Archive for the ‘Uncategorized’ Category

I’ve had some opportunity to view the NESDIS Snowfall Rate (SFR) Products today, in particular, to see how it performs during the central Plains/Midwest snowstorm.  These products are being delivered by SPoRT to several collaborative offices in the CONUS and Alaska for evaluation during the current winter.

Background info:  the Merged SFR product contains NSSL Multi-Radar Multi-Sensor (MRMS) precipitation data with insertions of polar-orbiter derived precipitation rate data when those are available.  The precipitation rate data from the polar-orbiters is available in AWIPS in individual swaths or contained within this merged product (in the merged product, the MRMS data replace the polar-orbiter data).  The data are available in AWIPS as liquid equivalent values or a snowfall rate with three distinct snowfall-to-liquid ratios: 10:1, 18:1, 35:1.  To learn more about this product, you may click here to see training material provided by researchers at NESDIS and SPoRT.

So, let’s take a quick look at some of the data today and I’ll share a few comments and thoughts.  This first image is the Merged SFR product valid at 1130 UTC with METAR plots (yellow) at 12 UTC.


Image 1.  NESDIS POES Merged Snowfall Rate (10:1) valid 1130 UTC 2 Feb 2016, METAR plot valid 12 UTC 2 Feb 2016.


Without any polar orbiting data available at this time, this image contains only the MRMS precipitation data.  In the image (Image 1), notice the band of heavier precipitation stretching roughly west-east across southern Nebraska and Iowa, and the relatively tight precipitation gradient in southern Iowa.  At the time of this image, notice no snowfall was occurring at the Des Moines location, per the SFR product or the 12 UTC METAR.  Pay particular attention to the discrepancy in times between the METARs and the SFR product at this point…there is a 30-minute offset.  Now, let’s look shortly later as a swath of polar orbiter data became available.


Image 2. NESDIS SFR Merged product valid 1140 UTC, NESDIS SFR swath data valid 1145 UTC, and METAR plots valid at 12 UTC 2 Feb 2016.

I have layered the imagery so that the polar imagery swath data are laid atop the Merged SFR product.  Notice that the polar orbiter derived data indicate a band of relatively heavier precipitation spreading northward into Nebraska and Iowa.  This is important because the polar orbiters observe precipitation within the clouds on average ~30 minutes before it manifests at the surface.  In this image (Image 2), notice that this band of heavier precipitation has now spread northward to include Des Moines and points to the west of there, where little to no precipitation was occurring earlier.  So, the NESDIS polar data suggested significant snowfall production was translating northward within the mid/upper cloud layer.  Knowing the data typically offer about a 30 minute lead time for observations at the surface, a forecaster could have surmised something about precipitation production aloft, intensity and overall storm evolution while obtaining more data about timing to impacts at a metro area.

The next image shows the timing of the arrival of the precipitation at Des Moines  per the merged SFR product and the Des Moines surface observation (Image 3).


Image 3.  NESDIS Merged SFR product valid 1230 UTC, METAR plots valid 1300 UTC 2 Feb 2016.

In image 2, remember that the SFR swath data indicated high snowfall rates, >1 inch/hr (per the 10:1 ratio…which may be understimated) directly over Des Moines and surrounding areas at 1145 UTC, while the Merged SFR above (Image 3) shows precipitation finally entering the city and the observation site at ~1230 UTC.  Notice that the Des Moines METAR showed light snow during the 1300 UTC observation (Image 3).

Let me point out something important here.  In the Merged SFR product, the satellite derived data are purposely delayed 30 minutes for insertion into the official delivered product.  This was decided as the configuration of the official product since precipitation in the satellite derived data typically precede the arrival of precipitation at the surface by about 30 minutes.  The thinking being that this apparent discrepancy would be observed between the MRMS data and the satellite derived data, and would lead to forecaster confusion.  That is understandable, especially for this latest experimental iteration of the SFR product.  However, after viewing these data in a few cases, I think it is advantageous that the satellite derived data contain important information about the evolution of snowfall and precipitation production aloft, well before it manifests at the surface.  The fact that satellite derived observations of precipitation rates precede the occurrence of snowfall at the surface by about 30 minutes, and if you noticed, by about one hour in this case, makes these satellite derived swath data operationally relevant and important.


Read Full Post »

This past weekend’s storm which brought record-breaking snow to the Mid-Atlantic and Northeast Corridor also brought something that gets the Earth Science Office at Marshall Space Flight Center (MSFC) excited…lightning from the view point of a camera lens aboard the International Space Station (ISS).

NASA Commander Scott Kelly (@CDRScottKelly) tweeted out this photo early Saturday morning from an overflight down the East Coast just before sunrise.


The corresponding satellite and lightning data show that the ISS camera captured a 4 stroke incloud lightning flash within the storm as the system pushed its way out to sea in the North Atlantic.


GOES East IR imagery from 0945 UTC on 23 January 2016. Red plus signs indicate the location of 4 incloud strokes as observed by the Earth Networks Total Lightning Network that represent the location of the flash in the ISS photo from Saturday.

Over the next year the weather enterprise will expand its capability to monitor lightning flashes from space in a similar manner to how the ISS captured this lightning flash. In the next year, two spaceborne lightning measurement instruments which NASA MSFC has played a major role in developing during many decades of hard work will be launched into space: the International Space Station Lightning Imaging Sensor (ISS-LIS) and the GOES-R Geostationary Lightning Mapper (GLM). These instruments will monitor energy from lightning flashes escaping the top of the cloud when a lightning flash occurs, utilizing a narrow oxygen emission line at 777.4 nanometers.

What does this mean for the public? Increased public safety and confidence in decisions which are affected by hazardous weather. Data from the ISS-LIS and GLM instruments will help scientists better understand the internal structure of all types of storms, helping develop better models for how storms grow, intensify and decay. Forecasters will be able to utilize flash rate information on storms acquired from these instruments to enhance severe weather prediction, determine where the heaviest snowfall rates are occurring in winter systems, or help reroute air traffic away from dangerous storms over the ocean. Most importantly, the ability to monitor the area of individual flashes will lead to better decisions on how to take shelter in an appropriate amount of time before the first lightning strike occurs in their area.

A special thank you to Mike Trenchard, Will Stefanov of Johnson Space Center for helping us acquire the ISS telemetry and camera information used to sync the meteorological observations with the lightning photo from Commander Kelly.

(Posted on behalf of the Earth Science Office)

Read Full Post »

The current winter storm unfolding across the eastern U.S. may be the storm that launches a thousand blog posts.  Well, maybe at least two or three on this site.  We’ll see.  Anyway, there may be multiple opportunities to assess the utility of a variety of Proving Ground (PG) data sets and imagery during this event.  Although not a PG suite of products, I’m going to start by taking a quick look at the SPoRT LIS soil temperature data that are being delivered to several NWS offices, including my own (Huntsville).

Below is a brief loop of 0.5 degree composite reflectivity imagery from the NWS NEXRAD network.  The great thing about AWIPS II, is that with some manipulation of time options, the radar loop can be layered over the latest LIS soil temperature data.  In this case, the latest data available were from 09Z this morning.


NASA SPoRT LIS 0-10 cm soil temperatures (image, F) valid at 09Z 22 Jan 2016, overlaid with WSR-88D 0.5 degree composite reflectivity, 1324Z-1512Z.  Surface observations in yellow. 

The soil temperature data have been color coded so that white colors represent the transition zone where temperatures are around freezing (28F to 34F), per the model, and impacts may be occurring or beginning to occur.  Of course, there are other factors to consider, such as precipitation type at the surface, and precipitation intensity.  If precipitation is liquid, then soil temperatures in the upper 20s will likely translate into  an icing event.  However, if temperatures are at or just below freezing, latent heat release generally will warm the very shallow layer near the surface, leading to eventual melting of any ice that accumulates.  If soil temperatures are above freezing, say around 34F, snow can still accumulate on the surface, as long as the snowfall rate exceeds the snow melt rate.

Looking at the data and imagery above, it would appear that much of Tennessee is under threat of some type of freezing precipitation at the surface, whether from accumulating snow or freezing rain.  Taking a closer look, the observation at Knoxville starts out at 36F in this loop, and finishes at 35F by the end of the loop with rainfall being reported.  Soil temperatures there are right around 32F to 33F at closer inspection.  So, no real good chance of freezing precipitation at the surface.  But, the data show forecasters that ground temperatures will not have to fall much before there are potential issues, especially as colder air begins to move into that area.  However, per reports from Nasvhille and many surrounding locations in middle and western Tennessee, snowfall is accumulating rapidly where soil temperatures (per closer inspection in AWIPS) are around 31F to 32F.  Meanwhile, reports of accumulating snowfall have been received from portions of northern Mississippi and southwestern Tennessee, where 0-10 cm soil temperatures are generally in the 34F to 38F range.  So, these are locations where the snowfall rate certainly exceeded the melt rate of the snow at the surface.  Nevertheless, the ground will continue to radiate these higher temperatures into the shallow snowpack and some melting from beneath will likely occur through the day.  These are other ways the data can be used…to determine the potential length of lingering impacts from recent snowfall.  Some limitations to the data currently?…they are only at 3 hourly resolution and have about a 2-8 hour latency.  We might want to reconsider higher temporal resolution during times of inclement winter weather, where more rapid updates would be more operationally useful.  Anyway, this is something for the SPoRT modeling team to explore.

These data are becoming a very valuable tool during operations at our office and several others.  Just yesterday, the Raleigh NC NWS office cited the data when assessing the potential for accumulating snow and ice in their forecast area.


Read Full Post »

On January 10 2016, an arctic front pushed east across our forecast area. Rain, associated with the front, mixed with snow and then changed to all snow from west to east during the afternoon and evening hours. Patchy snow showers or flurries then continued into the nighttime hours.

I have attached an SFR image (SFR_2233Z_011016.jpg) from 533 PM EST (2233 UTC). I have also attached two other images: RLX radar (Radar_2236Z_011016.jpg) and Surface Observations (Surface_23Z_011016.jpg). In general, the SFR data matches up quite well with the radar data as well as the surface observations.

The only area where the SFR data does not match up with the other observations is across eastern Kentucky and southern West Virginia. KBKW (Beckley WV) and KPBX (Pikeville KY) both reported
light snow while K6L4 (Logan WV) reported unknown precipitation. In contrast, the SFR data did not indicate any snow in the clouds.




Read Full Post »

So, the opportunity arose last evening to use the GOES-R Fog and Low Stratus Product (GOES-R FLS, produced by NOAA/NESDIS and UW-CIMSS) since I had aviation forecast duties.  Shortly after coming on shift, it was noticed that Marginal Visual Flight Rules (MVFR) conditions were forecast for the short-term at the KHSV Terminal Aerodrome Forecast (TAF) site in north central Alabama.  However, MVFR cloud bases were not occurring and had not quite developed as expected.  A look at the Fog and Low Stratus indicated that the low cloud bases were likely diminishing in coverage to the southwest of the KHSV TAF site (north central Alabama).  Other satellite, ground observation, and model data also suggested that this trend was likely to continue…at least for the next several hours.  I did like the fact that the GOES-R FLS product allowed for a relatively quick and effective assessment of the upstream conditions.  I was able to do a quick update to the TAF for the Huntsville airport partly due to the data/imagery made available by the FLS product.


Loop of GOES-R Fog and Low Stratus Product (MVFR Probability) from 19-2030 UTC 10 December 2015. The scale is located at the top left.  Notice that probabilities of MVFR conditions are generally low in northern Alabama and can be seen to decrease in coverage in upstream areas to the S-SW.  Ceiling/Vis surface observations are included in green. The observations from the Huntsville, AL airport (KHSV) can be seen in far north-central AL.



Read Full Post »

Ok, so maybe I’ll have some issues with copyright infringement here.  If the name of the post changes, then you’ll know.  Otherwise, no, this is not a new attempt at the Anthony Burgess novel nor is it an attempt to rewrite the script of the Kubrick movie.  Alas, this is a tale of flooding and soil moisture.  Yes, perhaps a little less dramatic than some setting in a futuristic dystopia.  Anyway, I’ll get to why I titled the post as I did eventually.  So, let’s get to it.

At the end of November and early in December, many locations in the Tennessee Valley received a fairly substantial rainfall.  The first image below shows the rainfall total for a 5-day period ending at 12Z on 2 December 2015.


Image 1. NEXRAD Level-III KHTX (Hytop, AL) Storm Total Precipitation for the period beginning 0843 UTC 27 Nov, and ending 1204 UTC 2 Dec 2015.

This image (Image 1) is centered on northern Alabama and surrounding locations.  Notice that amounts around 2.5 to 4 inches were common across much of the area.

So, let’s see what transpired during this event, and how soil moisture (per the SPoRT LIS) and streamflows responded during this more flood-prone time of the year.  The next image below shows the SPoRT LIS 0-200 cm Relative Soil Moisture (RSM) on the morning (12Z) of November 30th.  Values around 55% were common after the initial bout of rainfall (~1.5-2 inches) on the morning of the 30th.  Local use of the data at the NWS Office in Huntsville indicates that the risk for areal and stream flooding increases significantly once 0-200 cm RSM values surpass 55% coincident with >2.00 inches of rainfall.  As can be seen in the image below, values had reached 55% after the first round of rainfall.  All that was needed was an additional 2 or more inches of rainfall in about two days or less to put some streams at risk of flooding.


Image 2.  Close-up view of NE Alabama and adjacent areas, showing SPoRT LIS 0-200 cm Relative Soil Moisture, 1200 UTC 30 Nov 2015.  Urban areas and lakes are masked (black).  The stream basins that eventually went into flood include some of those in and around Huntsville and between Huntsville and Scottsboro. 

The area did receive the additional ~2 inches of rainfall.  In fact,  take a look at the rainfall that occurred over subsequent days, from 12Z on Nov 30th through 12Z on December 1st, in the next two images below.  Again, we’re focusing on northern Alabama here (sorry about the multiple scales and domains used in the series of images).


Image 3.  24-Hour StageIV Precipitation ending 12 UTC 1 December 2015. 


Image 4.  24-Hour StageIV Precipitation ending 12 UTC 2 December 2015. 

Additional rainfall totals in our basins of interest was about 2-3 inches.  So, how did rivers and streams in basins with >55% 0-200 RSM respond after the 2nd batch of heavy rainfall?  Well, several streams in these basins reached minor flood stage.  I won’t show all of the stream hydrographs here, but will show a representative case: the hydrograph from the Paint Rock River as measured at Woodville, AL (Image 5).


Image 5.  Paint Rock River hydrograph, as measured at Woodville, from about 18 UTC 28 November to 08 UTC 3 December 2015.  The blue dotted-line shows the river stage, the yellow, orange and red lines indicate action, minor and major flood stages, respectively.  Notice that the river rose above minor flood stage at about 00 UTC 2 December. 

In the hydrograph above, notice the sharp rise that took place on the 30th, eventually leading to the rise above minor flood stage on the evening of December 1st.  Forecasters here who are used to using the SPoRT LIS data in this manner had relatively high confidence in at least a minor flooding situation along this river and perhaps a couple of other rivers in the area.  As noted the the Area Forecast Discussion from the HUN office during the afternoon of November 30th, “LIS DATA OUTPUT FROM NASA SPORT WAS SHOWING SOIL MOISTURES OF AROUND 50 TO 60 PERCENT, WHICH COUPLED WITH AN EXPECTED ADDITIONAL 2 INCHES OF RAINFALL SHOULD LEAD TO A MAJORITY OF THE RAIN OCCURRING AS RUNOFF AND LEAD(ing) TO MINOR FLOODING…”

So, the use of these data for this purpose at NWS HUN has become routine, and important rainfall and soil moisture thresholds are now fairly well established.  Until a more robust and objective statistical study is conducted, involving inputs of soil moisture and rainfall and outputs of stream rises, we’ll continue to rely on these crude thresholds…which suffice in most cases for our purposes.  But, the river rises in some of our basins here, given these inputs of rainfall and soil moisture has become like clockwork, leading to increased confidence in eventual river stages, and importantly, when flood stage is likely to be reached (the thin orange line on the graph).  Thus, “Like Clockwork Above Orange”.  Ok, maybe that’s pretty corny, but it’ll have to work for now.


Read Full Post »

Viewing low-level clouds at night can be a problem when utilizing standard IR imagery available through GOES or most polar orbiters, especially when widespread high clouds are present.  However, Day-Night Band imagery available through the VIIRS instrument onboard the Suomi-NPP satellite offers this ability.  Well, I should add…when sufficient moonlight is present.  Last night was one such night.  First, let’s take a look at the 11-3.9 channel difference product (commonly known to forecasters as the “fog” product) with the standard yellow/gray/blue color curve (Image 1).


Image 1.  Suomi-NPP VIIRS 11-3.9 um product, 0717 UTC 30 Nov 2015.

Some low/mid clouds (yellows to medium gray) can be distinguished from higher clouds (blues) in portions of the image, especially over parts of the interior South.  However, it’s difficult to tell the extent of the low clouds with much confidence.  By the way, the values/symbols in white are automated ground observations.  the number at the bottom of the circles represents visibility (statute miles), while the numbers to the left indicate cloud base heights (in hundreds of feet).

Now, take a look below at the VIIRS Day-Night Band Reflectance product.


Image 2.  Suomi-NPP VIIRS Day-Night Band Reflectance product, 0717 UTC 30 Nov 2015

In the Day-Night Band Reflectance image above (Image 2), the extent of the clouds is more easily discernible.  However, it can be difficult to differentiate low clouds from high level clouds.  This is why I generally prefer the Day-Night Band Reflectance RGB for these purposes (Image 3).


Image 3. Suomi-NPP VIIRS Day-Night Band RGB product, 0717 UTC 30 Nov 2015

Notice in Image 3 above, it’s much easier to distinguish the low clouds (grays/yellow-grays) from high clouds (blues) due to the color-coding provided in the RGB.  Now, a forecaster can easily see the extent of the low cloud deck across the Southeast, and can tell where this line ends from southern Louisiana, trough central Mississippi, Alabama and into northern Georgia.  If you look back at Image 1, it would have been impossible to tell where the terminus of low clouds was situated.  Sure, observations (shown in white) will provide some clue for forecasters, but in areas with a dearth of observations, there would be no way to know for sure.

Read Full Post »

Older Posts »


Get every new post delivered to your Inbox.

Join 1,518 other followers