Introduction to Radar Based Nowcasting WOO Wang-chun Forecast - - PowerPoint PPT Presentation
Introduction to Radar Based Nowcasting WOO Wang-chun Forecast Development Division, Hong Kong Observatory E-mail: wcwoo@hko.gov.hk 26 July 2016 WMO WWRP 4th International Symposium on Nowcasting and Very-short-range Forecast 2016 (WSN16) What
WOO Wang-chun
Forecast Development Division, Hong Kong Observatory
E-mail: wcwoo@hko.gov.hk 26 July 2016 WMO WWRP 4th International Symposium on Nowcasting and Very-short-range Forecast 2016 (WSN16)
Quantitative Precipitation Estimate (QPE)
Quantitative Precipitation Forecast (QPF)
Lightning, Gust, Hail
Forecasts & Warnings
Regional Rainfall Map
Local Rainfall Map for the Public Local Rainfall Map for Forecasters
For Forecasters For Public For Public
For Internal Customer (Forecasters)
For Public
Rad Radar ar-base ased
– Correlation-based – Optical flow – Convolutional LSTM
afterwards
NWP-base ased
due to spin-up problem
afterwards
Accuracy Rain Gauge Radar Satellite Best Moderate Worst Spatial Resolution Discrete Continuous, up to 200 m Continuous, up to 500 m Type In-situ Remote Sensing Remote Sensing Spatial Coverage At point only Regional, effective up to 256 km (radius) Half the Globe (geostationery) Cost Cheap as single unit Expensive as network Expensive to operate Expensive to launch Cheap to use
Shor
ge War arni ning o
ntense se Rai ainst nstorm i in n Local alized Systems
:- Quant ntitative precip ipita tation e estim timatio ion ( (QPE)
– rada dar-bas ased, d, rai ainga gauge ge-base sed and b blending w with th s satellite cloud images
:- Retrie ieval l of echo m motio tion
– tracki cking by by max aximum cor
TREC) – tracki cking by by opt
cal flow –
ct-or
storm motion
OREC ECAST : :- semi-Lag agran angian an ad advecti tion t to extrapolate r rad adar ar r reflecti ctivity u ty up to
6 / 9 9 hour
UTPUTS : S :- computa tatio ion of gridde dded p d precipi pitation no nowcas cast (QPF) and and l loca cati tions of stor
cts o
n convectiv ctive w wind g gust, t, lightn tnin ing and and hail, suppor
to decision m
ng
UNCERTAINTY : :- probabilistic Q c QPF a and b blend nding ng wi with c h conv nvection
permitting N g NWP WP model
PRODUCTS : S :- now
prod
for i internal us users a and p pub ublic
calibration of radar reflectivity using real-time raingauge measurement.
reflectivity to rainfall rate
computed by Barnes successive correction or more advanced co- kriging algorithm
b
dBZi =b dBGi + 10log(a)
– interpolation with Gaussian weighting according to distance between data & estimation point – consider correction using residuals and grouping of rainguages B G G G G G G h
B : barnes estimation (mm) L : radius of influence N0 : number of gauge report Gi : i-th gauge report (mm) wi : weight of i-th gauge hi : distance between gauge and estimation point
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
G Gi
i
h h
K(x K(x0
0)
)
G Gi
i
G Gi
i
G Gi
i
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
R R
j j
G Gi
i
h h
K(x K(x0
0)
)
G Gi
i
G Gi
i
G Gi
i
=1 =1
co-Kriging estimate: ( ) ( ) ( )
N M i i j j i j
K x x G x R λ λ = +
∑ ∑
[ ]
{ }
2 2
seek to minimize: ( ) ( ) E K x G x σ = −
=1 =1
subject to constraints: ( ) 1 & ( ) 0
N M i j i j
x x λ λ = =
∑ ∑
=1 =1 =1 =1
Solution: ( ) ( , ) ( ) ( , ) ( ) ( , ), for 1, , ( ) ( , ) ( ) ( , ) ( ) ( , ), for 1, ,
N M i GG n i j GR n j G GG n i j N M i RG m i j RR m j R RG m i j
x x x x x x x x x n N x x x x x x x x x m M λ γ λ γ µ γ λ γ λ γ µ γ + + = = + + = =
∑ ∑ ∑ ∑
Actual Rainfall
512 km 512 km 1158 km 904 km 1804 km 1728 km TMS Radar composite through collaboration with Guangdong
Satellite Channels
Base Time : 2013-04-05 02: 12 HKT ( 6 hour forecast ) Actual Extrapolate with only Hong Kong Radars Extrapolate with Multi- Sensors
Sample Time : 2014-07-22 15:24
SWIRLS Ensemble Rainstorm Nowcast (SERN) SWIRLS Rainstorm Viewer
Thunder- storm with high gusts expected during this period Amber expected in 36 minutes with criteria M and S met
SWIRLS Severe Weather Viewer Actual Rainfall (QPE) Forecast Rainfall (QPF)
mobile app
2 hours at your location
– data from SWIRLS QPF
alerting service based on user location and expected rainfall
– update frequency – notification intervals – range of detection – forecast location
http://www.weather.gov.hk/nowcast/prd/api/
the Pearl River Delta region in the next 2 hours
http://maps.weather.gov.hk/ocf/
網址: : http://maps.weather.gov.hk/ocf/index_uc.html Rainfall nowcast
Click to display time series
Provide image and animation sequence of rainfall forecast map over HK and Pearl River Delta for the next 2 hours
39 dBZ 39 dBZ
Group echo identification
∑
∆ × = = = − = =
− ellipse
A dbZ I a b aE P a b a ab A ) / ( tan ) , 2 / ( 4 /
1 2 2
θ ε π ε π max dBZ <--> max rainfall ave dBZ <--> ave rainfall ave (max50% rainfall) ave (max25% rainfall)
x y
a (major axis) b (minor axis) θ (orientation)
V
(TREC speed and direction)
area eccentricity perimeter
total intensity
Based on moving speed, size, overlapping area
merging splitting translation
Searching radius
Two systems merging and moving steadily towards SE
T+0min T+6min T+12min T+18min
20°C:
0 to -20°C
+ +
+ + +
analysis
R A B D C A1 A2 D2 D1 B2 B1 C2 C1 R0
℃
D0
℃
A0
℃
B0
℃
C0
℃
– 60-dBZ TOPS > 3 km – 0-2km VIL < 5 mm
– Time-lagged ensemble of blending QPF
– time lagged ensemble of extrapolated sub-zero reflectivity fields based on optical flow
Aggregate latest 10 RAPIDS QPF according to exponential decreasing weights
Spread of radar rainfall nowcast via selecting various parameters in echo motion retrieval
rainfall nowcast ensemble of 36 members.
Select T+60 … 540 min nowcast
Radar-based / Multi- Sensor Deterministic Probabilistic Products
ESCAP/WMO Typhoon Committee Research Fellowship Scheme
Typhoon Committee Research Fellowship 2012
Separate the motion of TC before radar echo tracking Quantitative precipitation forecasts For TC
ACTUAL Forecast using TC Module
Severe Typhoon Vicente 13HKT on 23 July 2012 Verification (15 Cases in 2003-2012) Threshold = 1mm Threshold = 20mm
Rainstorm Analysis and Prediction Integrated Data-processing System
– QPF by semi-Lagrangian advection of radar echoes
– QPF by non-hydrostatic model
RAPIDS-NHM rainfall forecast
Meso-NHM
BC from ECMWF IFS forecasts
RAPIDS-NHM
Hong Kong
Hong Kong
Doppler velocity from radars in HK multi-layer wind retrieval (u,v) using radar mosaic CAPPI reflectivity volume for 1D retrieval (mosaic from HK + Guangdong radars)
RAPI DS-NHM
Shenzhen and Guangzhou
– JO is proportional to the square of difference between the observed radial velocity and the radial velocity derived from retrieved 3D wind field; – JB is proportional to the square of difference between the retrieved 3D wind field and the background; – JD is the anelastic mass constraint term; and – JS is the smoothness constraint of retrieved wind field using Laplacian of wind components.
Horizontal res.: 1 km Vertical res.: 500 m Reference: Data Assimilation of Weather Radar and LIDAR for Convection Forecasting and Windshear Alerting in Aviation Applications Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications (Vol. II), 2013, pp 527-554 DOI 10.1007/978-3-642-35088-7_22
Control
Pick up short-wave disturbance in coastal waters at 0400 UTC and develop another NE-SW band of simulated reflectivity
With radar wind retrieval
50
Control Expt (no radar)
Actual rainfall analysis
RAPIDS-NHM 3-hr acc. rainfall ending at 1630H With radar wind retrieval
SWIRLS nowcasts for 00:30 – 01:00H Blending SWIRLS + NWP
– Quantitative precipitation estimates – Quantitative precipitation nowcast (0-9 hr) – Severe weather parameters (lightning, hail, downburst)
– Improve very-short-range forecast
HKO’s Mascot