Evaluation of precipitation forecasts by polarimetric radar

Over the last years, weather services have developed a new
generation of high resolution mesoscale numerical weather
prediction (NWP) models with the aim to explicitly predict
convection. New methods are required to validate the representation
of precipitation processes in these NWP models against
observations. Polarimetric radar systems are especially suited for
model validation as they provide information on the intensity and
the microphysical characteristics of a precipitation event at a
high temporal and spatial resolution. However, the observations can
not be directly employed for model evaluation as polarimetric radar
systems do not explicitly measure the parameters represented in
microphysical parameterization schemes. In order to establish a
relationship and allow for a direct comparison between the model
parameters and the observations, the polarimetric radar forward
operator SynPolRad (Synthetic Polarimetric Radar) has been
developed. SynPolRad simulates synthetic polarimetric radar
quantities out of model forecasts which permits an evaluation in
terms of observed quantities. In a first step, the synthetic
reflectivity, LDR, and ZDR are computed from predicted bulk water
quantities and in a second step, the beam propagation in the model
domain is simulated under consideration of refractivity and
attenuation effects. In order to successfully employ SynPolRad for
model evaluation purposes, the link between the forward operator
and the mesoscale model has to conform as closely as possible to
the model assumptions. However, in the case of a polarimetric radar
forward operator not all the input parameters are defined by the
model. Within this work, these free parameters are derived on
theoretical terms accordingly to the model assumptions such that
the polarimetric quantities match the thresholds of a hydrometeor
classification scheme. Furthermore, special care is given to the
representation of brightband signatures. The application of
SynPolRad on two case studies proves the potential of the new
method. A stratiform and a convective case study are chosen to
assess the ability of mesoscale models to represent precipitation
in different dynamical regimes. LMK (Lokal-Modell-Kürzestfrist) and
MesoNH (Mesoscale Non-Hydrostatic Model) simulations considering
different microphysical parameterization schemes are evaluated. The
evaluation concentrates on the representation of life cycle,
intensity, and the spatial distribution of synthetic reflectivity,
LDR, and ZDR. Furthermore, hydrometeor types derived from the
observed and synthetic polarimetric quantities employing a
classification scheme are compared. Large discrepancies are found
between the model simulations and the observations. However, the
consideration of an additional ice hydrometeor category in the 3
component scheme significantly improves the performance of the LMK.