Nowcasting Lake Effect Snow

Objective Objective : To create nowcasts of present weather and specialized estimates of snowfall for towns, airports, nuclear power plants east of Lake Ontario.

This scenario was engineered so that you can run GARP for the entire event.

Start NTL/GARP for the 09dec95 case dataset. The time period of interest begins just prior to 14 UTC on 11 December 1995.

We start off with a shift change briefing. I'm leaving and you're presented with a full lake effect event underway. You have a full suite of satellite, radar, SAO and spotter reports showing where the lake effect was overnight.

Since you will be at the short-range desk, you will be estimating snowfall rates from all the data sources you have available to you.

Data sources Here's a summary of the information available to you:

• GOES-8 Imagery (Ch1 1km, Ch2 4km, Ch3 8km, Ch4 4km)
• NIDS radar data up to 17 UTC (RMX, BGM) [BREF, VEL, CREF, PRET]
• Surface and upper air data
• Model grids
• Written guidelines for snowfall estimation (included below)
• Maps and selected spotter reports (included below)

Back to menu

---

1. Observe the visibility or snowfall rate for a SAO or spotter. Use the table below for a comparison of visibility to snow rate . This relationship seems to work well for this event. So far GARP does not plot visibilities for you so I'll provide them down below.	snowfall intensity light moderate heavy extreme visibility(nm) gt 0.5 .5 -.25 .25 - 1/8 1/8 - 1/16 Rate (in/hr) lt 0.5 0.5 - 1 1 - 2 2 - 4
2. Ground truth and radar reflectivity : At selected sites for which you have observations, measure the reflectivity a few km upwind of your station, using the radar scan just prior to your analysis time (about 6 min earlier). We are trying to account for horizontal drift of falling snow from the radar observations (which depends on the distance to the radar). Figure that snow falls at about 2 m/s and the lowest scan is approximately 2km above lake level. For the instantaneous observations (e.g., SAO's) you can use a single reflectivity image. For spotter reports measuring snow over the past hour, you might want to look at a few sweeps and come up with a representative reflectivity.	In this example, I will pick Buffalo,NY. You can use GARP to locate a site by lat lon as well. In this case, I'm measuring the reflectivity right over Buffalo. However, I should measure values southwest of Buffalo by 5 to 10 km. The mean wind in the convective layer is southwesterly.
3. Radar reflectivity and satellite cloud tops : For the same area and time that you measured the reflectivity in the last part, now measure the GOES Ch4 IR cloud top temperature in a small area around your point. Again, one single image is good for an instantaneous ob but for a spotter report, you might want come up with a representative cloud top temperature for the time period of the observation. Also, look at other GOES-8 bands to determine whether there's glaciation occuring at cloud top and whether there is high visible reflectivity (e.g, deep or dense cloud). These indications will help you affirm that this cloud is producing snow.	Now I can see what cloud top temp corresponds to the reflectivity I chose in the last step. Again, I should've checked for a representative cloud top temp southwest of Buffalo by closer than 10km. Notice that there are numerous clouds with cold CTT's but are not showing up on radar. The visible image shows the brightest values that correspond more closely to the radar echos. Notice that the reflectivity image shows areas where cloud glaciation is occurring over the band. But not everywhere. GOES-8 VIS GOES-8 Reflecitivity image

Extending snow estimates beyond radar range: Now that you have an association from the ground report to the satellite-based cloud top temp you can estimate snowfall rates beyond radar range. The only (major) problem is there usually is much more cold cloud coverage than actual snowfall due to anvils, advective clouds, cirrus... Now you need to look at multiple bands to help you narrow down the areas where snow maybe falling.

Back to menu

After outlining of CTTs correlated with 1"/hr snows...
I check for areas of higher albedos --> deep clouds.
Then I outline areas of partial and complete cloud glaciation as shown by the 3.9um reflectivity image.
Then I overlay all the relevant composites. Where the high albedos overlay the CTTs corresponding to the highest reflectivities is where the heaviest snow is most likely occuring. The low 3.9um reflecitivies indicate the existence of glaciation and snow production. However, heavy snow can occur with medium or even high 3.9um reflectivities. Usually there is some amount of detectable cloud glaciation. At night, the 'fog product' is also a good detector of cloud phase but there is increasing noise with decreasing CTTs. Average imaging can also be useful for removing some of the transient clouds. Pull up this average image from 1545 to 1645 UTC. Finally, where all satellite-based indications of heavy snow reside, and the mean flow is upslope for at least several km is where the heaviest snow will fall.

Back to menu

On a sheet of paper, compare the

• radar reflectivity,
• GOES-8 10.7 micron cloud top temps,
• what the 3.9 micron reflectivity indicates,
• and what the relative visible brightness indicates

to the snow fall rate observed by spotters for Fulton,NY. Do the same for the METAR/SAO sites at Syracuse (SYR) and Rome (RME).

1400 UTC Obsevations below.

Here are the locations for each town I mentioned.

Continuing with the above example, this is what I would have (for a different date though)

Back to menu

---

Scenario at 1815UTC

Radar data is now unavailable.

Estimate the visibility (n. mi.) for Syracuse, NY (SYR) and Utica, NY (UCA). For Fulton, NY (43.19N, 76.25W), estimate the snowfall rate ("/hr).

  City    Time    Snow rate    Visibility

___________________________________________
___________________________________________
___________________________________________

Don't click here until you finished the 18UTC scenario.

Back to menu

---

Scenario at 2130UTC

Radar data still out...it appears that heavy snow collapsed the KRMX dome :)

The Oswego Co. Snow Removal Dept. called and would like a fax of a map of where to expect heavy snow at this time. Use the composite technique I showed in the example above . You should also have a road map copy where you can draw your composite.

BREAK
Now you have just gotten a critical ob from 9 Mile Pt. ( lat lon ) Nuclear Power station here. Do you want to revise your area of heavy snow?

Back to menu

---

OBSERVATIONS

1400 UTC Reports :

City      Time (UTC)   Snow (in)   Period (hr)    

Fulton      1345           5           1         
Fulton      1410          10           6  
Hannibal    1300           6           4        
Hannibal    1300           4          1.5  


Station   Time (UTC)   Visibility (n. mi.)   Present Wx 

  SYR       1350             0.2                 S  
  UCA       1350             0.5                 S  

Back up to 1415UTC

---