Aaron Reynolds WFO Buffalo Introduction All NWS radars have dual polarization capability. Dual Pol Expectations: Ability to determine Precip type. More info about intensity Drop/particle size AND Better.
Download ReportTranscript Aaron Reynolds WFO Buffalo Introduction All NWS radars have dual polarization capability. Dual Pol Expectations: Ability to determine Precip type. More info about intensity Drop/particle size AND Better.
Aaron Reynolds WFO Buffalo
Introduction
All NWS radars have dual polarization capability. Dual Pol Expectations:
Ability to determine Precip type.
More info about intensity
Drop/particle size
AND
Better Precipitation estimates...for RAIN
However...a NON-dual polarization equation is used for snow.
Introduction
0.5 degrees
Freezing level
Radar samples “RAIN” dual Pol QPE.
Introduction
Radar samples “SNOW” Pre dual Pol QPE. 0.5 degrees
Freezing level
Radar samples “RAIN” dual Pol QPE.
The Problem
• WFO CLE found: • High QPE bias • Primarily cool season • Above freezing level • Based on DP QPE only – would have led to issuance of flood warnings
The Problem
Before Dual Pol Non-Dual Pol QPE
The Problem
Before Dual Pol 1.04 in Youngstown Non-Dual Pol QPE 1.27 in, Lyndonville 1.11 in, Chili After Dual Pol Both show overestimates, but Dual Pol is MUCH worse (higher) What happened?
1.04 in, Youngstown Dual Pol QPE 1.27 in, Lyndonville 1.11 in, Chili
Hypothesis
Overestimate of QPE when the lowest radar slice samples above the melting layer (Cocks et al. 2012). Radar classified areas above the melting layer as “dry snow’”. Multiplied by 2.8 to derive QPE.
Station Selection
13 gauges identified Requirements:
Knowledge of gauge type.
Track record.
Proper exposure.
Record to hundredth of an inch.
10 -100 km range.
Mt. Morris, NY
Finding Events.
Event requirements:
Cold season months of October thru April.
Five gauges >= 0.10 for an event.
Data Collection
Dry snow
Data Collection
Dry snow
QPE
Data Collection
Dry snow
QPE
Gauge data.
Data collection
Brief periods of missing, or anomalous data were common which required case by case judgment. Data requirements:
90% of the hour had to be “Dry snow”.
Quality control of data
Preliminary cases were further screened for accuracy, keeping in mind gauge limitations in certain environments. Data quality requirements:
Wind >= 4 m/s 9 gauges w/o shield.
Heated tipping bucket issues.
Final check of data from cooperative observers and COCORAHs measurements.
Methodology
Calculations
A total of 383 hourly cases were identified, from 17 event days.
To calculate the dry snow coefficient we divided the dual-pol QPE by 2.8 to get a raw radar estimate.
This raw value was then compared to the actual gauge measurement, to calculate the ideal coefficient for that event.
Results
For all of the 383 cases, the average dry snow coefficient was 1.19. This was calculated from the sum of all dual-pol QPE compared to the sum of measured precipitation.
Results
QPE from Dual pol Radar compared with measured precipitation for dry snow.
Hourly Cases
383
DP Radar QPE using 2.8 dry snow coefficient [inches]
30.29
Legacy PPSE with dry snow coefficient removed [inches]
10.82
Measured Precipitation [inches]
12.90
Calculated Coefficient
1.19
Results
Results by precipitation type.
Event Precipitation Type
All Rain All Snow Mixed Events (all)
Hourly Cases
129 53 201
Calculated Coefficient
1.42
1.53
1.00
Results
Results by distance from radar.
Site Cases
Close (<75 km) Far (>75 km) 119 264
Distance (km) Ideal Coefficient (calculated)
1.0
1.3
Preliminary Conclusions
This research supports:
-2.8 coefficient is too high.
The mean coefficient:
-[1.19] may not be the ultimate answer.
Errors in the HCC: -Mixed precipitation.
-All rain/snow events 1.5 would probably be most representative.
How do we handle this?
-Additional research from other locations.
Additional Research Planned
Field test beginning this winter to test different coefficients. Several office will be participating. Any other comments or questions?