Halloween Blizzard of 1991

October 31st, 2009 |

GOES-7 Visible (0.65 um) images, 31 October and 01 November 1991 [click to play animation]

GOES-7 Visible (0.65 um) images, 31 October and 01 November 1991 [click to play animation]

The Halloween Blizzard of 1991 was an early-season storm that moved north from the Gulf of Mexico to the upper Great Lakes. Unseasonably cold air allowed the rich moisture-laden airmass to deposit a long band of snow from the Panhandle of Texas northeastward to western Lake Superior. Many early-season snow total records were broken, and single-storm records fell at Minneapolis (28.4″) and Duluth (36.9″) Typically storms from the Gulf of Mexico do not move due north; however, eastward motion of this system was blocked by a large nor’easter off the coast of New England (the so-called “Perfect Storm”).

In the visible loop above, notice the rapid melting of snow deposited by the system in the Texas Panhandle, despite record cold (30 and 31 October 1991 are the only October days in Amarillo history when the surface temperature stayed below 30 F all day). Snowcover in South Dakota (the Missouri River stands out) also speaks to the chill in the airmass on the cold side of the storm. A larger-scale visible animation is available here.


The 1991 “Halloween” storm is the “single storm record for the metropolitan (Twin Cities)” area. A comparison of a GOES-7 Infrared and visible image on November 1, 1991 at 21 UTC.

A multi-day GOES-7 infrared (window) animation starting on October 31, 1991.
A GOES-7 visible band animation over parts of 2 days starting on October 31, 1991 .


These NOAA GOES-7 data was accessed via the University of Wisconsin-Madison SSEC Data Services, using the McIDAS-X software.

Snow cover in Colorado, Kansas, and Nebraska

October 23rd, 2009 |

MODIS visible and 2.1 µm near-IR snow/ice channel images

MODIS visible and 2.1 µm near-IR snow/ice channel images

AWIPS images of the MODIS visible and 2.1 µm near-IR “snow/ice” channels (above) showed areas of snow cover across parts of eastern Colorado, far northwestern Kansas, and Nebraska on 23 October 2009. Snow is a strong absorber at the 2.1 µm wavelength, so it appears very dark on the snow/ice channel image.

The corresponding MODIS Land Surface Temperature product (below) revealed significantly colder LST values in the middle to upper 30s F (darker green colors) where the snow cover was deeper. There was a lack of surface reports in the exact areas of deeper snow cover, except for Limon in eastern Colorado (station identifier KLIC), which was reporting a surface air temperature of 39º F at the time.

MODIS Land Surface Temperature product + surface METAR data

MODIS Land Surface Temperature product + surface METAR data

Snowfall amounts from this particular storm (which moved through the region on 22 October) included 15 inches at Elizabeth, Colorado, 12 inches at Brady, Nebraska, and 4 inches at Saint Francis, Kansas. These locations are marked on a MODIS Red/Green/Blue (RGB) true color image from the SSEC MODIS Today site (below, displayed using Google Earth).

MODIS true color image (displayed using Google Earth)

MODIS true color image (displayed using Google Earth)

Hurricane Rick at Category Five Intensity

October 19th, 2009 |

Eastern Pacific Hurricane Rick, shown above near peak intensity at sunset on 17 October 2009, is the second strongest hurricanes on record in the eastern Pacific — weaker only than 1997’s Linda. Sustained winds at this time were estimated to be 180 miles per hour, and the central sea level pressure was estimated to be 906 mb. Note in the visible imagery the presence of gravity waves in the cirrus shield that makes up the central dense overcast (CDO). In addition, as noted in the Tropical Prediction Center discussion issued near this time, the stadium effect in the Hurricane eye is readily apparent.

Rick formed out of a tropical disturbance southwest of the Gulf of Tehuantepec (a loop of 3-hourly water vapor imagery here, and a loop of 6-hourly 11-micron imagery here show an interesting flare-up of convection in the Gulf of Tehuantepec in the days before Rick formed. It is worth pondering how that convection influenced Rick’s early and rapid growth). The evolution from strong tropical depression (here, at 2100 UTC on 15 October) to minimal hurricane (here, at 1500 UTC on 16 October) to category 4 hurricane (here, at 1500 UTC on 17 October to category 5 hurricane, above, was rapid indeed and speaks to the ideal environment through which the disturbance traveled. Consider the image below from the CIMSS Tropical Weather Website.


The image shows that the theoretical minimum to the central pressure in the region through which the system traveled was below 880 mb! (This value is a function of sea surface temperature, and of atmospheric thermodynamic profiles as described here. Note that Rick was moving across ocean waters with surface temperatures close to 30 C as it intensified rapidly. Wind shear as the storm rapidly intensified time was also very low (as diagnosed by Satellite winds). Very warm ocean waters and low vertical wind shear are key ingredients in allowing the strengthening of tropical systems.

The ideal environment resulted in a category 5 storm with a very tall circular ring of convection around the eye. The GOES-11 10.7-micron image, below, shows temperatures of nearly -80 C (the purple pixels within the grey) in the tallest convection around the eye.


(Added: Note in the water vapor and infrared imagery loops, above, the presence of what looks to be a binocular-shaped eye. This is an artifact of the interpolation used to blend GOES-12 and GOES-11 imagery to combine one cohesive picture. In individual images from either satellite, only a single eye is present).


Polar orbiting satellites, such as NOAA-19, give high-resolution images of the storm. The 10.8-micron example above, from 2020 UTC on 17 October, as the storm neared its peak intensity, shows pixels northwest of the storm center (this NOAA-19 pass is ascending, so north is towards the bottom of the image) with brightness temperatures of -84 C. Note also the more circular aspect ratio that comes from the polar-orbiter’s more top-down view, versus the Geostationary satellite’s oblique view. Visible imagery, below, at 0.65 and 0.86 microns, from the NOAA-19 AVHRR instrument, show better storm structure as well.



MODIS imagery from the Terra and Aqua satellites can also be used to investigate the storm. Unfortunately for this storm, the Aqua overpass granule split was right across the storm eye (granules are created so that the vast amount of data created by the satellite are more easily transportable). Gluing the two images together does not re-capture all the missed points, but it does give a good representation of the storm intensity here. A later MODIS image from TERRA, below, from 1755 UTC on 18 October (that is, about a day after the image from Aqua), below, shows a somewhat cloudier, but still quite distinct, eye. At this point, Rick has passed its peak in intensity.


(added: Jesse at Accu-Weather has other imagery of Rick here).

Lake Effect Showers near Lake Michigan

October 16th, 2009 |


A cold airmass over the Great Lakes states and northeast that allowed an unprecedentedly early snowfall over central Pennsylvania (see precipitation totals ending at 1200 UTC on 16 October here) is also supporting the development of Lake-effect precipitation downwind of Lake Michigan. In the loop above (click the loop to enlarge), pay attention to the cumuliform cloud development over the relatively warm waters of Lake Michigan, and note how the straight lines of clouds move south-southwestward towards the lake counties of Wisconsin and Illinois. These cumuliform clouds contain showers as noted in the radar imagery here. The plot of surface data (below) at 1800 UTC shows a prevailing northerly or north-northeasterly flow at the surface, consistent with the motion of the lake-effect clouds. Note also the plots from moored buoys over the Lakes (plotted in red) that show temperatures in the low 50s (11-12 Celsius) over Lake Michigan and mid 40s (7-8 Celsius) over cooler Lake Superior.


The 850-hPa temperatures observed by radiosondes over the midwest at 1200 UTC on 16 October (here) show temperatures cooler than -5 Celsius over Lake Michigan. In general, a temperature difference of at least 13 C between 850 hPa and the Lake Surface is looked for when forecasting lake-effect precipitation. Observations in the surface plot certainly support a temperature difference exceeding 13 C between the lake surface and 850 hPa. Estimates of lake surface temperature from satellite have been hampered lately by persistent cloudiness over the Great Lakes basin. An MODIS estimate from 1800 UTC today, however, located here, does show temperatures in the 50s over Lake Michigan, and somewhat cooler waters over Lake Superior. This cooler lake temperatures in Lake Superior may explain in part the lack of lake-effect clouds just downwind of Lake Superior over the eastern part of the upper Peninsula (It is likely that other forcings may be suppressing cloud formation there as well; note that at the beginning of the visible imagery loop, cloud streets do extend from Lake Superior into the upper Penisula of Michigan just to the east of Marquette; these cloud streets dissipate with time, suggesting strong subsidence in the region).

Development of lake-effect clouds in mid-October is a reminder of what is to come in the near future as cooler and cooler airmasses develop over Canada and move southward over the still-warm Great Lakes.