Transcript (PDF)
Advanced AHW Diagnostics Ryan D. Torn, Univ. Albany SUNY Chris Davis, NCAR • AHW Assimilation System Diagnostics • AHW Forecast Diagnostics • Ready-to-go HWRF Diagnostics Pressure Tendency • Measure of the imbalance in initial conditions • Computed using high-frequency model output. Files processed using short fortran program that writes netcdf tendency file. • Plotted via NCL Observation Verification • Large-scale model errors and biases • Windowing by region can demonstrate more detailed model bias information • Computed within AHW assimilation system, could extend to HWRF if grids are made available • NCL or Matlab plotting Near-TC Verification • Compute errors relative to obs. near TC. Primarily for diagnosing near-storm environmental errors • Done from fortran program that windows observations based on distance to TC • Aggregation using fortran program Inner-Core Verification Wind Speed Temperature Prior Analysis Obs. • Comparison against aircraft and dropsonde obs. in vortex-relative space. Eliminates position-related errors. • Computed via fortran program that reads WRF files and obs. from DA system and writes model estimates at observation locations. • Plotting with Matlab, with transition to NCL Lagrangian Diagnostics Metrics: circulation, thickness, humidity, vertical shear, divergence, precipitation, average winds • Used extensively during PREDICT to evaluate processes that lead to genesis • Each Lagrangian metric is computed using a separate fortran program, could be merged into single application • Plotting via NCL Histogram Verification • Shows conditional biases • In assimilation system, computed via separate fortran program, aggregated via Matlab script, output as netcdf • AHW forecasts have allin-one fortran program • Calculations can include equitable threat scores • Plotting accomplished via Matlab with transition to NCL Davis et al. 2010, WAF Climatology Comparisons • Shows whether the model can replicate the observed distribution of a quantity, shows systematic biases • Easy to compute once you have quantities Davis et al. 2010, WAF Environment Shear Remove effect of storm from estimate of vertical shear between two levels (subscripts 1 and 2) by removing vertical difference of storm-related vorticity (z) and divergence (d) z z for r r0 2 ( 2 1 ) 2 1 ; 0 r r 0 2 1 0 on lateral boundaries d d for r r0 2 ( 2 1 ) 2 1 ; 0 r r 0 2 1 0 on lateral boundaries v kˆ ( 2 1 ); v ( 2 1 ) 200 mb – 900 mb wind difference (and 900 mb relative vorticity): EPAC developing composite venv v2 v1 v v Due to storm 200 mb – 900 mb wind difference with storm removed: EPAC developing composite Davis et al., 2008: MWR HWRF Histograms • Convert ATCF to netcdf • Histogram constructed via Matlab • Plotting either via Matlab or NCL HWRF Histograms • Convert ATCF to netcdf • Histogram constructed via Matlab • Plotting either via Matlab or NCL Future Diagnostics • H*WIND comparisons (either RMS difference or quadrant-by-quadrant fraction coverage, similar to Marchok calculation)