Seasonal Forecast for Summer 2019 for Hong Kong

  • Slides: 51
Download presentation
Seasonal Forecast for Summer 2019 for Hong Kong Wing-hang, CHAN Hong Kong Observatory

Seasonal Forecast for Summer 2019 for Hong Kong Wing-hang, CHAN Hong Kong Observatory

Outline • Review of seasonal forecast for JJA in 2018 • Seasonal forecast for

Outline • Review of seasonal forecast for JJA in 2018 • Seasonal forecast for JJA in 2019 • Annual outlook 2019 and recent development

Review of seasonal forecast for JJA in 2018 11

Review of seasonal forecast for JJA in 2018 11

Review of JJA in 2018 Forecast Actual Temperature Normal to above normal Above Normal

Review of JJA in 2018 Forecast Actual Temperature Normal to above normal Above Normal (28. 8 °C) Rainfall Normal to below normal Near normal (1415 mm) Rainfall anomalies(%) in JJA 2018 Hong. Kong JJA Climate Temperature: 28. 3 – 28. 6 ° C Rainfall: 1201 -1509 mm

Seasonal forecast for JJA in 2019 11

Seasonal forecast for JJA in 2019 11

Seasonal Climate Forecast ENSO consideration (El Niño) : JJA Temp. Rainfall FC Cat N

Seasonal Climate Forecast ENSO consideration (El Niño) : JJA Temp. Rainfall FC Cat N to AN N to BN Normal to warm Normal to dry Prognostic charts: JJA Temp. Rainfall FC Cat N to AN Normal to warm Normal to wet Forecasts from indices JJA Temp. Rainfall FC Cat N to AN Near N Normal to warm Near normal Consensus of major centres: JJA Temp. Rainfall FC Cat N to AN Normal to warm Normal to wet

1. ENSO consideration 11

1. ENSO consideration 11

ENSO Status 8

ENSO Status 8

ENSO F/C Nino 3. 4 SST anomaly (deg C) AMJ MJJ JJA JAS ASO

ENSO F/C Nino 3. 4 SST anomaly (deg C) AMJ MJJ JJA JAS ASO SON OND NDJ DJF Average, all models 0. 9 0. 8 0. 7 9

ENSO impact on HK climate Large scale: Source: NOAA Climate. gov Local statistics: Impact

ENSO impact on HK climate Large scale: Source: NOAA Climate. gov Local statistics: Impact on seasonal rainfall/temperature distribution (1950 -2018) Temperature (El Niño) Below normal Normal Above normal JJA 2 4 10 Rainfall (El Niño) Below normal Normal Above normal JJA 6 7 3 Source: HKO 10

1950 -2018 Observations JJA ENSO-neutral (32 cases) El Nino (16 cases) La Nina (11

1950 -2018 Observations JJA ENSO-neutral (32 cases) El Nino (16 cases) La Nina (11 cases) Temp. (o. C) 28. 4 28. 1 Rf (mm) 1265 1161 1164 The difference of RF in El Nino years and La Nina years are not statistically significant. The difference of Temp. in El Nino years and La Nina years are not statistically significant. 11

2. Prognostic charts: 11

2. Prognostic charts: 11

JJA El Niño Years Composite (1981 -2018) Z 500_std_anom 500 wind_anom C A 1982,

JJA El Niño Years Composite (1981 -2018) Z 500_std_anom 500 wind_anom C A 1982, 1983, 1987, 1991, 1993, 1997, 2002, 2009, 2014, 2015 A 13

JJA El Niño Years Composite (1981 -2018) 850 h. Pa wind anom. Rainfall anom.

JJA El Niño Years Composite (1981 -2018) 850 h. Pa wind anom. Rainfall anom. 1982, 1983, 1987, 1991, 1993, 1997, 2002, 2009, 2014, 2015 14

ECMWF 2019 JJA FC Charts 500 h. Pa GPH std. anom. 500 h. Pa

ECMWF 2019 JJA FC Charts 500 h. Pa GPH std. anom. 500 h. Pa wind anom. C No significant anti-cyclonic flow near HK 15

ECMWF 2019 JJA FC Charts 925 h. Pa wind + SST std. anom. Rainfall

ECMWF 2019 JJA FC Charts 925 h. Pa wind + SST std. anom. Rainfall std. anom. moisture supply 16

3. Consensus of major centres 11

3. Consensus of major centres 11

JJA Temperature 11

JJA Temperature 11

ECMWF N – Warm NOAA N – Warm JMA N – Warm 11

ECMWF N – Warm NOAA N – Warm JMA N – Warm 11

UKMO N – Warm CMA/BCC N – Warm APCC N – Warm MJJ 20

UKMO N – Warm CMA/BCC N – Warm APCC N – Warm MJJ 20

C 3 S multi-system N – warm WMO LC N – warm MJJ 21

C 3 S multi-system N – warm WMO LC N – warm MJJ 21

JJA Rainfall 22

JJA Rainfall 22

ECMWF N – wet NOAA Uncertain JMA N – wet (very close to no

ECMWF N – wet NOAA Uncertain JMA N – wet (very close to no signal) 23

UKMO N – wet CMA/BCC N – wet APCC N – wet MJJ 24

UKMO N – wet CMA/BCC N – wet APCC N – wet MJJ 24

C 3 S multi-system N - wet WMO LC N - wet 25 MJJ

C 3 S multi-system N - wet WMO LC N - wet 25 MJJ 17

Tercile Probabilities Summary Forecast category of JJA from different centres 2019 JJA   TEMPERATURE

Tercile Probabilities Summary Forecast category of JJA from different centres 2019 JJA   TEMPERATURE   RAINFALL CMA/BCC ECMWF JMA UKMO NOAA/CPC APCC C 3 S WMO LC   N – Warm   N – Warm(MJJ) N - wet Uncertain N - wet (MJJ) * N – Warm/Cool/Wet/Dry : Normal to Warm/Cool/Wet/Dry 26

4. Consensus categorical forecast by pre-season indices (PSI) and contemporary indices (CTI)

4. Consensus categorical forecast by pre-season indices (PSI) and contemporary indices (CTI)

Predictors for JJA RF in last year PSI: • DJF UMI and Ji index

Predictors for JJA RF in last year PSI: • DJF UMI and Ji index • DJF SST index (SSTa-SSTb) CTI: • JJA EC UMI • JJA EC DMI SSTa SSTb Ji: 1000 h. Pa V over 10 -30 N, 115 -130 E UMI: 1000 h. Pa V over 7. 5 -20 N, 107. 5 -120 E DMI: anomalous SST gradient between (50 E-70 E and 10 S-10 N) and (90 E-110 E and 10 S-0 N). @Indian Ocean 28

30 -year running correlation with JJA rainfall PSI : related to winter monsoon CTI

30 -year running correlation with JJA rainfall PSI : related to winter monsoon CTI : related to summer monsoon and [email protected] ocean PSI CTI The correlations are decreasing in recent years 29

 • Aim New Approach of Predictor Search • Out-sampling. Training/searching period is out

• Aim New Approach of Predictor Search • Out-sampling. Training/searching period is out of the verification period. • A more generic and automatic way to find predictors. • Elements considered • U/V/T/Z anomaly at standard levels(from 1000 h. Pa to 50 h. Pa) 1 • T 2 m/p. Rate/SST/sea-ice 2 /u 10 m/v 10 m/mslp anomaly • Cluster • • Construct correlation map (moving window of 40 years) Find grid points with confidence level >=95%. Group neighbouring grid points into cluster Discard clusters smaller than 500*500 sq. km • Principal Components • Consider first half set of the EOFs (lower order) • Adopt PC with confidence level >=95% Note: 1. 150 h. Pa, 100 h. Pa and 50 h. Pa are excluded in current EC 5 data. 2. sea-ice is excluded in current EC 5 data. 30

Clustering grid points of high correlation (ECMWF VS HK JJA rainfall ) • Clusters

Clustering grid points of high correlation (ECMWF VS HK JJA rainfall ) • Clusters map (combined all pressure levels) • ECMWF JJA model FC data VS HK JJA rainfall 500 h. Pa V wind -ve correlation (S’lies -> less rainfall) STR extend to west -> Less rainfall 31

Consensus Forecast of PSI and CTI Normal to warm Near normal rainfall 32

Consensus Forecast of PSI and CTI Normal to warm Near normal rainfall 32

Performance of PSI and CTI (% of correct forecast) 1997 -2018 Random JJA Temp

Performance of PSI and CTI (% of correct forecast) 1997 -2018 Random JJA Temp JJA RF ~70% 2009 -2018 Random JJA Temp JJA RF ~70% Persist Past 21 PSI+CTI NA NB Yr Trend 91% 82% 41% 64% 91% 64% 86% 64% Persist Past 21 PSI+CTI NA NB Yr Trend 100% 70% 20% 90% 100% 70% Forecast for 2019 JJA Forecast Temperature Normal to above normal Rainfall Near normal 33

Seasonal Climate Forecast ENSO consideration (El Niño) : JJA Temp. Rainfall FC Cat N

Seasonal Climate Forecast ENSO consideration (El Niño) : JJA Temp. Rainfall FC Cat N to AN N to BN Normal to warm Normal to dry Prognostic charts: JJA Temp. Rainfall FC Cat N to AN Normal to warm Normal to wet Forecasts from indices JJA Temp. Rainfall FC Cat N to AN Near N Normal to warm Near normal Consensus of major centres: JJA Temp. Rainfall FC Cat N to AN Normal to warm Normal to wet June - August 2019 (preliminary FC) Temperature: Normal to above normal Rainfall: Normal to above normal

Annual outlook 2019 and recent development

Annual outlook 2019 and recent development

Annual outlook issued in March 2019 Annual rainfall in Hong Kong Normal to above

Annual outlook issued in March 2019 Annual rainfall in Hong Kong Normal to above normal (between 2300 and 2900 mm) Number of tropical cyclones entering 500 km of Hong Kong Near normal (4 to 7) Onset of tropical cyclone season June or after Chance of annual mean temperature reaching the top 10 positions High Jan Feb Mar Apr 18. 1°C 20. 1°C 21. 0°C 24. 7°C 1. 8°C 3. 3°C 1. 9°C 2. 1°C 36

The impact of long-range temperature forecasts on electricity load forecasting 长期温度预报对预测用电量的影响 陈永铿 1 李国梁2

The impact of long-range temperature forecasts on electricity load forecasting 长期温度预报对预测用电量的影响 陈永铿 1 李国梁2 庄家乐 2 唐恒伟 1 李细明1 1香港天文台 Hong Kong Observatory 2香港电灯有限公司 Hongkong Electric Company, Limited 37

Thank you 38

Thank you 38

Appendix 39

Appendix 39

40

40

Verification of long-range temperature forecasts 长期温度预报的验证 Correlation coefficients between electricity load person and monthly

Verification of long-range temperature forecasts 长期温度预报的验证 Correlation coefficients between electricity load person and monthly mean temperature 全港人均用电量与月平均温度相关系数 Long-range temperature forecasts is relatively skillful in warm month(Jun-Sep). While skill over Mar, Apr and Dec is not good enough. 长期温度预报在温暖月份(6至 9月)有一定技 巧,个别的月份例如3 、4和12月的表现则 不太理想 It shows positive correlation over warm months. 同时,用电量在温暖月份与月平 均温度较有正相关 41

GRCM charts MJJ-RF MJJ-TT GRCM 20190316 - 0321 ENS mean (6 members) HC base

GRCM charts MJJ-RF MJJ-TT GRCM 20190316 - 0321 ENS mean (6 members) HC base time: 1982 -2009, 0318 & 0323 (2 members) 42

Reference forecast for skill evaluation • Climate JFK • Jan, Feb temperature known •

Reference forecast for skill evaluation • Climate JFK • Jan, Feb temperature known • Mar-Dec climate Normal distribution of Mar-Dec temp Target depends on Jan & Feb obs and records at the time • Performance metric – BSS • BSS > 0 : better than ref. f/c (max BSS at 1) • BSS < 0 : worse than ref. f/c 43

Deterministic event forecast Forecast occurrence of event if the probability exceeds a certain threshold

Deterministic event forecast Forecast occurrence of event if the probability exceeds a certain threshold No. of correct forecasts in 1997 -2018 (prob threshold in brackets) Event EC (50%) NC (50%) Avg of EC, NC (50%) Climate JFK (20%) Always “Yes” Always “No” Top 1 - - 2 20 Top 5 17 15 15 15 10 12 Top 10 16 16 16 6 44

Clustering grid points of high correlation (reanalysis data VS HK JJA rainfall ) •

Clustering grid points of high correlation (reanalysis data VS HK JJA rainfall ) • Clusters map (combined all pressure levels) • JRA 55 DJF+MA reanalysis data VS HK JJA rainfall 45

Clustering grid points of high correlation (reanalysis data VS HK JJA rainfall ) •

Clustering grid points of high correlation (reanalysis data VS HK JJA rainfall ) • Clusters map (combined all pressure levels) • JRA 55 JJA reanalysis data VS HK JJA rainfall 46

JJA Dry Years Composite (1981 -2018) Z 500_std_anom 500 wind_anom C A A 1981,

JJA Dry Years Composite (1981 -2018) Z 500_std_anom 500 wind_anom C A A 1981, 1983, 1984, 1989, 1990, 1991, 1992, 1993, 1996, 2002, 2004, 2011, 2012, 2015 C 47

JJA Dry Years Composite (1981 -2018) 850 h. Pa wind anom. NE’lies anomaly 1981,

JJA Dry Years Composite (1981 -2018) 850 h. Pa wind anom. NE’lies anomaly 1981, 1983, 1984, 1989, 1990, 1991, 1992, 1993, 1996, 2002, 2004, 2011, 2012, 2015 48

JJA Wet Years Composite (1981 -2018) Z 500_std_anom 500 wind_anom A C 1994, 1995,

JJA Wet Years Composite (1981 -2018) Z 500_std_anom 500 wind_anom A C 1994, 1995, 1997, 2001, 2005, 2008, 2017 A 49

JJA Wet Years Composite (1981 -2018) 850 h. Pa wind anom. SE’lies anomaly 1994,

JJA Wet Years Composite (1981 -2018) 850 h. Pa wind anom. SE’lies anomaly 1994, 1995, 1997, 2001, 2005, 2008, 2017 50

JJA La Nina Years Composite (1981 -2010) Z 500_std_anom 500 wind_anom A C 1985,

JJA La Nina Years Composite (1981 -2010) Z 500_std_anom 500 wind_anom A C 1985, 1988, 1999 51