Status: Finalised Material from: Linus, Tim, Esti
1. Impact
The cyclone Stella passed the U.S east-coast on 14 March, bringing snow, sleet, freezing rain and rain on different part of the region. In front of the cyclone, a large parts of the activities in Washington D.C and New York City closed in preparation to large amount of snow. In the end most of the precipitation fell as rain in Washington (but with some freezing rain) and much less snow that expected in NYC. However, parts of upstate New York got more than 25in (~1 metre) snow.
2. Description of the event
3. Predictability
The plot below shows the accumulated snowfall from NOAA over 48 hours. During the event Central Park in NYC got 20 cm. Other snow total can be found here: snowtotals.txt
3.1 Data assimilation
The plot below shows the vertical sounding from Long Island measured 14 March 12z below, with the first guess and analysis included. The is an inversion in the profile around 1 km height that brings the temperature above the freezing point. With the cooler layer below it would have given favourable conditions for ice pellets.
3.2 HRES
The plots below show 24-hour accumulated precipitation for 14 March and the MSLP valid 12z the same day from HRES with different initial times.
3.3 ENS
The plots below show EFI and SOT for precipitation for 14 March.
The plot below shows the evolution of 24-hour precipitation forecasts for New York City for ENS (box-and-whisker) and HRES (red dot) and the model climate (red box-and-whisker).
The plots below show the probability of different precipitation types for Washington from 12 March 00z (left) and NYC from 13 march 00z (right). The plot for Washington shows a risk for freezing rain while the plot for NYC shows the risk for ice pellets.
The animation below, from DT 00Z 13th, shows "most probable precipitation type" at different instants, with associated probabilities for that type. It was created retrospectively on 27 Mar, following some SYNOP-based calibration of mimimum rate for different precipitation types (these are lower for ice-related ppn types than for rain). The algorithm shows dry (no shading) as most probable if prob of dry >50%. Darker colours highlight areas of greater confidence in the forecast type (higher probs). Clearly colours get lighter near New York, especially on Long Island, during the height of the storm, with wet snow and other types favoured at some pixels at some times. So the ENS was uncertain about ppn type, notably in southern parts of the city extent.
The plot below show the cyclone feature product for maximum snowfall.
3.4 Monthly forecasts
3.5 Comparison with other centres
The plots below show 24-hour precipitation for 14 March and MSLP valid 12z from ECMWF (left) and NCEP (middle) and UKMO (right).
+24h
+36h
+48h
+60h
+72h
+84h
+96h
+108h
+120h
The plots below shows the evolution of 24-hour precipitation for NYC (left) and Albany (right) for 13 March from ECMWF (red), UKMO (blue) and NCEP (green). For NYC the signal was picked up at a similar stage (9-10 March), while the northern extend was picked up later (11 March by ECMWF). NCEP picked up the northern extent earlier (9-10 March)) but lost the magnitude on 11 and 12 March. It is agreement with the NCEP HRES results and it is also worth noting the lower spread in the NCEP ensemble.
The plots below show the position of the cyclone in all ensemble members at 14 March 18UTC and the track from +/-12 hours. The colour of the symbol indicates the depth of the cyclone. the analysis included in red and ECMWF HRES in green. The left plots show ECMWF forecasts, middle from NCEP and right from UKMO. The ECMWF forecast had a clear majority of the members to the east of the analysis track for all three initial times. The NCEP forecast from 11 March 00z had the members more to the west but was shifted again to the east in the forecast from 12 March 00z.
+66h (12 March 00z)
+90h (11 March 00z)
+114h (10 March 00z)
4. Experience from general performance/other cases
5. Good and bad aspects of the forecasts for the event
- The forecast predicted the existence of cyclone well from 9-10 March and onwards
- HRES forecasts before 11 March missed the northerly extent of the precipitation
- Overprediction of snowfall also in the last forecasts
6. Additional material
Status:Ongoing analysis Material from: Linus, Tim, Esti
Picture
1. Impact
The cyclone Stella passed the U.S east-coast on 14 March, bringing snow, sleet, freezing rain and rain on different part of the region. In front of the cyclone, a large parts of the activities in Washington D.C and New York City closed in preparation to large amount of snow. In the end most of the precipitation fell as rain in Washington (but with some freezing rain) and much less snow that expected in NYC. However, parts of upstate New York got more than 25in (~1 metre) snow.
2. Description of the event
3. Predictability
The plot below shows the accumulated snowfall from NOAA over 48 hours. During the event Central Park in NYC got 20 cm. Other snow total can be found here: snowtotals.txt
3.1 Data assimilation
The plot below shows the vertical sounding from Long Island measured 14 March 12z below, with the first guess and analysis included. The is an inversion in the profile around 1 km height that brings the temperature above the freezing point. With the cooler layer below it would have given favourable conditions for ice pellets.
3.2 HRES
The plots below show 24-hour accumulated precipitation for 14 March and the MSLP valid 12z the same day from HRES with different initial times.
3.3 ENS
The plots below show EFI and SOT for precipitation for 14 March.
The plot below shows the evolution of 24-hour precipitation forecasts for New York City for ENS (box-and-whisker) and HRES (red dot) and the model climate (red box-and-whisker).
The plots below show the probability of different precipitation types for Washington from 12 March 00z (left) and NYC from 13 march 00z (right). The plot for Washington shows a risk for freezing rain while the plot for NYC shows the risk for ice pellets.
The animation below, from DT 00Z 13th, shows "most probable precipitation type" at different instants, with associated probabilities for that type. It was created retrospectively on 27 Mar, following some SYNOP-based calibration of mimimum rate for different precipitation types (these are lower for ice-related ppn types than for rain). The algorithm shows dry (no shading) as most probable if prob of dry >50%. Darker colours highlight areas of greater confidence in the forecast type (higher probs). Clearly colours get lighter near New York, especially on Long Island, during the height of the storm, with wet snow and other types favoured at some pixels at some times. So the ENS was uncertain about ppn type, notably in southern parts of the city extent.
The plot below show the cyclone feature product for maximum snowfall.
3.4 Monthly forecasts
3.5 Comparison with other centres
The plots below show 24-hour precipitation for 14 March and MSLP valid 12z from ECMWF (left) and NCEP (middle) and UKMO (right).
+24h
+36h
+48h
+60h
+72h
+84h
+96h
+108h
+120h
The plots below shows the evolution of 24-hour precipitation for NYC (left) and Albany (right) for 13 March from ECMWF (red), UKMO (blue) and NCEP (green). For NYC the signal was picked up at a similar stage (9-10 March), while the northern extend was picked up later (11 March by ECMWF). NCEP picked up the northern extent earlier (9-10 March)) but lost the magnitude on 11 and 12 March. It is agreement with the NCEP HRES results and it is also worth noting the lower spread in the NCEP ensemble.
The plots below show the position of the cyclone in all ensemble members at 14 March 18UTC and the track from +/-12 hours. The colour of the symbol indicates the depth of the cyclone. the analysis included in red and ECMWF HRES in green. The left plots show ECMWF forecasts, middle from NCEP and right from UKMO. The ECMWF forecast had a clear majority of the members to the east of the analysis track for all three initial times. The NCEP forecast from 11 March 00z had the members more to the west but was shifted again to the east in the forecast from 12 March 00z.
+66h (12 March 00z)
+90h (11 March 00z)
+114h (10 March 00z)
4. Experience from general performance/other cases
5. Good and bad aspects of the forecasts for the event
- The forecast predicted the existence of cyclone well from 9-10 March and onwards
- HRES forecasts before 11 March missed the northerly extent of the precipitation
- Overprediction of snowfall also in the last forecasts