Presentation on theme: "Tropical Cyclone Intrinsic Variability & Predictability Gregory J. Hakim University of Washington 67th IHC/Tropical Cyclone Research Forum 6 March 2013."— Presentation transcript:
Tropical Cyclone Intrinsic Variability & Predictability Gregory J. Hakim University of Washington 67th IHC/Tropical Cyclone Research Forum 6 March 2013 Q: What is the TC predictability limit? A: We do not know.
Weather Predictability Limits No theory or predictability limits exist for the TC forecast problem no basis for projecting improvements; devoting resources; etc. Lorenz (1982)
Predictability of First and Second Kind applied to tropical cyclone prediction Two types of predictability (Lorenz 1975) : First kind: initial conditions –E.g. weather forecasts with fixed SST Second kind: boundary conditions –E.g. ENSO; CO2, aerosol, orbital, etc. forcing on climate Applied to tropical cyclones: First kind: “intrinsic” TC-scale initial conditions –Internal storm dynamics Second kind: environmental “boundary conditions” –SST, shear, dry air intrusions, etc.
Motivation Tropical cyclone forecasts: Track: steady improvement –better large-scale models & data assimilation Intensity: much slower improvement –despite improved large-scale environment –cf. ``environmental control'’ Emanuel et al. (2004) Why? Need to understand intrinsic variability. –variability independent of the environment –what aspects are predictable? What timescales? –data assimilation key to realizing predictability, but first need to know limits.
Method Idealized numerical modeling –Necessary to control environment –CM1 model (George Bryan) –Axisymmetric and 3D (not shown; similar to axi) Simulate statistically steady state –Extremely long simulations (500 days) –Robust sampling Variability: EOFs & regression Predictability: inverse modeling & analogs
Maximum Wind Speed “superintensity” is a transient effect wide range of intensity in steady state
Azimuthal wind variability Bursts of stronger wind that move inward Dominant period ~4-8 days
Azimuthal wind leading EOFs EOF1: radial shift of RMW EOF2: intensity pulsing at RMW
RMW variability linked to far field Bands of stronger/weaker wind move radially inward Eyewall replacement cycles
Predictability Autocorrelation Analogs (divergence of similar states) Linear inverse modeling Estimate M statistically (least squares) Verify forecasts from independent data
Predictability: LIM Predictability limits: Clouds: ~12-18 hours Azimuthal wind: ~ 2-3 days radial wind azi wind temperature cloud water
Analog Forecasts Fully nonlinear model Similar results to LIM –larger initial error due to limited sample
Comparison against operational forecasts (NHC) Coincidence? Already at predictability limit? Intrinsic variability dominates error?
Conclusions Intrinsic variability –Promotes understanding how environment affects storms –convective bands form in the environment and move inward Intrinsic predictability –~48-72 hours –environment can add or subtract from this limit –compares closely with operational forecast errors Basic research needed! Hakim, G. J., 2011: The mean state of axisymmetric hurricanes in statistical equilibrium. J. Atmos. Sci., 68, 1364--1376. Hakim, G. J., 2013: The variability and predictability of axisymmetric hurricanes in statistical equilibrium. J. Atmos. Sci., 70, in press. Brown, B. R., and G. J. Hakim, 2013: Variability and predictability of a three-dimensional hurricane in statistical equilibrium. J. Atmos. Sci., 70, accepted.