Application of Numerical Weather Prediction Prognoses to Operational

  • Slides: 28
Download presentation
Application of Numerical Weather Prediction Prognoses to Operational Weather Forecasting in Hong Kong Pre-CAS

Application of Numerical Weather Prediction Prognoses to Operational Weather Forecasting in Hong Kong Pre-CAS Technical Conference on "Environmental Prediction in the Next Decade: Weather, Climate, Water and the Air We Breathe" Incheon, Republic of Korea 16 -17 November 2009 Hilda Lam Hong Kong Observatory Hong Kong, China

Content • NWP prognoses available at HKO in the past 30 years • Their

Content • NWP prognoses available at HKO in the past 30 years • Their impact on forecast performance • A few recent cases of unsatisfactory forecasts • Concluding remarks

Global model prognoses availbale at HKO since 1980’s Year NWP centres/ products Model resolution

Global model prognoses availbale at HKO since 1980’s Year NWP centres/ products Model resolution Forecast range Via 1985 ECMWF, UKMO 5 to 2. 5 deg T+72 GTS 1989 ECMWF 2. 5 deg T+72 GTS 1996 JMA GSM 2. 5 deg, 1. 25 deg T + 120 GTS 1999 ECMWF 2. 5 deg T + 120 GTS 2002 NOAA NCEP 1 deg T + 96 Internet 2004 ECMWF EPSgram Full resolution T+240 Internet 2006 ECMWF 2. 5 deg T + 24 GTS 2007 NOAA NCEP 1 deg T + 240 Internet 2009 JMA 0. 5 and 0. 25 deg T + 216 Internet ECMWF 1 deg and some o. 25 deg T + 168 Internet

Current NWP models used in HKO forecast up to (hours) model grid runs (Z)

Current NWP models used in HKO forecast up to (hours) model grid runs (Z) ECMWF 2. 5 deg 00, 12 T+240 1 deg 00, 12 T+168 0. 25 deg 00, 12 T+42 1. 25 deg 00, 06, 12, 18 T+192 T+84 0. 5 deg 00, 12 T+216 T+84 0. 25 deg 00, 12 T+216 T+84 UKMO(EGRR ) 2. 5 deg 00, 12 T+120 NCEP 1 deg 00, 12 T+240 ORSM 60 km 00, 06, 12, 18 T+72 ORSM (20 km) 20 km 00, 03, …, 21 T+42 JMA run at 12 Z others

Objective verification of weather forecasts at HKO • Current scheme used since 1984 •

Objective verification of weather forecasts at HKO • Current scheme used since 1984 • Each of the following elements is verified against actual and assigned an accuracy score: Max and min temp, rainfall, wind speed, state of sky, visibility • Weighted mean of element scores will give the forecast accuracy scores. • Different weighting is applied for the elements depending on the season, to reflect its importance in public’s perception

Monthly mean verification score and 12 -month running average of local weather forecasts for

Monthly mean verification score and 12 -month running average of local weather forecasts for next day (score range: 0~100)

Monthly mean verification score of 7 -day forecasts (score range: 0~100)

Monthly mean verification score of 7 -day forecasts (score range: 0~100)

Tropical Cyclone Warnings • Area of responsibility of HKO is bounded by the white

Tropical Cyclone Warnings • Area of responsibility of HKO is bounded by the white line, 10 -30 N and 105 -125 E. • Warnings issued every 3 hours for 72 hours

HKO tropical cyclone forecast track error

HKO tropical cyclone forecast track error

Forecast track of Chanchu – based on 12 UTC 12 May 2006 • HKO’s.

Forecast track of Chanchu – based on 12 UTC 12 May 2006 • HKO’s. TC information processing system (TIPS) – forecast tracks of various models and the multi-model ensemble • Black line : postanalysed track

A prolonged cold spell in early 2008 over south China

A prolonged cold spell in early 2008 over south China

For Hong Kong • Longest cold spell in 40 yrs (24/1 – 16/2) •

For Hong Kong • Longest cold spell in 40 yrs (24/1 – 16/2) • Daily min temperature below 12 C • Mean min temp was 9. 9 C, second lowest during the same period since record began in 1885. • Continuous push of cold air from Siberia into southern China and moist air transported from the South China Sea and even as far as the Indian Ocean, brought continuously cloudy, rainy and cold weather to the region.

72 -hrs forecast pressure change in 24 hrs (PC 24) Initial : 12 UTC

72 -hrs forecast pressure change in 24 hrs (PC 24) Initial : 12 UTC daily 24 Jan - 17 Feb 2008 forecast and actual PC 24 ECMWF JMA

Mean temperature error field (in deg) of 72 -hr f/c Initial : 12 UTC

Mean temperature error field (in deg) of 72 -hr f/c Initial : 12 UTC 24 Jan - 16 Feb 2008 daily JMA sfc (left) & ECMWF 850 h. Pa (right)

EPSgram 10 -day forecast with base time at 12 UTC on 22 Jan

EPSgram 10 -day forecast with base time at 12 UTC on 22 Jan

EPSgram 10 -day forecast with base time at 12 UTC on 29 Jan 20

EPSgram 10 -day forecast with base time at 12 UTC on 29 Jan 20 Observed temperature (°C) 15 10 5

EPSgram 10 -day forecast with base time at 12 UTC on 7 Feb

EPSgram 10 -day forecast with base time at 12 UTC on 7 Feb

Fengshan Best track 18 -25 June 2008

Fengshan Best track 18 -25 June 2008

Forecast tracks of Fengshan predicted by global models 00 Z 19 June 2008 00

Forecast tracks of Fengshan predicted by global models 00 Z 19 June 2008 00 Z 23 June 2008 00 Z 22 June 2008

ECMWF 500 h. Pa GPH 12 UTC on 24 June 2008 T+72 forecast analysis

ECMWF 500 h. Pa GPH 12 UTC on 24 June 2008 T+72 forecast analysis

Combination of Typhoon Ketsana and NE monsoon over the south China coast

Combination of Typhoon Ketsana and NE monsoon over the south China coast

Ketsana & monsoon - surface analysis 27, 29 Sep & 1 Oct 2009

Ketsana & monsoon - surface analysis 27, 29 Sep & 1 Oct 2009

Ketsana and monsoon – satellite imageries 27, 29 Sep & 1 Oct 2009

Ketsana and monsoon – satellite imageries 27, 29 Sep & 1 Oct 2009

T+12 H forecast of 12 -hourly rainfall by ECMWF model for the grid closest

T+12 H forecast of 12 -hourly rainfall by ECMWF model for the grid closest to HK - actual rainfall recorded at the HKO in purple

surface prognoses valid for 12 UTC on 30 Sep T+96 T+48 T+72 T+24

surface prognoses valid for 12 UTC on 30 Sep T+96 T+48 T+72 T+24

Radar imageries at 07 -12 UTC on 30 Sep 2009 Amber rainstorm warning issued

Radar imageries at 07 -12 UTC on 30 Sep 2009 Amber rainstorm warning issued

Concluding remarks • NWP advancement >more accurate forecast with longer lead time • Helped

Concluding remarks • NWP advancement >more accurate forecast with longer lead time • Helped raise public expectation • Occasional situations with satisfactory forecasts by models are more difficult to handle • It may be beneficial to foster closer cooperation between operational forecasters and model developers

Thank you

Thank you