270 likes | 462 Vues
Preliminary Results from CaRD10 ( California Reanalysis Downscaling at 10km ) . Masao Kanamitsu and Hideki Kanamaru (Scripps Institution of Oceanography) This work is funded jointly by PIER/NOAA/NSF/ESC. Objective.
E N D
Preliminary Results from CaRD10(California Reanalysis Downscaling at 10km) Masao Kanamitsu and Hideki Kanamaru (Scripps Institution of Oceanography) This work is funded jointly by PIER/NOAA/NSF/ESC
Objective • Reconstruction of the high-spatial resolution /high-temporal scaleanalysisofatmosphere and land covering the state of California, neighboring states and ocean for global change study. • Longest possible analysis • Highest possible resolution
Methodology • Dynamical downscaling • Using a complex atmospheric model to interpolate large scale analysis to regional space and time scale. • Advantage • The fields obtained are dynamically, thermodynamically and hydrologically consistent (difficult for statistical technique). • Can be used to “understand” the dynamics and physics. • Disadvantage • Model dependent. Accuracy uncertain. Importance of validation.
Model and Data • Scripps Experimental Climate Prediction Center Hydrostatic Global to Regional Spectral Model (G-RSM). Highest possible spatial resolution of ~10km. • NCEP/NCAR Global Reanalysis as a large-scale forcing. Only analysis that goes back to 1948. • Apply Scale Selective Bias Correction technique to preserve the large-scale forcing field within the domain. • No other observations, except SST, are used. • Does not incorporate change in land use. • Responses due to changes in large scale atmospheric circulation and SST. • Hourly output.
1. CaRD10 Jan. 1948-Aug. 2005 Completed! Ran on NCAR and multiple national super computing center machines. 2. Extended North American region. In progress Running on Earth Simulator machine in Yokohama, Japan. To be completed by 2006. Two Experimental Phases
Challenges • Computations – solved • Validations and Diagnostics – this presentation • Sensitivity tests – in progress • Improvements – future work
Hourly Buoy Wind Speed Red: CaRD10 Black: Buoy Buoy #2-b42
Daily Mean Buoy Wind Speed Red: CaRD10 Black: Buoy Buoy #2-b42
12Z Jul 21, 1992 00Z Jul 22, 1992 00Z Jul 21, 1992 Catalina Eddy 36.0N CaRD10 31.5N 121W 116.5W Mesoscale data assimilation Thompson et al, 1997
July 10m Wind Diurnal Variation NARR CaRD10
July 2m Temp Diurnal Variation NARR CaRD10
Coast stations Valley stations Mountain stations Selected Stations
Coast Coast Coast Valley Valley Valley Valley Valley Valley Mountain Mountain Seasonal change of correlation of monthly mean 2m T
Coast Coast CoastValley Valley Valley Valley Valley ValleyMountain Mountain Seasonal Change of Correlation of Monthly Mean Precipitation
ObsCaRD10 Decadal Variationin Seasonal Change of Precipitation Coast Red: 1975-99 Black: 1950-74 Valley Mountain
January 2mT trend comparison (°K/Year)(Obs. 1950-1996, CaRD10 1950-1999)
Diurnal Cycle of January T2m Green: 1975-99 Black: 1950-74
Conclusions • California region downscaling is complete for 1948-present (Aug. 2005). • The quality of the downscaled analysis is very promising. • Excellent in winter • Reasonable in summer but not quite good for some variables over some regions. • The downscaled analysis is useful for global change research and applications. • Further validation of analysis is necessary.