Korea Meteorological AdministrationKorea Meteorological Administration

Yong-Sang Kim, Chun-Ho Cho, Oak-Ran Park, Hyeon LeeYong-Sang Kim, Chun-Ho Cho, Oak-Ran Park, Hyeon Lee

One of the greatest deficiencies of numerical weather prediction models is their lack of skill in predicting clouds and precipitation in the early portions of their forecast period because of the “spin-up” problem.

Even today, most mesoscale model initialize forecasts with zero cloudiness, and accept the “spin-up” time between initial time and the first reasonable cloud distributions and precipitation rates.

In this study, KLAPS(Korea Local Analysis and Prediction System) is used to improve the initial values of liquid water species (cloud water, cloud ice, rain water, snow, and graupel) in the MM5.

We introduce the diabatic initialization algorithm including cloud analysis and the impact study to using as the initial data of numerical model is also explained.

1. Introduction1. Introduction1. Introduction1. Introduction

A Three Dimensional Cloud Analysis for Diabatic Initialization of Mesoscale Model and its Impact Study

6. Result6. Result6. Result6. Result Fig. 7 shows that the performance of precipitation forecast is improved by assimilating these va

riables (HOT experiment), compared to COLD experiment (i.e. an experiment without assimilating the retrieved variables).

Considering the statistical verification of POD (Probability of Detection) from several rainfall casr studies, the HOT experiment showed significant improvement during the first 6 hour forecasts compared to the COLD experiment (Fig. 8).

World Weather Research Program Symposiumon Nowcasting and Very Short Range Forecasting5-9 September 2005Meteo-France Toulouse France

Acknowledgements: Acknowledgements: This study was performed by using the Forecaster’s Analysis System developed by the KMA principal project “International Cooperative Development of Advanced Forecasting System” and the “Development of physical initialization technique for the very short range weather forecast” one of the International Joint R&D Projects.  

Corresponding Author:Corresponding Author: Yong-Sang Kim, Information Management Division, Korea Meteorological Administration, 460-18 Shindaebang-dong, Dongjak-gu, Seoul, 156-720, Kore

a Tel : +82-2-834-6248 /Fax : +82-2-834-6249 /Email :yskim@kma.go.kr

Fig. 4. The final 3-D cloud analysis of the 5 phases of cloud and precipitation (a) and cloud type and cloud omega (b).

KLAPS was designed to provide a computationally efficient method of combining all available sources of meteorological information into three dimensional depiction of the atmospheric state, with an emphasis on Nowcasting.

The analysis routines consist of surface analysis, 3-dimensional analyses of wind, temperature, moisture and cloud.

Fig. 1 shows the schematic diagram of KLAPS process : 1)KLAPS analysis is performed for every 6 hour using synoptic and asynoptic observations, include Geostationary Meteorological Satellite data, METAR from local airports, commercial airline weather reports from Aircraft Communications Addressing and Reporting System(ACARS). 2) KLAPS analysis data is used not only for nowcasting but also for initial data of local scale numerical model.

KLAPS has been increasingly used as a means to initialize numerical weather prediction models because of its data ingest and quality control capabilities.

2. Introduction of KLAPS2. Introduction of KLAPS2. Introduction of KLAPS2. Introduction of KLAPS

4. The data Assimilation using QG equation including 4. The data Assimilation using QG equation including cloud amountcloud amount4. The data Assimilation using QG equation including 4. The data Assimilation using QG equation including cloud amountcloud amount The balance procedure uses 3-dimensional

variational formulation to combine the background fields with the initial KLAPS univariate analyses of the mass and momentum fields (temperature, wind) while considering the cloud vertical motions(eq.1) and the concept is illustrated at Fig. 5.

3. The 3-Dimensional Cloud Analysis Algorithm3. The 3-Dimensional Cloud Analysis Algorithm3. The 3-Dimensional Cloud Analysis Algorithm3. The 3-Dimensional Cloud Analysis Algorithm

The cloud analysis utilize a variety of observation data, including GOES satellite imagery, aircraft observation, surface observation (METAR), and radar reflectivity data (Fig. 2a).

Fig. 2b shows the flow diagram of KLAPS 3-dimensional cloud analysis (Albers, et al., 1996). The 3-dimensional clouds are analyzed by adding the observation step by step and Fig. 3 shows the example of 3-dimensional cloud distribution at each step (Fig. 3)

Ztop

depth

Cu

Cu

CuZbase

Wmax

KLAPS Cloud Analysis

Cloud liquid water and Cloud ice

Cloud type decision

Eq.3

BL uu BL vv

0

Du

Cloud ~

BD ~

BD uuu ~

BD vvv ~

Eq.(2) compute

、、、、 ~~~~vu

Take into Eq.(1), (3),(4),(5)

、、、 vu

Mixing ratio of cloud liquid water and cloud ice; Reset RH=100% in cloud region

D

BL uu BL vv

BL

0BL uu BL vv

0

Dv

Eq.1

Eq.4

Eq.1

Eq.2

Eq.3

Eq.4

Eq.5

A strong constraint of mass continuity is applied as part the 3DVAR cost function, such that the horizontal wind field, is slightly adjusted to produce divergence fields consistent with cloud vertical motion fields.

In addition, to prevent the introduction of excessive gravity waves during the initial few time steps of model integration, the cost function also minimizes the time tendencies of the horizontal wind fields.

Fig. 6 is the process of solving Eq.1, Eq.2, and Eq.3 by successive correction method.

cloud sounding

Analyze horizontally

Compare T0 and Ts. Add or delete layer until Ts

and T0 are consistent

Add cloud if radar echo above existing cloud base and high

er than dBZ threshold

Decrease or add visible & 3.9 um

satellite cloud cover

Background cloud

cover

METAR

GOES IR

RADAR Ref.

GOES VIS & 3.9 um

Final 3D Cloud

Fig. 2. (a) The schematic diagram illustrating the contributions of the various data sources for 3-dimensional cloud analysis and (b) the flow diagram of KLAPS 3-dimensional cloud analysis algorithm (Albers et al., 1996)

Fig. 1. The Schematic diagram of KLAPS Process

Synoptic weather chart

High Resolution Analysis

Radar

NWPD

ATOVSRSRL

Satellite

GTS SYNOP , SHIP, TEMP , BUOY , METAR

KMETAR

AWS

Wind Profiler

BUOY

ACARS

JMA

FSL

AMO

RSD

IMD

GDAPSGDAPS

KLAPS KLAPS (sfc,3d ananlysis & cloud)(sfc,3d ananlysis & cloud)

Initial data for NWP

MM5 (18,6,2 MM5 (18,6,2 km)km)

Homepage orHomepage orFAS systemFAS system

MM5 (18 km)

For Nowcasting

        

        

          

                 (1)

2222

22

22'22

ˆˆˆˆ

)ˆˆˆ()ˆ()ˆ(

)ˆ()ˆ()ˆ()ˆ(

BBvBuB

vuvu

OOvvOuuOJ

VV

pyxtt

cVk j i

V

+1st +met

+ir

+rad +visRadar echo

Fig. 3. The example of 3-dimensional cloud distribution according to adding the observational data

After producing the 3-dimensional cloud distribution, it diagnoses the mixing ratios of each hydrometeor species, performs cloud typing based on stability and temperature information, and assigns appropriate vertical motion profiles to each cloud analysis field (Fig. 4, Table 1).

Table 1. Cloud type and drop size specifications (Albers et al., 1996)

q= qs

c

Fig. 5. The schematic diagram of balance algorithm between wind, mass and cloud derived omega

Fig. 6. The solving process of 3DVAR formalism

Forecast hour

2 4 6 8 10 12

PO

D (

abov

e 1m

m/1

hr)

0.0

0.2

0.4

0.6

0.8

1.0

COLDHOT

Fig. 7. The 1-hourly accumulated precipitation of each 1,2,3 hour forecast (1st column, a,d,g, COLD experiment; 3rd column, c,f,I, HOT experiment) and radar retrieved precipitation images (2nd column, b,e,h) related with each output.

Fig. 8. Probability of detection of precipitation (threshold at 1mm) for 8 rainfall cases.

a b

a b

Eq.1

)ˆ(ˆ)ˆˆˆˆˆˆ(ˆ uDvfuuuvuvuuuuu xbppbbyybbxxbt Eq.2

Eq.3

AvNfuu xy /)

2)~(~(

AuNfvv yx /)

2)~(~(

2~ p

)~~~(22pyxpp AvuA

))~()~((2 yx uNvNf

vNAfu

Af x ~2~

2~

xy uNvNA )~()~(~

AAf

y 22