1. HiRAD: A Real-Time System to Estimate Weather Conditions at High Resolution Presented to the WG/WIST … June 7, 2006

2.  HiRAD is a system to derive accurate, high-resolution surface weather observations  Uses multiple inputs, including in situ and remote sensors  Sensible weather estimates are central to the system  Phase II and III operational at TWC  On Local on the 8’s and weather.com HiRAD Summary

3. HiRAD System Estimate 2 Estimate 1 Final Wx Estimate Post Processing Radar Calibration & Combination Radar Fingerprinting Lightning RUC 2.5 -km Downscaling Satellite Data Corrections with recent obs Climo Data Surface Observations HIRAD

4.  Operational today @ 10,000 sites  2.5 km resolution gridded version is operational and flowing to MET systems  Observations updated every 20 min; processing in 5 min’s or less.  Gridded data soon available on many platforms  These samples are phase III gridded data Operational System

5. 81 Tiles running on 33 AMD dual-core server farm:

6. 2.5 km Grid intersections in metro DC 2.5 km Grid intersections in metro DC

7. National Scale Examples

8. HiRAD Local Examples Front Range ChinookLake Effect Snows

9. DDEQC Schematically.. DEC-05 DDEQC statistics for all METAR ICAO’s used within HiRAD SiteVarStdErrScaling KATLT3.74F 2.5 KATLTd2.90F 2.5 KATLU4.12 mph 3.0 KBOST… etc… METAR (LDM/NOAAPort) Feed KLWD 221953Z AUTO 04005KT 4SM BR OVC001 M01/M02 A3029 RMK AO2 SLP272 T10061017 TSNO $= KMLU 221953Z 34006KT 10SM -RA BKN006 BKN010 OVC016 08/07 A3011 RMK AO2 CIG 003V007 SLP196 P0016 T00780067= KODX 221953Z AUTO 01004KT 10SM CLR 00/M06 A3032 RMK AO2 SLP294 T00001061= KSPS 221952Z VRB03KT 10SM FEW018 SCT060 BKN100 10/M01 A3021 RMK AO2 SLP230 T01001006 $= K9V9 221952Z AUTO VRB03KT M03/M10 A3035 RMK AO2 SLP301 T10331100 PWINO FZRANO TSNO $= KTOL 221952Z 04004KT 10SM CLR 03/M05 A3038 RMK AO2 SLP296 T00281050= DDEQC & other QC Filter Logic Rejection/Inspection Log.. ddeqc2006 012512050 7.log KIPJ (72314038) DewPoint1142 UTC3.20000026.1434402.5000008.000000 Not Available (-99999) ddeqc2006 012512250 5.log KFQD (72315045) DewPoint1200 UTC3.20000025.9037972.5000008.000000 Sunny (3200) ddeqc2006 012512250 5.log KIPJ (72314038) DewPoint1202 UTC3.20000027.3936202.5000008.000000 Not Available (-99999) FAIL ? PASS ? Core HiRAD Analysis

10. Bad wind gust.. KEVV 251754Z 27009KT 10SM SCT027 03/M04 A3042 RMK AO2 SLP304 T00331039 10033 21017 50015 KEVV 251654Z 32010KT 10SM CLR 02/M04 A3042 RMK AO2 SLP305 T00221044 KEVV 251554Z 31010KT 10SM CLR 01/M05 A3041 RMK AO2 SLP301 T00111050 $ KEVV 251454Z 32011G142KT 10SM CLR 00/M05 A3037 RMK AO2 SLP289 T00001050 51023 KEVV 251354Z 29011KT 10SM CLR M01/M06 A3035 RMK AO2 SLP281 T10061056 KEVV 251254Z 29008KT 9SM CLR M01/M06 A3033 RMK AO2 SLP275 T10111056 $ Bad wind speeds or gusts generally attributed to typos (?), and less to instrumentation. They are rare but often egregious.

11. A typical day in DDEQC.. On January 25, 2006.. Fairly typical day; i.e. representative of total sample. Total Observations: 75,781 Flagged/Withheld:141 Hit Ratio0.18% Unique ICAO43 VariableCount % T5841.1% T d 7956.0% U (sust. and gust)4 1.8%

12. WIST thoughts w.r.t. HiRAD..  Variational assimilation is a powerful technique  Shelters or isolates raw OBS on the input side.. however, single bad observations can now effect areas instead of single points..  Results can be provided as uniform outputs in time and space (grids)..  Downstream applications need only know their actual location on the earth’s surface and understand how to relate this location to the gridded data.

13. WIST thoughts w.r.t. HiRAD (cont)  The sensible weather (present weather, visibility, precipitation type and rate) should receive the lion’s share of attention  Is it possible to have an equivalent of the NLDN or USPLN for other weather variables? That is, very high resolution and very reliable and accurate, but low cost and only a sparse network of actual instruments?  Latency. Start with 0.00 minutes and resist every second of time that is added to the total delay and the resulting time of dissemination.

14. WIST thoughts w.r.t. HiRAD (cont)  We fear adding mesonet and other secondary obs sources to HiRAD because of bias, reliability, and q/c implications.  The knock on road sensors is reliability and quality control.  If these obstacles are overcome, it is an extremely valuable source of information.  Other gee-whiz sources must be demonstrated to be reliable, robust, and unbiased. Even 1% error rates are unacceptable and will eat you alive.