1-minute Mesoscale Domain Sector GOES-19 (GOES-East) True Color RGB images from the CSPP GeoSphere site (below) showed ice that had formed in far southern Lake Michigan (within the nearshore waters of Wisconsin, Illinois and Indiana) on 20 January 2026.

1-minute GOES-19 Visible images with plots of surface winds and surface air temperatures (below) depicted the cold air temperatures — from single digits below 0ºF to single digits above 0ºF — that were present near the Lake Michigan coast shortly after sunrise. Along the eastern coast of the lake, surface air temperatures were in the 10-16ºF range, and lake effect cloud bands were producing snowfall in far southwestern Lower Michigan.

The GOES-19 Visible image at 1510 UTC (below) included an overlay of Metop-C ASCAT wind barbs — which showed westerly winds of 20-25 kts that were causing the ice to slowly drift away from the coast.

A toggle between the 1500 UTC GOES-19 Visible image with and without an overlay of the 1500 UTC GOES-19 Sea Surface Temperature (SST) derived product (below) indicated that SST values were as warm as 3.54ºC in the open waters just east of the sea ice.

A 30-meter resolution Landsat-8 Natural Color RGB image visualized using RealEarth (below) provided a more detailed view of the lake ice in the vicinity of Chicago.

View only this post Read Less

Unseasonably cold temperatures were found in the early morning hours of 20 January 2026. This map of observed temperatures from 0800 UTC (2:00 AM CST) shows that a tongue of cold air has plunged southward with subzero temperatures found even in the southern portions of Minnesota and Wisconsin while single-digit temperatures were common throughout the Midwest.

Of particular interest, however, are the isolated temperatures between -11 and -18 in southwestern Wisconsin. Are these temperatures observational outliers with no physical basis, or are they really representative of the conditions on the ground? Let’s use satellites to see if we can learn more about what’s going on.

We’ll start with a regional view of Band 13 from GOES-19. This is the standard 10.3 micron infrared band used to identify the positions of clouds at all hours of the day. However, interpreting this band during abnormally cold conditions can be a challenge, because the color scale used to identify higher-altitude clouds paints the cloud-free surface in the same colors because the temperatures are so low. This animation shows this quite well: there is clearly a lake effect band on the western shore of Michigan’s lower peninsula, and yet it has the same cyan and blue colorization as the immobile parts of Minnesota and Wisconsin.

The infrared-observed surface temperatures of the upper midwest are clearly heterogeneous. But what is contributing to all of that variability? To analyze this, let’s adjust the color scale of the infrared image. This is a still frame from the 0800 UTC image above, merely adjusted to cnhance the dynamic range over the observed surface temperatures. Light is warmer than dark.

Compare the patterns in the infrared image above to this terrain map from the Wisconsin Department of Natural Resources. Note how the colder (darker) regions on the infrared map in southwestern Wisconsin correspond to the valleys in the terrain map, with that pair of -11 F temperatures confined to the Wisconsin River valley. The -18 F temperature is found in the valley of the La Crosse River. Both of these rivers flow through the Driftless Area, the rugged heart of the upper midwest that avoided being flattened by glaciers in the most recent ice age.

We commonly associate increasing altitude with decreasing temperatures. So why are the valleys colder than the tops of the ridges? This is a result of cold air pooling and katabatic flow. The night was clear and calm; see how the surface winds in the earlier figure were between 0 and 5 knots everywhere. In the absence of synoptic scale forcing, smaller scale forcing can take effect. In this case, the air at the top of the terrain experiences strong radiative cooling. After all, there were no clouds above to absorb and emit outgoing infrared radiation. Cold air is also dense air, and so these radiative-cooled air masses flowed downhill into the valleys, where they pooled. This downward flow of radiatively-cooled air is known as a katabatic wind. On the scale of the relatively shallow terrain of the Driftless the flow would be difficult to measure directly, but in mountainous regions they can be quite significant.

Based on all of this, it is likely that the anomalously low temperatures in southwest Wisconsin aren’t false readings, but actually representative of their local microclimate.

View only this post Read Less

5-minute CONUS Sector GOES-19 (GOES-East) Water Vapor and Infrared images (above) showed the development of a standing wave cloud over northeast Minnesota and Lake Superior on 19 January 2026. This cloud feature was formed by a vertically-propagating internal gravity wave that resulted from the interaction of post-frontal northwesterly surface winds with the topography of the shoreline — the terrain quickly drops from an elevation of about 2000 feet above sea level (over northeastern Minnesota) to about 600 feet above sea level (over Lake Superior) in a very short distance.

The coldest cloud-top infrared brightness temperature associated with this standing wave cloud was -53ºC — which roughly corresponded to the tropopause (at an elevation of 6.9 km), as depicted in 1200 UTC rawinsonde data at International Falls (below).

Similar standing wave clouds have been documented here on the blog, on 19 January 2022 and 08 January 2019.

View only this post Read Less

Strong northerly winds created blowing snow and blizzard conditions across parts of eastern North Dakota and western Minnesota — particularly within the Red River Valley — on 17 January 2026. 5-minute CONUS Sector GOES-19 (GOES-East) Near-Infrared “Snow/Ice” images (above) showed the development of Horizontal Convective Rolls (HCRs) where winds were channeled down the Red River Valley (topography) — with gusts in excess of 30 knots, reducing the surface visibility to less than 1 mile due to blowing snow at several METAR sites.

GOES-19 Blowing Snow RGB images created using Geo2Grid (below) provided a more distinct view of the HCRs.

===== 18 January Update =====

Even stronger winds in the wake of a cold frontal passage (with gusts in excess of 40 knots) produced a second day of blowing snow and blizzard conditions across eastern North Dakota and western Minnesota on 18 January — and HCRs were once again very prevalent in GOES-19 Blowing Snow RGB images (above), across a larger area than the previous day. The surface visibility was reduced to near zero at some METAR sites.

View only this post Read Less