Warm SST Bias over SE Pacific in CFS CMIP Simulations

1. Assessing CFS Model Bias Associated with the Marine Stratus over Southeastern PacificP. Xie1), W. Wang1), W. Higgins1), M. Cronin2), P.A. Arkin3), and R. Weller4)1) NOAA Climate Prediction Center2) NOAA Pacific Marine Environmental Laboratory3) ESSIC, Univ. of Maryland4) Woods Hole Oceanographic InstitutionAcknowledgements:Mingyue Chen, Ake Johansson, and Suranjana Saha

2. Warm SST Bias over SE Pacificin CFS CMIP Simulations

3. Questions for This Year: • How long does it take for the warm bias to develop in the CFS forecasts from its initial condition to what is observed in CFS/CMIP? examining CFS hindcasts • To what extent the warm bias is caused by the insufficiently simulated stratus clouds over SE Pacific? performing CMIP simulations with radiation over SE Pacific corrected

4. Examining the Evolution of Warm SST Bias • Looking at CFS hindcasts with different forecast leading times and comparing them the with initial (GODAS/R2) and balanced conditions (CMIP) • Daily and monthly climatology of CFS hindcasts for a 23-year period from 1981 to 2003 • Initiating date / month • Leading time from 1-45 days and 0 – 8 months for the daily and monthly hindcast climatologies, respectively

5. Evolution of the Warm SST Bias[Monthly Climatology of CFS Hindcasts] • Averaged with respect to different leading months using hindcasts with all different initiating months; • Warm bias starts from coastal areas and spreads steadily toward NW; • After 8 months, warm bias reaches very close to its balanced condition (CMIP)

6. Evolution of the Warm SST Bias[Daily Climatologyof CFS Hindcasts] • Warm bias grows quite fast along the coast during the first month

7. Mean SST Bias over SE Pacific • Evolution of mean warm SST bias over the SE Pacific as seen from the daily (top) and monthly (bottom) CFS hindcasts • Warm SST bias reaches 80% of that in the CFS/CMIP after 8 months

8. Bias Evolution in the Hindcasts with Different Initial Months • Warm SST bias starts at different coastal regions and evolves through different processes

9. SEP Mean Bias Evolution for Hindcasts with Different Initiating Times • Warm bias develops relatively quickly in the first month • Warm bias grows very fast during boreal spring • Warm bias reaches close to the balanced condition from the spring of the second year • Bias evolution seasonally dependent

10. Bias Evolution Near the Coast [76oW;13oS] • Warm bias reaches close to its balanced condition in the first month • Warm bias may grow at a rate of more than 0.1oK/day during the first 30 days of hindcasts (e.g. in September)

11. Bias Evolution Over Open Ocean [85oW;20oS] • Warm bias develops relatively fast in the first 30 days (but not as much as that over the coastal region) • Bias evolution regionally dependent

12. Evolution of Cloudiness in the CFS Models • Averaged with respect to different leading times using hindcasts with all different initiating months • Cloudiness reached very close to its balanced condition in the first month

13. Evolution of Net Radiation in the CFS Model • Net downward radiation biased in the initial condition • Adjusted to its balanced condition in a month • Positive radiation bias over a narrow band along the coast

14. Radiation and SST Biases for Different Months • Monthly hindcasts climatology for the first targeting month • Seasonal variations in location and distribution for both the SST and the radiation biases during early months of forecasts • The SST and the radiation biases are co-located through the seasonal cycle • Excessive incoming radiation caused by insufficiently simulated stratus clouds is a major contributor to the warm SST bias

15. Radiation and SST Biases in the Hindcasts of Diff. Leading Months • Positive radiation bias confined within coastal areas throughout the forecast periods • Warm SST bias spreads out from the coastal region toward NW during the forecast period • Other mechanisms responsible for the spreading the warm bias

16. Examining the Impacts of the Insufficiently Simulated Stratus Clouds over SE Pacific • Quantifying the stratus clouds’ contributions to the warm SST bias under balanced condition (CMIP) • Performing CFS CMIP simulations with radiation budget over SE Pacific corrected with daily climatology of observed radiation (SRB) • Comparing radiation-corrected CMIP simulations with the standard CMIP runs

17. Annual Mean Bias • The warm SST bias over the SEP is reduced by about half

18. Interannual Variability • Interannual standard deviation of monthly SST • Tropical interannual variability is weakened with the reduction of the warm SST bias over SE Pacific

19. ENSO • NINO3.4 index • Intensity of ENSO reduced compared to the standard CFS CMIP simulations and observations

20. Answers (?) • How Long Does it take to develop the warm bias? • Seasonal and regionally dependent • Faster during spring and reaches close to balance spring of the 2nd year • More rapidly over coastal region than over open ocean • On average, the warm bias reaches 80% of its balanced condition after 8 months • To what extent the insufficiently simulated stratus clouds play in forming and maintaining the warm bias • Triggering the warm bias over the narrow band along the coast • Providing excessive energy to the model ocean over the coastal region • Other mechanisms (oceanic and ar-sea interactions) responsible for re-distributing the energy and spreading the warm SST bias • Warm bias reduced by ~half if the excessive incoming radiation is corrected • Insufficiently simulated stratus clouds not the only factor contributing to the warm SST bias