Use of WSR88 D radar data to evaluate

Use of WSR-88 D radar data to evaluate regional model forecasts Sandra Yuter North Carolina State University October 2005

Outline Comparison of 3 D radar and forecast model output n Mixed precipitation n Draft plan for observations in Carolinas n 2

Use of observations for model evaluation and diagnosis What structural characteristics of storms are most important to reproduce in numerical models? n What subset of characteristics can we most reliably observe? n How do we distinguish between details that are unique to an individual storm versus those that are repeated among a group of storms? n 3

Routine Daily comparison of observations to forecast model output n Directly comparable products derived from observations and model output yield objective measures of: Confidence in forecast model output for particular storm type n Model strengths and weaknesses n Evaluation of proposed model changes n Diagnosis of error sources n 4

What should the model/observations comparison products look like? 5

Model to Observation Comparison of Surface Rainfall 1 km cloud resolving model with explicit microphysics (ARPS) of Ft. Worth Texas storm for time=0 (Smedsmo et al, 2005) 6

Volumetric comparison for accumulated storm totals 7 Smedsmo et al. (2005)

Evaluation of Model Output must be 3 D! n Surface fields necessary but not sufficient for comparisons n Operational WSR-88 D radar can provide 3 D precipitation structure and wind field information Supplemental vertically-pointing radar can provide fine-scale information on freezing level and subgrid scale variability n 8

Prototype Concepts in Portland, Oregon Area 120 km radius from KRTX 9

METEK Inc. Radar in Scholls, OR Ku-band (1. 25 cm wavelength) Cost ~ $16 K Resolution 150 m Measurements of: n Doppler velocity n d. BZ- attenuates in moderate to heavy rain 10

Microphysical and Observational Context Fall velocity Radar reflectivity Snow Layer 0 C Melting Layer Rain Layer Vertically-pointing radar Ground Surface 11

UTC Time (hours) m/s Height (m) d. BZ Variable freezing level UTC Time (hours) 12

UTC Time (hours) m/s Height (m) d. BZ Variable freezing level UTC Time (hours) 13

UTC Time (hours) m/s Height (m) d. BZ Variable freezing level UTC Time (hours) 14

What sub area around radar can we use for comparisons? Radar Visibility: n Total n Partial n Blocked 15

Interpolate Polar Radar Data to Cartesian Coordinate System WSR-88 D Precip Scan • Spatial coordinates similar to model • Minimize range dependence of radar data 16

31 Dec - 1 Jan storm Wind Field (radial velocity) 1. 0 km altitude Freq. of Echo 13 d. BZ 17

31 Dec - 1 Jan storm Wind Field (radial velocity) 3. 0 km altitude Freq. of Echo 13 d. BZ 18

17 - 18 Jan 2005 storm Wind Field (radial velocity) 1. 0 km altitude Freq. of Echo 13 d. BZ 19

17 - 18 Jan 2005 storm Wind Field (radial velocity) 3. 0 km altitude Freq. of Echo 13 d. BZ 20

B. Colle MM 5 output Distance (km) WSR-88 D Observation Distance (km) 21

Findings Orography limits radar’s visibility, compare with model over subarea of domain n Interpolate radar data to Cartesian grid n minimize range dependence n common coordinate system with model n n Variable freezing level height in winter complicates use of quantitative d. BZ statistics 22

Suggested Observed 3 D Characteristics for Forecasts of Winter Storms to Reproduce From WSR-88 D: n Wind field pattern (radial velocity) n Precipitation frequency pattern From vertically-pointing radar n Freezing level altitude (location, time) 23

PARSIVEL disdrometer: simultaneously measures diameter and fall speed of particles Löffler-Mang and Joss (2000) 24

Mc. Kenzie Bridge, Oregon (494 m MSL) MKB Observation Locations Steamboat Springs, Colorado (3200 m MSL) SPL 25

MKB 17 December 2001 Rising Temperature: 2. 5 o. C to 5. 5 o. C Rain r g Lump graupel, Locatelli and Hobbs (1974) Log 10(# of particles) Fall Velocity (m/s) Rain, Berry and Pranger (1974) d Unrimed dendrites, Locatelli and Hobbs (1974) D (mm) 26

SPL 27 February 2003 Falling Temperature: -5 o. C to -10 o. C r g Log 10(# of particles) Fall Velocity (m/s) Dry Snow d D (mm) 27

MKB 19 December 2001 Steady Temperature: 0. 0 o. C r g Log 10(# of particles) Fall Velocity (m/s) Period C d D (mm) 28

MKB 18 -19 December 2001 Falling Temperature: 0. 5 o. C to 0. 0 o. C r g Log 10(# of particles) Fall Velocity (m/s) Period B d D (mm) 29

Precipitation Classification Fall Velocity (m/s) Rain r g Not Rain d Ambiguous D (mm) 30

MKB 18 -19 December 2001 r g Log 10(# of particles) Fall Velocity (m/s) Rain Subset d D (mm) 31

MKB 18 -19 December 2001 r g Log 10(# of particles) Fall Velocity (m/s) Not-Rain Subset d D (mm) 32

Relative proportions of rain subset of particles during mixed precipitation at Mc. Kenzie Bridge, OR. Air Temp 0°C 0. 5°C-0°C 1. 5°C 0. 5°C % rain particles of total conc (# m-3) 5% 23% % rain volume of total vol. 1% 4% Average rain rate (mm hr-1) 0. 09 1. 0 93% 74% 2. 3 Statistics are calculated for subset of particles with 1 mm ≤ D ≤ 10 mm. 33

34

North Carolina Planned New Observations n Initial Phase (Winter 2005 -2006) n n n Clayton DAQ site (Spring 2006) n n n Micro. Rain. Radar at Raleigh NWSFO Precip instrument test including PARSIVEL Micro. Rain. Radar PARSIVEL precip type Mt. Mitchell (Winter 2006) n n n Micro. Rain. Radar PARSIVEL precip type Snow measurement (hopefully) 35

Long Term Possibilities n NOAA Hydrometeorology Testbed (HMT) moves to Appalachians (~2010) Likely Tar River Basin n Many observation resources n Seasonal operations over several years n n Carolina Experiment (2015? ) in coordination with NOAA HMT intensive field phase 36

Archived Level 2 WSR-88 D data is being used for research! n Preferred VCPs for precip n VCP 11 – 14 elevation angles n VCP 21 – 9 elevation angles n VCPs to avoid n VCP 121 – smears rapidly moving storms n VCP 12 – faster antenna rotation degrades Vr and Z data 37

Cooperative studies n Important to Operations n Interesting for Basic Science n Practical n Able to make quality observations n High potential to make progress 38

The End 39