APEC Climate Center (APCC) Climate Prediction and Information Activities: New Developments
Total Page:16
File Type:pdf, Size:1020Kb
___________________________________________________________________________ 2008/SOM3/ISTWG/SYM/031 Agenda Item: 4-05 APEC Climate Center (APCC) Climate Prediction and Information Activities: New Developments Submitted by: APEC Climate Center (APCC) APEC Climate Symposium Lima, Peru 19-21 August 2008 20 August, 2008 APEC Climate Symposium, 2008 Abstract 4-05 APCC Climate Prediction and information activities: New developments Dr.Karumuri Ashok, APCC, Korea ABSTRACT The talk covers the new operational efforts and developmental projects in APCC since the beginning of 2008. The APCC has been issuing the rolling monthly 3-month forecast since 2008, with due verification. It has also been supporting KMA by providing downscaled forecasts for 60 Korean stations. The APCC has also initiated experimental 1-tier 6-month MME prediction, slated on seasonal basis, and so far developed two forecasts. The SNU, SINTEX-F (FRCGC), UH data (all three datasets provided by CliPAS), NCEP, and POAMA datasets are used in this efforts. To boost this effort, APCC has also started to develop an in-house coupled prediction model based on the CCSM3. To do this, APCC has developed a coupled SST nudging scheme. The presentation will briefly touch about other ongoing development projects such as a probabilistic downscaling system, a drought monitoring system etc. The presentation will also introduce issues such as the need for longer hindcast data, need for more 1-tier forecast datasets etc. The climate conditions of since 2007 spring will also be briefly presented, along with the performance of the APCC MME system for some of the seasons. The presentation will also seek guidelines on future activities of APCC. The APCC climate prediction activities: New Developments Australia Asia Brunei Darussalam Canada Chile People’s Republic of China Pacific Hong Kong, China Indonesia Economic Cooperation Japan Korea Malaysia Mexico Climate New Zealand APEC Climate Center (APCC) Papua New Guinea Peru for Climate Information Services to Society, Philippines enter Russia C Busan, Republic of Korea Singapore Chinese Taipei Thailand United States Viet Nam 1 APEC Climate Center 2008 projects Operations • Operationalization of monthly 3-month forecast since January, with upgraded hindcast verification. • Downscaled forecast provision at 60 Korean stations to KMA with due hindcast verification • New monitoring products Development • 6-month MME seasonal prediction • In-house development of coupled prediction system • Probabilistic downscaling 2 APEC Climate Center 1 Operational Products Enhancement of MME forecast for APCC economies Æ 12 times/yr Operation schedule 10 ~ 15 16 ~ 18 19 ~ 23 production of MME MME forecast outlook & climate Data acquisition & analysis forecast highlights services Multi-institutional cooperation and sharing of high-cost data 3 APEC Climate Center Models Start End SST Specification Ensemble Name/Economy Variables Model Designation Focal Point/E-Mail Year/Month Year/Month (Hindcast/Forecast) (H/F) BCC NCC-CGCM Peiqun Zhang 1983/Jan. 20008/Present PREC,T850,Z500 Predicted SST/ 8/8 China T63L16 [email protected] COLA T2M, PREC, MSLP, T850, U850, COLA AGCM v2.2.6 OISSTv2/ Daniel Paolino 1981/Dec. 2002/Dec. 10/10 U.S.A. V850, U200, V200, Z500, OLR T63L18 IRI SST Forecast [email protected] CWB T2M, PREC, MSLP, T850, U850, OISSTv2/ Jyh-Wen Hwu 1979/Jan. 2003/Dec. CWB T42L18 5/10 Chinese Taipei V850, U200, V200, Z500 IRI SST Forecast [email protected] GCPS T2M, PREC, MSLP, T850, U850, KMA/SNU SST Forecast/ Jeong-Seok Lee 1979/Jan. 2008/Present GCPS T63L21 4/4 Republic of Korea V850, U200, V200, Z500 KMA/SNU SST Forecast [email protected] GDAPS T2M, PREC, MSLP, T850, U850, Won-Tae Yun 1979/Jan. 2008/Present GDAPS T106L21 OISSTv2/OISSTv2 1/20 Republic of Korea V850, U200, V200, Z500 [email protected] HMC SL-AV Persistent SST/ Dmitry Kiktev 1979/Mar. 2003/Dec. T2M, PREC, MSLP, T850, Z500 10/10 Russia 1.125 x 1.406, L28 Persistent SST [email protected] T2M, PREC, MSLP, T850, U850, IAP9L-AGCM Observed SST/ Huijun Wang IAP 1979/Jan. 2004/Jan. 7/7 V850, U200, V200, Z500, OLR 5.0 x 4.0, L9 IAP-TOGA SST Forecast [email protected] T2M, SST, PREC, MSLP, T850, U850, Observed SST/ David G. Dewitt IRI 1979/Jan. 2005/Dec. ECHAM4.5 T42L19 24/24 V850, U200, V200, Z500, OLR Forecasted SST [email protected] Kiyotoshi Takahashi T2M,SST,PREC,MSLP,OLR,T850, GSM Persistent SST/ JMA 1979/Jan. 2006/Dec. 51/51 Kiyotoshi.takahashi- Z500,U850,V850,U200,V200 TL95, L40 Persistent SST [email protected] PREC, MSLP, T850, U850, V850, METRI AGCM Persistent OISST/ Won-Tae Kwon METRI 1979/Jan. 2008/Present 10/10 U200, V200, Z500 5.0 x 4.0, L17 Persistent OISST [email protected] MGO T2M, SST, PREC, MSLP, T850, U850, Observed SST/ Vadim Matyugin 1979/Jan. 2004/Dec. MGOAM2 T42L14 6/10 Russia(((( V850, Z500, OLR Persistent SST [email protected] MSC T2M, PREC, MSLP, T850, U850, RPN GEM Persistent ERA40-SST/ Normand Gagnon 1969/Jan. 2003/Dec. 10/10 Canada V850, U200, V200, Z500 1.875 x 1.875, L50 Persistent CMC SST [email protected] NASA T2M, SST, PREC, MSLP, T850, U850, NASA-GSFC Predicted SST/ Myong-In Lee 1993/Feb. 2007/Sep. 6/18 U.S.A.. V850, U200, V200, Z500, OLR 2.5x2.0, L34 Predicted SST [email protected] NCEP T2M, SST, PREC, MSLP, T850, U850, Predicted SST/ Jae-Kyung E. Schemm 1981/Dec. 2003/Feb. NCEP CFS T62L64 15/15 U.S.A.. V850, U200, V200, Z500, OLR Predicted SST [email protected] POAMA T2M, SST, PREC, MSLP, T850, U850, Predicted SST/ Oscar Alves 1981/Dec. 2003/Feb. POAMA 1.5 T47L17 15/15 Australia V850, U200, V200, Z500, OLR Predicted SST [email protected] 4 APEC Climate Center 2 High Performance Computing System Korea Meteorological Administration Korea Institution of Science & (Cray X1E) Technology Information (IBM) - IBM System p5 595 server - Storage System : IBM 366 TB DS4700 - Shared storage : 460 TB - Speed : 37 TF - Memory : 1024 GB memory per 1 node 5 APEC Climate Center APCC Dynamical Climate Forecast System (Tier2) Training Phase Forecast Phase t=0 N MME Real-time S = O + ∑ai (Fi − Fi ) i=1 Prediction train 2 The weights are computed at each grid point by minimizing the function: G = ∑()St − Ot t=0 6 APEC Climate Center 3 Procedure of MME Seasonal Prediction System Collection : 17 Models Quality Check Composite: Deterministic Forecast (4 kinds of schemes) Probabilistic Forecast Verification: Application: Graphic •Previous prediction • Index forecast •Hindcast • Statistical downscaling Outlook: Interpretation and Description of Global/Regional Prediction Dissemination: Web information & Backup of the Data and Doc. 7 APEC Climate Center Deterministic MME Schemes Simple Composite Method : 1 ′ P = F SCM Simple composite of individual forecast ∑ i with equal weighting M i Stepwise Pattern Projection Method : 1 ′ Simple composite of individual forecasts, after ˆ SPPM P = ∑ Fi correction by SPPM (improved version of CPPM) M i Multiple Regression Method: ′ MRG Optimally weighted composite of individual forecasts. P = a F The weighting coefficient is obtained by SVD based ∑ i i i regression Synthetic Multi-Model Super Ensemble Method: 1 ′ Weighted combination of statistically corrected ˆ SSE P = ∑αi Fi multi model output M i 8 APEC Climate Center 4 Couple Pattern Projection Method Forecast procedure – Pointwise regression method Y(t) = αXP(t) + β Predictor field X(i, j, t) XP(t) ( model forecast) Predictand Y(t) Y(t) ( observation) Coupled pattern projection coefficient 1 Σ (Y(t) – Ym)•(X(i,j,t)-Xm(i,j)) XP(t) = Σ COR(i,j)•X(i,j,t) COR(i,j)= N i,j σX(i,j) • σY Calculate regression coefficient α, β in training period, forecast a regional climate from projection coefficient of the coupled pattern onto forecast field. 9 APEC Climate Center Implemented MME-S in APCC ¾ SPPM and MME-S was tested on ¾ SPPM and MME-S was tested on prediction of 850 hPa temperature Advantage to using SPPM2 prediction of 850 hPa temperature precipitation using APCC hindcast precipitation using APCC hindcast data for the period 1983-2003 and (1) Computational Estimates data for the period 1983-2003 and operational forecast data for 2006 (per 1 model, 1 variable, 22 years) operational forecast data for 2006 and 2007. and 2007. ¾ SPPM code was transferred to CPPM- OLD version : 72 ¾ SPPM code was transferred to APCC and is now part of the hours APCC and is now part of the CPPM – New version : 12-15 Automated Forecast System. Automated Forecast System. hours MME-S Procedure SPPM v2 : 5 hours (suggestion: If you use 8 cpu simultaneously, it takes 10 STEPSTEP 1: 1: Applying Applying statistical statistical correction correction hours for all models’ usingusing SPPM SPPM to to individual individual models models hindcast and forecast and (1) Prior prediction selection two variables) (2) Second Step: Pattern Projection (3) Optimal choice of prediction (2) Improved skill, especially for STEPSTEP 2: 2: Simple Simple multi-model multi-model precipitation compositecomposite using using available available predictions predictions 10 APEC Climate Center 5 Probabilistic MME Scheme Defining Normal fitting method terciles - For the middle/upper tercile boundary : mean plus 0.43 times the standard deviation Æ μ + 0.43σ - For the lower/middle tercile boundary : mean minus 0.43 times the standard deviation Æ μ -0.43σ Forecast probability A Probability of Above-normal N Probability of Near-normal B Probability of Below-normal 11 APEC Climate Center Monthly 3-Month Forecasts Operationalized The APEC CLIMATE CENTER 4-deterministic and one probabilistic Climate Outlook for August-October, 2008 BUSAN, 23 July 2008 - Synthesis of the latest computer model MME forecasts carried out each month. forecasts for August to October, 2008 at the APEC Climate Center (APCC), located in Busan, Korea, predicts warmer than Best deterministic forecast for that month normal temperatures in Western Europe, Mediterranean, parts of selected on hindcast verification.