verified, and impacts during the start of the morning commute period resulted in a 2-hour school delay.


Alaskan Sunshine Doesn’t Burn the 24-hour Microphysics Product

In collaboration with the Geographic Information Network of Alaska (GINA) at the University of Alaska, NASA/SPoRT generates two VIIRS and MODIS microphysics satellite products for use by the National Weather Service in Alaska in assessing the presence of low stratus clouds: the RGB Night-Time Microphysics product, or simply NT Micro, and the 24-hour Microphysics product, or 24hr Micro. Being RGBs, these two products result from combining a number of satellite channels. Both RGBs use the 12.0-10.8 micron difference as the red channel, and both use the 10.8 micron signal as the blue channel. But the NT and 24hr Micro products diverge in what they use for the green channel: the NT Micro uses the classic 10.8-3.9 micron difference (the legacy “fog product” that has been employed by meteorologists for years), while the 24hr Micro uses a 10.8-8.7 micron difference. The motivation behind this divergence in approach is that the 8.7 micron channel is not affected by solar reflectance, while the 3.9 micron channel definitely is affected by solar reflectance. As a consequence, the NT Micro changes its appearance as night gives way to day and is not usable once the sun comes up, while the 24hr Micro provides a consistent depiction of the clouds both day and night.

Patchy low stratus clouds covered parts of southwestern Alaska on December 17, 2014. What follows are two animations taken from AWIPS at the NWS office in Fairbanks, Alaska that toggle images from around 15Z and 21Z, with the image near 15Z occurring during darkness and the image near 21Z occurring during the light of day. While there isn’t much sunshine in southwestern Alaska in December, there is still enough to cause trouble for the NT Micro with its reliance on the 3.9 micron wavelength in the green channel.

toggle NT Micro

The top animation is a toggle of the NT Micro. Note how at 15Z, in darkness, the green-yellow colors indicate the low stratus in southwest Alaska. But at 21Z, in daylight, these same clouds appear pink despite the continued presence of the low clouds. Observations from the village of Sleetmute (METAR identifier PASL), Alaska are a case in point. The ceiling and visibility plots at a number of METAR sites are overlaid in light blue text, with Sleetmute plotted just northwest of the yellow zone number “152” at the center of the image. At 15Z Sleetmute has good visibility but a ceiling down at 500ft overcast. At 21Z Sleetmute still has good visibility, and the ceiling remains low, in this case at 400ft broken 1100ft overcast. That’s not much of a change in the METAR over this span of time despite the huge change in the appearance of the NT Micro product between 15Z and 21Z.

The bottom animation toggles the 24hr Micro over the same span of time, from 15Z to 21Z, and there isn’t much change in the appearance of the 24hr product from darkness to daylight. The deck of low clouds over southwest Alaska appears bright yellow in this product, day or night.

toggle 24hr Micro

Every product has its strengths and weaknesses. At night the NT Micro shows more details and is perhaps preferable to the coarser looking 24hr Micro. But during the transition from darkness to daylight the NT Micro is not usable, while the 24hr Micro provides consistent imagery. The 24hr Micro product may prove increasingly useful in Alaska as the months progress and the long dark Alaskan winter gives way to a summer of almost continual daylight.

More Fog…This Time the Nighttime Microphysics RGB is the Better Product

So, with the moon now passing into the waning crescent phase, the Day-Night Band imagery is less operationally useful, at least for the detection of fog and other lower level cloud types.  That is, at least until the moon is back into the waxing gibbous phase.  Nevertheless, when cirrus clouds aren’t present, the Nighttime Microphysics RGB has proven to be a very valuable tool for the detection of fog and other low-level clouds.  Just this morning a forecaster at the Huntsville, AL WFO was able to use the imagery not only for the detection of fog, but also to aid in the issuance of a special weather statement about the fog.  The image below valid at ~724 UTC (0224 am CDT) 17 Oct shows the fog (whitish-aqua colors) lying across the valley areas of NE Alabama and adjacent areas of southern Tennessee and NW Georgia.

MODIS Nighttime Microphysics RGB 724 UTC 17 October 2014


Around the time of this image, the visibility in the foggy locations had decreased to ~1/4 – 1/2 SM or less.  Notice the fog in the DeKalb Valley is fainter than the fog in areas to the north and west.  Not only is the DeKalb Valley more narrow, but the fog was likely more shallow.  This feature of the imagery can also help to guide forecasters in assessing the longevity of the fog once sunrise breaks.  Over time, forecasters can develop a sense of pattern recognition with the varying degrees of color shading and tailor forecasts to better match the time of dissipation.  In this case, the fog in the DeKalb Valley began to dissipate significantly by about 1430 UTC, while  the deeper and more expansive fog to the north and west lasted about an hour longer.

First Hurricane-Force Low in the East Pacific

From September 20 through September 23, 2014, the Ocean Prediction Center (OPC) was monitoring the development of the season’s first hurricane-force extratropical storm in the East Pacific.  Models were suggesting a marginal hurricane-force wind event would unfold well west of the Pacific Northwest, near 140W longitude, north of 40N latitude.  OPC is routinely using satellite data to monitor and forecast these strong ocean storms.  On this particular event, OPC forecaster James Kells collaborated with Michael Rowland and David Kosier on if and when to pull the trigger on the hurricane-force warning.

GOES-15 6.5 um water vapor animation showing the evolution of the hurricane-force low.

GOES-15 6.5 um water vapor animation showing the evolution of the hurricane-force low.

The above animation shows the evolution of the hurricane-force low, with an eye-like feature evident near the end of the loop.  By 1200 UTC on the 23rd, it was forecast to develop hurricane force winds (64 knots or greater) just west of Oregon near 140W.  During the production of the 1200 UTC OPC Surface Analysis, there was question of whether or not the winds had reached hurricane force intensity. The ASCAT pass from ~0600 UTC showed a large area of 50-55 knot winds in the strong cold advection south of the low center, and the GFS model indicated that the system was still developing.  The GFS 0-30m boundary layer winds also indicated a very small area with hurricane force intensity.

Advanced Scatterometers A and B overlaid on GOES-15 Infrared imagery showing storm force winds at ~0600 UTC on 09/23/14.

Advanced Scatterometers A and B overlaid on GOES-15 Infrared imagery showing storm force winds at ~0600 UTC on 09/23/14.

The 1130 UTC MODIS RGB Air Mass product was timelier, and it showed an area of downward momentum south of the low with the deep purple shading. The corresponding water vapor image was less clear with upper level moisture obscuring the downward motion just beneath it.   In addition, there were no surface reports south of the low center as there were no buoys moored nor drifting in that vicinity.  Furthermore, most ships were aware of the danger and navigated away from the region neglecting the possibility of a surface report in the area of question.

Aqua MODIS RGB Air Mass image from 1130 UTC on 09/23/14.

Aqua MODIS RGB Air Mass image from 1130 UTC on 09/23/14.

A cross-section of the 1200 UTC 09/23/14 GFS model potential temperature and dew point temperature was taken through the low center in order to analyze the depth of the stratospheric intrusion, and also to gauge the magnitude of the downward momentum.  It showed a deep stratospheric intrusion to roughly 500 hPa, and it corroborated the strong downward momentum indicated by the imagery.  The RGB Air Mass image showed the intensity of the downward momentum through the red/purple coloring and gave a good indication of the stronger winds aloft mixing down toward the surface.  The imagery increased confidence with classifying the system as a hurricane force low.

The 1200 UTC 09/23/14 GFS vertical cross-section of potential temperature and dewpoint showing the downward transport of drier air associated with the tropopause fold.

The 1200 UTC 09/23/14 GFS vertical cross-section of potential temperature and dewpoint showing the downward transport of drier air associated with the tropopause fold.

The 1200 UTC 09/23/14 OPC surface analysis.

The 1200 UTC 09/23/14 OPC surface analysis.

~ Guest blogger, James Kells (OPC)

MODIS Nighttime Microphysics RGB Observations from This Morning…

It is said that a picture says a thousand words…well in this case let’s just say 434 words, as are contained in this post. Anyway, I’d like to point out six features in this morning’s Nighttime Microphysics RGB.  The image below (MODIS Nighttime Microphysics RGB) showed several features of varying degrees of operational relevance.

MODIS Nighttime Microphysics RGB with annotations valid 0755 UTC 16 July 2014

MODIS Nighttime Microphysics RGB with annotations valid 0755 UTC 16 July 2014


A myriad of cloud features can be observed, including fog in the valleys of central Appalachia, deep convective clouds along the Florida coast, patches of thin and thick cirrus over north-central Alabama, and low stratus clouds in Missouri…to name just a few.  Sure, this isn’t an exhaustive list of the potential cloud features to observe, but showcases the ability to contrast effectively between different cloud types.  Of perhaps significant interest is the ability to see the contrasting airmasses displayed across the Southeast region.  Notice the  pinkish colors north and west of the yellow curved line that stretches from central Louisiana to southern Virginia.  This represents a lower relative contribution of blue color, or lesser longwave radiation at the 10.8 µm wavelength, which is indicative of cooler temperatures.  To the south and east of this line, much more blue is apparent, which is thus indicative of warmer temperatures.   Surface observations valid at about the same time have been overlaid with the RGB image to provide temperature data context.  Air and dew point temperatures are around 10 degrees F cooler behind the line/front, but notice that the northerly wind shift is still on the south/east side of the line at such locations as Montgomery, AL and Columbus, GA.  At those locations, dew point temperatures were still 70 and 71 F, respectively, with air temperatures at 72 F.  So, the gradient in temperatures still lingered behind the surface front and is well depicted in the RGB imagery.  This type of information can be valuable to forecasters, as temperature, moisture, and wind characteristics are often complex in the vicinity of surface fronts.  Thus, while wind shifts may be observed initially, as in this case, the imagery shows the location of the temperature gradient much better.

The importance of this type of imagery is that it offers a much more effective assessment of meteorological phenomena than existing GOES imagery.  The only problem currently is the limitation of available imagery to forecasters, since these are from polar-orbiting platforms (Terra, Aqua, Suomi NPP), and thus provide just a few snapshots per night over a given location.  Nevertheless, the imagery form the VIIRS and MODIS instruments offer added value to existing GOES imagery and serve as valuable teaching and preparatory aids for future GOES-R and JPSS missions.

MODIS and VIIRS Products for Fog Detection in the TN Valley

I didn’t have a chance to make this post last week when the imagery were more time-relevant.  Nevertheless, I wanted to point out another example of the usefulness of MODIS and VIIRS imagery over current GOES imagery and show the usefulness of exciting products and imagery to come!  First, let’s take a look at the color-enhanced GOES-IR image below from the morning (0715 UTC) of June 20th.

Color-enhanced GOES-IR (11um) image valid 0715 UTC 20 June 2014

Image 1.  Color-enhanced GOES-IR (11 µm) image valid 0715 UTC 20 June 2014


I’ve placed the yellow circles in the image for a reason, which you’ll see below.  Further down, I’m going to show areas of fog displayed in the MODIS and VIIRS imagery, and granted, this is not the standard GOES channel difference (11-3.9 µm) typically used for making fog assessments.   However, this post is meant to show current (MODIS / VIIRS) and future capabilities (GOES-R / JPSS) that will make fog detection and cloud differentiation much more easy for the operational forecaster.  So, in the image above, fog is nearly unidentifiable as it was in the 11-3.9 µm channel difference image that morning (not shown).  Mainly high cirrus clouds can be observed scattered across the region.  Now, let’s take a look at the MODIS “fog” product, or channel difference (11-3.9 um) product valid at about the same time (Image 2).

Color-enhanced MODIS 11-3.9 u m product valid 0718 UTC 20 June 2014

Image 2.  Color-enhanced MODIS 11-3.9 µm image valid 0718 UTC 20 June 2014

Notice that in the same areas we can now begin to see low clouds (indicated by yellow colors) scattered around the valleys of the southern Appalachian region.  While the GOES-East imager is capable of detecting larger scale fog often in the valleys in the eastern circle, fog in the valleys in the western circle present challenges for the current GOES-East instrument, and is often not shown very well (even in the standard 11-3.9 µm channel difference).    Next, let’s take a look at a VIIRS Nighttime Microphysics RGB valid at about the same time.

VIIRS Nighttime Microphysics RGB valid 0723 UTC 20 June 2014

Image 3.  VIIRS Nighttime Microphysics RGB valid 0723 UTC 20 June 2014

In the RGB imagery it is much easier to detect the extent of the fog, making the operational forecast process much more effective.  Notice also that it is possible to see the fog through the higher clouds around the TN/GA/NC border region.  Not only does the resolution of the VIIRS and MODIS instruments allow for superior fog detection, but the RGBs in particular offer tremendous operational advantages.  As a user of RGBs for about 2 years now, I am convinced that this type of imagery has a relevant and needed place in future operational forecasting.  Of course, it will take time for forecasters to become accustomed and adjust to the new imagery, but it will happen.