Report on CTB CFS Test and Evaluation Team Activities Team Leads: Jae-Kyung Schemm and Shrinivas Moorthi CPC and EMC, NCEP/NWS/NOAA 32nd Climate Diagnostics and Prediction Workshop October 22-26, 2007 Tallahassee, Florida
CFS Test and Evaluation Team - One of three components of CTB Project Team Mission Perform sensitivity experiments and evaluate the results. Experiments are designed to accelerate the development of CFS. Team Membership: EMC and CPC federal and contract employees, and CTB funded investigators. Two types of experiments are performed on the CTB computer i. CTB funded projects ii. Internal projects proposed by CTB T & E Team members, reviewed and approved by CST Balance will shift from ii) to i) as CTB grants increase.
FY07 accomplishments - 17 experiments proposed and approved by CST. - Most experiments finished and preliminary evaluation done. (4 presentations and 7 posters) - CTB T&E Team has fostered successful collaboration between EMC and CPC. - This should evolve further in FY08 as the competitive grants program grows and additional PIs join CTB. FY07 experiments have provided some insight and recommendations towards the next CFS and CFS RR.
Selected results from experiments and evaluations
Test of RRTM shortwave radiation code (Hou and Saha) 1.AER RRTM-SW radiation scheme has been incorporated into the new GFS/CFS. Also included are other physical processes such as solar-cycle, aerosol effect (including historical volcanic origin) and historical CO2. 2.T126/MOM3 version was run from 01/01/1948 for 60 years. 3. Comparison of global mean SST, T2m anomalies, and global map of SST climatology shows promising improvement over the control model. - Served as a benchmark for other GFS evaluations.
CFS subseasonal hindcast run ( Vintzileos and Pan ) - reported by A. Vintzileos in this session 1.60-day forecast runs in 5-day intervals during May-August, Impact of horizontal resolution (T62, T126, T254) and initial condition (GDAS, CDAS2) examined. Principal findings and recommendations 1.New CFS needs to be accompanied by its own reanalysis. 2.Impact of model resolution detected in tropical Atlantic and Africa - downscaling of operational forecasts could be beneficial in those sectors.
25-year 10-member 6-month T126 CFS runs –Four configurations of T126 CFS: A) CFS/OSU/GR2: - OSU LSM, ICs from GR2 (CONTROL) B) CFS/Noah/GR2: - Noah LSM, ICs from GR2 C) CFS/Noah/GLDAS: - Noah LSM, land ICs from T126GLDAS/Noah D) CFS/Noah/GLDAS-Climo: - Noah LSM, land ICs from GLDAS/Noah climo –25 summers:( ) Initial conditions: 00Z daily from Apr ,30, and May 1-3 –10 winters:(83, 88, 89, 90, 98, 00, 01, 02, 03, 04) Initial conditions: 00Z daily from Nov 29-30, Dec 1-3, Dec For summers 1999 & 2000 only for ensemble size and lead time test Evaluations focused on impact of land state initialization on precipitation and surface air temperature over North America, and global SST - GLDAS land initial condition necessary for Noah LSM Implication - New CFS reanalysis crucial for new CFS CFS with Noah LSM and GLDAS(Mitchell, Yang and Meng) - will be presented by R. Yang on Wednesday Impact on CFS of: A) new land model (Noah LSM vs OSU LSM) B) new land initial conditions (GLDAS vs GR2)
Rerun of the offline Noah Land Model for (Fan and van den Dool) Improvements in the model, vegetation class table and input precipitation datasets. Homogeneous input for both retrospective (from 1948) and real-time analysis. Main findings Soil moisture anomalies and annual cycles are improved in top two layers, but the dry end of annual cycles are about 1~2 month late. Anomalies in lower layers are comparable with old run, however, the amplitudes of annual cycles much larger than Obs as in older version. Slight changes in evaporation and runoff are found in northern US. About 10~20 mm/mon increase (decrease) in evaporation (runoff) is seen in southeastern US.
T382 CFS runs for hurricane season (Saha, Thiaw) AGCM operational NCEP GFS LSM - Noah LSM OGCM - GFDL MOM3 All runs initialized with NCEP/DOE R2 and NCEP GODAS at 0Z, May 15, AGCM spatial resolution in T382L64. Companion runs made in T126L64 and T62L64 also. This experiment has been extensively analyzed by the Evaluation Team members. 1 presentation and 3 posters on tropical storms and monsoon prediction.
Western North PacificNorth Indian AtlanticEastern North Pacific Examples of Storm Tracks for 4 NH Basins
AtlanticBasin Correlations Total: r=0.38 ’81-’93: r=0.15 ’94-’06: r=0.37
EasternPacificBasin Correlations Total: r=0.16 ’81-’93: r=0.04 ’94-’06: r=0.02
SSTA in MDR ASO 2 month lead Year CoCo T62 T126 T382 OBS New CFS Runs 3 different resolutions ICs: May 15 (1 member) Correlation (CFS, OBS) T62 T126 T382 Linear trend removed Warming trend contributes to the high correlation. SST
U200–U850 in MDR ASO 2 month lead Year m/s T62 T126 T382 New CFS Runs 3 different resolutions ICs: May 15 (1 member) T62 T126 T382 OBS Correlation (CFS, OBS) Linear trend removed Lower resolution better wind shear (single run) U200–U850
Summary CFS in T382 resolution exhibits fairly robust climatological seasonal cycle of tropical cyclone over four NH basins. Warming trend and intensification of hurricane activity in the Atlantic basin captured. Increased horizontal resolution may lead to improved vertical wind shear prediction over the Atlantic MDR. Ensemble members needed for better depiction of large scale circulation. Coupled air-sea interaction: important
CFS E. Pacific SST bias corrections (Xie, Wang) Impact of radiation flux error over the Southeastern Pacific examined. SST bias reduced by half from surface radiation bias correction.
CFS hindcast rerun with corrected GODAS for ( W. Wang, Xue, Schemm ) Reruns with June and December initial conditions - Evaluations on SST and atmospheric variables performed. Mean SST is reduced by degree due to GODAS corrections - reduces warm biases but worsens cold biases, particularly for June IC’s. Changes in precipitation bias are confined in the Tropics. SST hindcast skill is improved, most prominent in the Pacific and Indian Ocean for December IC’s and in the Atlantic Ocean for June IC’s. No significant changes in skill over North America for 2m temperature and precipitation. Provided analysis with April IC’s for hurricane seasonal outlook issued in May 2007.
JAS JFM OND SST Indices Hindcast Skill in (red improvement)
SST hindcast averaged in the hurricane main development region (MDR) for July-November starting from June I.C.. The SST hindcast is reduced (green) in and increased (red) in due to GODAS corrections. CFS realistically simulated the peaks of the index in 1987, 1995, 1998 and , and valleys in and 1994, which suggests a potential contribution of the CFS forecast for the hurricane outlook.
North America
Summary CTB T&E Team has fostered successful collaboration between EMC and CPC. FY07 experiments provided some insight and recommendations towards the next CFS and CFS RR. This successful collaboration should evolve further in FY08 with funded projects for the next+1 CFS.
For FY08 - Main focus : To participate in CFS reanalyasis and reforecasts - To experiment with and evaluate the impact of suggested changes to CFS by funded external proposals.