Author Archives: Scott Lindstrom

Use MODIS data at High Latitudes

GOES-West IFR Probabilities, MODIS IFR Probabilities and surface topography over northern Alaska, ~2000 UTC on 14 January 2014 (click image to enlarge)

GOES pixels grow to large sizes at high latitudes, such as those found over northern Alaska. Consequently, the IFR Probabilities can give information that is difficult to interpret. Data from the polar-orbiting MODIS instrument (on board Aqua and Terra satellites), in contrast, have nominal 1-km resolutions even at high latitudes. In the toggle of imagery above, the MODIS IFR Probabilities suggests low clouds just north of the Brooks Range in northern Alaska. In contrast, the IFR Probabilities based on GOES-15 are difficult to interpret.

Consider using the MODIS-based IFR Probabilities. At very high latitudes, data from polar orbiters is more frequent than at mid-latitudes. Thus, there is a benefit from higher spatial resolution without an onerous loss of temporal resolution as happens in mid-latitudes.

Widespread Advection Fog over the Midwest

GOES-East IFR Probabilities and surface plots of visibilities/ceilings and surface analysis of dewpoint at 0202, 0402, 0615, 0802, 1002 and 1215 UTC on 10 January 2014 (click image to enlarge)

The northward movement of moist air over a snow-covered surface allowed for widespread advection fog in the midwest overnight from January 9th to 10th. The animation, above, shows GOES-R IFR Probabilities at 2-hour time steps. Included in the plots are surface observations and cloud ceilings (documenting the widespread region of IFR conditions) and the RTMA Dewpoint analysis that shows the slow northward movement of dewpoints at the surface. As this moist air moves over the cold snow-covered surface (the snow analysis from the National Operational Hydrological Remote Sensing Center is below), advection fog is a result. The GOES-R IFR Probability fields do a fine job of outlining where the IFR conditions are observed.

Analysis of snow depth from NOHRSC, 0600 UTC, 10 January 2014 (click image to enlarge)

Note in the animation above how the presence of higher clouds moving up from the southwest affects the IFR Probability fields. As high clouds overspread the advection fog, satellite data can no longer be incorporated into the GOES-R IFR probability algorithm, and IFR Probabilities drop, in this case from values near 90% to values near 55%.

Polar-orbiting data can also give information about low clouds and fog. Temporal resolution is far superior to geostationary, as shown below. In cases of small-scale fog, polar orbiter data can give important information by identifying the first region of a developing fog. In large-scale cases such as this, high-resolution data can better identify edges to the fields. The MODIS data in this case does show high probabilities over the midwest; the brightness temperature difference field shows evidence of high clouds from central Iowa southwestward. As with the GOES data, the presence of high clouds results in lower IFR Probabilities.

Toggle between MODIS-based IFR Probabilities and Brightness Temperature Difference at 0814 UTC 10 January 2014 (click image to enlarge)

Suomi/NPP data, below, from the Day/Night band shows widespread cloudiness over the midwest. The clouds are illuminated by the moon, nearly full, setting at this time in the west. Shadows are being cast by high clouds on the lower clouds over western Minnesota. The brightness temperature difference fields from Suomi/NPP are very similar to the MODIS data. In contrast to MODIS, the VIIRS instrument does not have a water vapor sensor, so the IFR Probability algorithms are not directly transferable to Suomi/NPP VIIRS data.

Toggle between Suomi/NPP Day/Night band and Brightness Temperature Difference at 0737 UTC 10 January 2014 (click image to enlarge)

IFR Conditions over Oklahoma and Kansas

GOES-R IFR Probabilities from GOES-13 (upper left), GOES-13 Brightness Temperature Difference (10.7 µm – 3.9 µm) Fields (upper right), GOES-R Cloud Thickness from GOES-13 (lower left), Suomi/NPP Day/Night Band (lower right), ~2345 UTC 8 January 2014 (click image to enlarge)

IFR Conditions developed over portions of Kansas and Oklahoma (and adjacent states) overnight. How did the GOES-R IFR Probability field diagnose the development of this event? At ~0000 UTC, the traditional method of low stratus/fog detection (Brightness Temperature Difference) showed two regions over the Plains, one centered over Oklahoma, and one over the Kansas/Nebraska border. IFR Probabilities at the same time covered a smaller area; Wichita in particular had low IFR Probabilities despite a brightness temperature difference signal, and Wichita did not report IFR conditions at the time.

Note that high clouds are also present in the Brightness Temperature Difference field over western Kansas, the panhandles of Texas and Oklahoma, and New Mexico. In the enhancement used, high clouds are depicted as dark greys.

As above, but at 0202 UTC 9 January 2014 (click image to enlarge)

By 0200 UTC (above), the high clouds have moved over parts of Oklahoma and Kansas. Consequently, there are regions over central Oklahoma and south-central Kansas where the brightness temperature difference field is not useful in detecting low stratus/fog that is occurring. The IFR Probability fields suggest the presence of low clouds despite the lack of satellite data because the IFR Probability Field also uses output from the Rapid Refresh Model that suggests saturation is occurring in those regions. Model data are also used where satellite data suggest low clouds/stratus are present to delineate where surface ceilings/visibilities are congruent with IFR conditions. As at 2345 UTC, Wichita is not reporting IFR conditions, although the brightness temperature difference field suggests IFR conditions might exist. The IFR Probability field correctly shows low values there.

As above, but at 0802 UTC 9 January 2014. Suomi/NPP data is a toggle of Day/Night band and 11.35 µm – 3.74 µm brightness temperature difference (click image to enlarge)

By 0802 UTC, high clouds have overspread much of Oklahoma, yet IFR conditions are occurring at several locations. IFR probabilities nicely depict the widespread nature of this IFR event. Probabilities are reduced in regions where high clouds are present because the algorithm cannot use satellite predictors of low clouds/stratus there. Both the Day/Night band and the brightness temperature difference field give information about the top of the cloud deck — it’s hard to infer how the cloud base is behaving. The addition of Rapid Refresh model information on low-level saturation helps better define where IFR conditions are present.

As above, but at 0945 UTC 9 January 2014. Suomi/NPP data is a toggle of Day/Night band and 11.35 µm – 3.74 µm brightness temperature difference (click image to enlarge)

By 0945 UTC, above, the time of the next Suomi/NPP overpass, the high clouds have started to move eastward out of Oklahoma. Consequently, satellite data can be used as one of the predictors in the IFR probability field, and IFR Probabilities over Oklahoma increase. By 1215 UTC (below), higher clouds have east out of the domain, and IFR Probabilities are high over the region of reduced ceilings/visibilities over Oklahoma. The algorithm continues to show lower probabilities in regions over Kansas where the Brightness temperature difference signals the presence of low stratus/fog but where IFR conditions are not present. Again, this is because the Rapid Refresh model in those regions is not showing low-level saturation.

As above, but at 1215 UTC 9 January 2014 (click image to enlarge)

Note in the imagery above how the presence of high clouds affects the GOES-R Cloud Thickness. If the highest cloud detected is ice-based, no cloud thickness field is computed. GOES-R Cloud Thickness is the estimated thickness of the highest water-based cloud detected. If ice clouds are present, the highest water-based cloud cannot be detected by the satellite. Cloud Thickness is also not computed during twilight conditions. Those occurred just before the first image, top, and about an hour after the last image, above.

IFR Probabilities in Extreme cold, Continued

GOES_IFR_Reg_11-3.9_Sat_20140106_0231loop

Toggle between GOES-R IFR Probabilities from GOES-13 and GOES-13 Brightness Temperature Difference (10.7 µm – 3.9 µm), with surface observations and ceilings plotted, ~0230 UTC on 6 January 2014. (click image to animate)

The coldest air of the season has plunged into the central part of the US. And as noted before, extreme cold does have an influence on the IFR Probability fields because of how the pseudo-emissivity is computed. Consider this effect of cold on the fields as you interpret them. Note also that in the daytime, when visible imagery can be used to augment the cloud mask, IFR Probabilities are low in very cold airmasses.

IFR Probabilities over the Pacific Northwest with a front

GOES-R IFR Probabilities from GOES-15 (upper left), GOES-15 Brightness Temperature Difference (10.7 µm – 3.9 µm) Fields (upper right), GOES-R Cloud Thickness from GOES-15 (lower left), Suomi/NPP Brightness Temperature Difference (lower right), times as indicated, 3 January 2014 (click image to animate)

One benefit of the GOES-R IFR probabilities is its consistency from hour to hour. In the animation above, the region of higher IFR Probabilities associated with a southward-propagating front over Oregon shows good hour-to-hour consistency. In contrast, the Brightness Temperature Difference field (upper right in the figure) suffers from the presence of higher clouds (denoted in the enhancement by darker regions). As the high IFR Probabilities expand southward into southern Oregon, reported visibilities/ceilings decrease towards IFR conditions. In the animation, regions of high clouds show up in the Cloud Thickness product as regions of missing data: Cloud Thickness is of the highest water-based cloud. If the highest cloud detected by satellite is mixed-phase, or ice, Cloud Thickness is not computed. Cloud Thickness is also not computed in the 1600 UTC imagery because that time is near sunrise, and cloud thickness is not computed in twilight conditions.

Suomi/NPP provided a view of the scene as well, and the Day/Night Band showed the band of frontal clouds well. The brightness temperature difference field suggests that the cloud band was not necessarily low cloudiness, although the higher IFR Probabilities (and reduced ceilings and visibilities) testify to the presence of low clouds underneath the middle- and higher-level clouds.

GOES-R IFR Probabilities from GOES-15 (upper left), GOES-15 Brightness Temperature Difference (10.7 µm – 3.9 µm) Fields (upper right), GOES-R Cloud Thickness from GOES-15 (lower left), Toggle between Suomi/NPP Day/Night band and Brightness Temperature Difference (lower right), ~1000 UTC, 3 January 2014 (click image to enlarge)

Fog over snow in the upper midwest

Suomi/NPP Brightness Temperature Difference (11.35 µm – 3.74 µm) Brightness Temperature Difference and Day/Night Band, 0840 UTC 27 December 2013 (click image to enlarge)

At times of low lunar illumination, it can become increasingly difficult to discern regions of clouds and snow in the Suomi/NPP day/night band, shown above, toggling with the brightness temperature difference. Nevertheless, careful perusal of the image reveals cloud edges over northeastern Wisconsin and through central lower Michigan that are confirmed by the brightness temperature difference field. In contrast, the whiter region that stretches southwestward from Des Moines towards extreme northeastern Kansas has no signal in the brightness temperature difference field. This is snow on the ground (vs. little snow to the northwest).

GOES IFR Probabilities computed with GOES-13 data, hourly from 0615 UTC through 1302 UTC, 27 December 2013 (click image to enlarge)

IFR Probability fields and GOES-Based brightness temperature difference fields are produced to aid in the detection of low clouds. In the animation above, higher IFR probabilities are centered on north-central Wisconsin where, intially, IFR conditions are not quite met (according to the plotted observations of ceiling and visibilities). However, as the night progresses, ceilings lower and visibilities decrease as IFR conditions do develop in regions where IFR Probabilities are high. The IFR probabilities roughly overlap the region where IFR conditions exist.

Note that encroachment of higher clouds in from the west, starting around 0800 UTC, means that satellite data cannot be used in the IFR Probability algorithm. Because only model data are used, IFR probabilities drop from values at/above 80% to values between 50 and 60% even as IFR conditions come to be more widespread. For this reason, it is important when interpreting IFR Probabilities to be alert to the presence of high clouds.

IFR Probabilities give a much more better approximation of where fog/low stratus may be occurring than a simple brightness temperature difference field. The toggle between the GOES-R IFR Probabilities and the GOES-13 Brightness Temperature Difference, below, gives testimony to this.

IFR Probability fields in extreme cold

GOES-R IFR Probabilities from GOES-15 with surface ceilings/visibilities and GOES-15 Brightness Temperature Difference (10.7 µm – 3.9 µm) Fields with surface plots at 1800 UTC 26 December 2013 (click image to enlarge)

The image toggle above shows IFR Probability and the Brightness Temperature Difference Field over northern Alaska. Plotted METAR observations show very cold surface temperatures in the -30 to -50 F range. At such cold temperatures, the pseudo-emissivity computation can become noisy because a very small change in 10.7 µm radiance (used to compute the 3.9 µm radiance) can cause a large change in 3.9 µm brightness temperature. (The effect is shown graphically below — a small change in radiance at 3.9 µm leads to a very large temperature change). This noise can lead to a speckled appearance to the IFR probability fields. This effect can also occur in the northern Plains of the United States when surface temperatures dip below zero.

Radiance (y-axis) vs. Brightness Temperature (x-axis) for 3.8 µm (left) and 10.7 µm (right)

(update) Below is the IFR Probability in the heart of a Polar Airmass over northwestern Ontario.

GOES-R IFR Probabilities from GOES-13 with surface ceilings/visibilities and GOES-13 Brightness Temperature Difference (10.7 µm – 3.9 µm) Fields with METAR plots at 1215 UTC 29 December 2013 (click image to enlarge)

The influence of high clouds

GOES-R IFR Probabilities from GOES-15 (upper left), GOES-15 and GOES-13 Brightness Temperature Difference (10.7 µm – 3.9 µm) Fields (upper right), GOES-R Cloud Thickness from GOES-15 (lower left), GOES-13 and GOES-15 6.7 µm enhanced water vapor imagery (lower right), all at ~0700 UTC 16 December 2013 (click image to enlarge)

High clouds in the atmosphere limit the ability of satellites to sense the presence of low clouds, as this example from December 16 2013 on the coast of Oregon demonstrates. Both the brightness temperature difference product and the water vapor imagery show signatures that accompany high cirrus. When cirrus is present, the brightness temperature difference field cannot be used to isolate regions of fog and low stratus because the satellite is detecting radiation from the highest emitting surface (the cirrus) not the fog/low stratus beneath. The IFR Probability field, however, uses both cloud information and Rapid Refresh Data, and the model data can fill in regions where satellites give no useful information, such as the lower Columbia River Valley around Astoria. Because satellite data are not used as a predictor, probabilities are lower. Remember how the presence of high clouds affects things when you interpret the IFR Probability fields.

GOES-R Cloud Thickness is not computed under high clouds. The GOES-R Cloud Thickness is the thickness of the highest water-based cloud deck. If a cirrus deck is present, or if twilight conditions are present, GOES-R Cloud Thickness is not computed.

Pseudo-emissivity at 3.9 µm

The traditional (or heritage) method of detecting fog/low clouds is a brightness temperature difference (BTD) product. The difference between brightness temperatures at 10.7 µm and 3.9 µm highlights water-based clouds because those clouds do not emit as a blackbody at 3.9 µm, so the inferred temperature (computed assuming a blackbody emission) is colder than that temperature computed using 10.7 µm radiation, because clouds emit radiation at 10.7 µm more like a blackbody.

At night, the GOES-R IFR Probabilities use a pseudo-emissivity at 3.9 µm in lieu of the 3.9-11µm BTD to highlight regions of water-based clouds at low levels. The 3.9 µm Pseudo-emissivity is the ratio of the observed radiance at 3.9 µm to a computed 3.9 µm blackbody radiance that is based on the observed 10.7 µm brightness temperature. In other words, the observed 10.7 µm brightness temperature is computed from the 10.7 µm radiance. That computed brightness temperature is then used to compute a 3.9 µm radiance that would be detected if the emitter was a blackbody.

The 3.9 µm pseudo-emissivity produces a satellite signature for low water clouds similar to the brightness temperature difference but is used instead because it is less sensitive to scene temperature. Skill scores for fog/low stratus detection are higher when the pseudo-emissivity is used in the algorithm to find regions of low clouds/fog than when the brightness temperature difference field is used.

Advection Fog over the Upper Midwest

Toggle between nighttime GOES-R IFR Probabilities from GOES-13 and GOES-13 Brightness Temperature Differences (10.7 µm – 3.9 µm) at 0215 UTC on 4 December 2013 (click image to enlarge)

Dense advection fog developed in the upper midwest on Tuesday 3 December 2013 and persisted into December 4th as a Colorado Cyclone moved into central Wisconsin, drawing moist air over cold ground. The IFR Probability Product, a product that fuses together the 3.9 µm pseudo-emissivity (nighttime only) from satellites (a signal similar to the 3.9-11 µm brightness temperature difference field, which gives little information on low clouds in this situation) with model data from the Rapid Refresh (which suggests widespread fog), accurately depicts the large region of advection fog that led to dense fog advisories over parts of Wisconsin and Iowa and surrounding states (see below).

Cropped Screenshot from http://www.crh.noaa.gov at 1414 UTC on 4 December 2013 that shows widespread Dense Fog Advisories over the Upper Midwest

Daytime GOES-R IFR Probabilities computed from GOES-13, 2145 UTC on 3 December 2013 (click image to enlarge)

The dense fog was present late in the day on December 3rd, 2013, and the IFR Probability fields reflected that. However, in the image above, there are isolated pixels with very low probabilities mixed in with the high probabilities over Wisconsin and surrounding states where advection fog was widespread. Why?

This storm had multiple cloud layers, which can make detection of low cloud difficult from satellite (above) when sun angles are low (sunrise/sunset). The IFR Probability image above is at 3:45 PM local time, and the sun is low in the sky. Deep shadows are being cast and the dark shadowed regions in the visible are misinterpreted by the cloud-clearing algorithm as clear skies. During the day the GOES-R fog/low stratus algorithm relies on the cloud mask to determine where clouds are. Where clear skies are detected (erroneously, in this case), IFR Probabilities are not calculated because fog/low stratus are not expected to be present. Thus, if you see pixelated fields such as the one above, and the sun is low in the sky, this likely means cloud shadows are causing the cloud mask to erroneously return clear sky, which in turn leads to very low IFR probabilities. The animation below cycles through IFR Probability and visible imagery (with a regular enhancement and with a low-light enhancement). After sunset, the cloud shadows are gone and the probability field fills in (as can be seen in the 0215 UTC imagery at the top of this post).

Daytime GOES-R IFR Probabilities computed from GOES-13 at 2145 UTC on 3 December 2013 and the corresponding Visible Imagery (click image to enlarge)

GOES-R Fog Product Examples

Fog detection fusing GOES, Terra/Aqua or Suomi/NPP Satellite data with Model output