Color-enhanced GOES-15 Imager Water Vapor Imagery (click image to play animation)

The Sensor Processing Subsystem (SPS) for GOES-15 (operational as GOES-West) was changed at 20:45 UTC on March 12, 2012. This change that should have been transparent to the users introduced an error to the calibrated data that was especially apparent in the Imager Water Vapor Channel (Channel 3). The SPS is used to generate the GVAR data stream. The error was traced to a database issue in the new SPS that included the incorrect coefficients for the conversion to brightness temperature for the on-board blackbody. The blackbody temperature is used as part of the calibration of the IR bands. As a result, water vapor brightness temperatures were about 5% too cool for approximately five days. Other Imager channels were not so dramatically affected and Sounder Channels were not affected.

For example, in the image toggle above, the ‘Before’ image, at 20:30 UTC 12 March, shows the warmest brightness temperatures over the eastern Pacific Ocean west of Baja California. The brightness temperatures in that yellow region cooled by 2 K over those 30 minutes. Similarly, cold brightness temperatures over cirrus shield off the coast of Oregon/California cooled by 1-2 K. (The cold cloud tops over southwest Oregon also cooled, but that cooling is likely influenced by synoptic forcing). In contrast, Sounder data (below), shows little cooling between 20:02 UTC and 21:02 UTC on March 12th 2012. Brightness temperatures in the warm band west of Baja peak near 244.5 K in both images; brightness temperatures in the cirrus shield off the coast of California remain near 218 K (and actually show signs of warming between 2002 and 2102 UTC).

Color-enhanced GOES-15 Sounder Water Vapor Imagery (click image to play animation)

Color-enhanced GOES-15 Imager and Sounder Water Vapor Imagery

The four-panel image above also highlights the change in Imager data, and the static nature of the Sounder data. The enhanced Imager data, in the two panels on top, shows less yellow (that is, cooler temperatures) at 2100 UTC vs. 1800 UTC. Sounder data shows very little change over the same three hours.

The change in Brightness temperatures was first brought the attention of NOAA scientists via email from ECMWF. In preparation for the planned routine use of GOES-15 Imager data in the global model at ECMWF (starting in April of 2012), that center is monitoring observed and modeled radiances, and a big change started late in the day on March 12, 2012. This website shows the monitoring over the past three weeks. A similar site that shows how NCEP monitors Satellite Data is here. In addition, NOAA/NESDIS STAR compares GOES Imager observations to observations from high-spectral resolution measurements.

The image above shows GOES Imager Water Vapor Mean brightness temperatures over water from 90 S to 90 N and 110 W to 170 W. Note the big drop in mean temperature late in the day on March 12, and the big increase last in the day on March 16 2012. Comparisons between geo-located GOES-15 and MTSAT Mean Water Vapor Brightness temperatures, below, show good agreement until 2045 UTC on March 12, when the SPS change that accessed incorrect values in a different database was implemented. Database values were checked in a test starting at 21:00 UTC on March 14th until before the 00:52 UTC image on March 15th before the correct database values were used starting again at 2045 UTC on March 16th. (Link). At that time in both graphs, GOES-15 brightness temperatures return to more appropriate values.

Interpretation of the Sounder vs. Imager water vapor imagery should be colored by the spectral response functions (SRF) of the two instruments. The GOES-15 Imager SRF for water vapor is a broad one centered near 6.5 micrometers. The three GOES-15 Sounder water vapor channel SRFs are much narrower and are centered at 6.5, 7.0 and 7.4 micrometers. In addition, nominal resolution for the GOES-15 imager is 4 kilometers at the sub-satellite point versus something closer to 10 kilometers for the GOES-15 Sounder.

Special thanks to Cristina Lupu at ECMWF for alerting us to this issue.

View only this post Read Less

MODIS 0.65 µm visible channel image + MODIS Land Surface Temperature product

On 18 March 2012, the maximum temperature at Winner, South Dakota (station identifier KICR) reached 92º F (33º C), which was just shy of the State’s all-time record high temperature for the month of March (which was 94º F or 34º C, set at Tyndall in 1943).

An AWIPS image comparison of MODIS 0.65 µm visible channel data with the corresponding MODIS Land Surface Temperature (LST) product (above) showed pockets of very warm LST values in the 95-105º F range (darker red color enhancement) across parts of South Dakota at 19:10 UTC (1:10 pm local time). Differences in soil type and vegetaion density can contribute to higher LST values — and there is no direct 1:1 correspondence between LST values and air temperature values measured in an instrument shelter at a height of 5 feet off the surface. Daily record high temperatures were also set on 18 March at other locations in central and eastern South Dakota: Pierre (88º F), Sioux Falls and Aberdeen (85º F), Sisseton and Watertown (82º F), and Mobridge (81º F).

Land surface temperature values were not available in the immediate vicinity of Winner at the time of the MODIS images, due to patches of thin cirrus cloud over that area — the cloud mask prevents the calculation of LST products in cloudy areas. These patches of thin cirrus clouds could be seen as colder IR brightness temperatures (yellow to green color enhancement) in the MODIS 11.0 µm IR channel image (below). The warmer north-to-south oriented line located to the northwest of Winner was the burn scar from the Okreek fire which burned on 05 October 2011.

MODIS 11.0 µm IR channel image

View only this post Read Less

MODIS true-color Red/Green/Blue (RGB) images

250-meter resolution Terra and Aqua MODIS true-color Red/Green/Blue (RGB) images from the SSEC MODIS Today site (above) revealed the presence of an undular bore along the leading edge of a southward-moving cold front (also known regionally as a “pneumonia front“) on 15 March 2012.

AWIPS images of the MODIS Sea Surface Temperature (SST) product (below) indicated that SST values in the central portion of Lake Michigan (behind the cold frontal boundary) were generally in the upper 30s to low 40s F (darker blue color enhancement). At Racine in southeastern Wisconsin (station identifier KRAC), the surface air temperature dropped from 69 F to 47 F in 2 hours after the front passed. Just to the north, the high temperature of 72 F at Milwaukee (station identifier KMKE) set a record high for the date — then the temperature there dropped into the upper 40s F a few hours later.

MODIS Sea Surface Temperature product + Surface frontal analysis

During the previous night-time hours, the southerly to southwesterly flow of unseasonably warm air (with dew points in the 50s F) over the cold waters led to widespread lake fog across almost all of Lake Michigan, as was seen in a 03:22 UTC MODIS fog/stratus product (below).

MODIS fog/stratus product

A few hours later (at 07:33 UTC),  isolated convective rain showers were moving across the southern half of Lake Michigan — these showed up at the darker gray to black enhanced features on the MODIS fog/stratus product image (below). The corresponding MODIS Instrument Flight Rules (IFR) and Low Instrument Flight Rules (LIFR) Probability products indicated a number of areas with IFR and/or LIFR probabilities in excess of 80-90% (darker red color enhancement).

MODIS fog/stratus product + IFR Probability product + LIFR Probability product

View only this post Read Less

Regular readers of this blog will be quick to recognize the GOES imagery above as a low cloud detection product that exploits the differences in emissivity properties for water droplets that exist between 3.9 and 10.7 µm, two radiation bands that are detected on the GOES imager. The emissivity differences mean that 10.7 µm brightness temperatures will be warmer than 3.9 µm brightness temperatures, so a difference field will highlight where low clouds exist. This can be done with GOES imagery, above, or with MODIS imagery, below. Note that the existence of the low cloud may or may not suggest fog: only the top of the cloud is detected; whether or not the cloud rests on the ground cannot be determined easily from satellite.

Both the GOES and MODIS imagery show slow expansion to the low cloud field over the 3 hours, as might be expected given slow cooling at night. However, careful inspection reveals a variety of regions that show obstructions to visibility but no indication of fog or low clouds. For example, Montgomery (KMGM), Mobile (KMOB) and Tuscaloosa (KTCL) all show fog observations in the absence of a clear signal of detected low cloud. Similar observations occur over eastern Georgia and the Carolinas.

New low-cloud detection algorithms that incorporate model information (from the RUC, or, in the near future, the Rapid Refresh) can quantitatively describe the evolution of the low cloud and fog field. These algorithms were initially developed (and trained) using GOES data and are distributed to the AWIPS environment here at CIMSS. The loop above shows probabilities of Low IFR visibilites over the deep south during the morning. (Click for the 0400 UTC and 0700 UTC examples). Because the GOES-R product synthesizes model and satellite data together, better fog/low cloud detection occurs in regions where high clouds make low cloud detection difficult. Note how the probability of a visibility obstruction increases during the course of the night, and how places that develop fog are among the first to see a LIFR signal.

The visible imagery from GOES-13, below, shows the characteristic erosion of the low clouds, from outside in, during the course of the subsequent day. As usually happens, cumulus development is suppressed in regions where low clouds persist during the morning.

GOES-13 Visible (0.63 µm) (click image to play animation)

View only this post Read Less