再分析资料对比

https://www.wcrp-climate.org/ICR4/posters/Subramanian_UA-36.pdf

Results
 Initial state of synoptic wind is different in all the three
forcing data especially in the wind speed magnitude.
 Vortex initialization in the model is also very different
from each other in strength, position and shape.
 The vortex eye is clearer in MERRA at 850 mb level
but is much weaker at 500 mb level compared to FNL
and ERA-I.
 Track error analysis shows GFS does better than both
MERRA and ERA-I with MERRA being the second
best.
 MERRA based simulations were able to produce
better MSLP and Vmax for Laila and Phet. They remain
very different for the first 48 hours of simulations but
later converge. Intensity is generally over-predicted
and GFS performs better than ERA-I and MERRA.
 Precipitation patterns differ, with MERRA producing
less rainfall than GFS and ERA-I.
 No single reanalysis data emerges wholesome as
each storm responds differently to each reanalysis
data.