overestimation of precipitation. This can produce range-dependent biases in the
radar rainfall estimates [11, 13, 15, 53].
A number of techniques have been proposed to deal with the VPR and BB effects
on radar measurements, including climatological corrections depending on seasons
[54, 55] or rain types [56], characterization of VPR by estimating the altitude and
peak of the BB [57], and retrieval of the VPR by filtering the beam-sampling effects
from the comparison of radar reflectivity at different distances and altitudes
[58, 59]. Fabry and Zawadzki [60] studied the structure of the BB using long-term
observations from a vertically pointing radar with high spatial (15 m) and temporal
(2 s) resolutions. Their results highlighted the importance of the shape, density and
fall speed of the ice particles in the existence of a BB. Andrieu and Creutin [58]
proposed an inverse method for retrieving VPR from a two-elevation scanning radar
based on reflectivity ratio (reflectivity at high elevation divided by reflectivity at low
elevation) function. However, this method assumes VPR homogeneity. Further
improvements in the VPR retrieval methods have been developed by using volume
radar scans (scans taken a different elevation angles) [61–65]. A full-volume scan
radar provides an estimate of the VPR, which can be smoothed by the characteristics
of the radar beam. Bellon et al. [66, 67] highlighted the influence of the spatial
heterogeneity of VPRs on the resulting corrections associated with volumetric
sampling strategy. Kitchen et al. [57] developed a method to correct the BB by
using an idealized reflectivity profile convolved with the radar beam power profile.
The current UK operational correction method is based on Kitchen et al.’s [57]
algorithm which relies on forecasts of freezing level heights in addition to a fixed BB
thickness of 700 m. However, this approach does not allow for spatial and thickness
irregularities in the BB which can occur due to atmospheric variability. Smyth and
Illingworth [68] have emphasized that it is important to use a correction procedure
which uses different VPRs for different precipitation types (i.e. stratiform and
convective). Moreover, the advance of polarimetric radar enabled new techniques
to classify hydrometeors for BB correction, such as the decision tree method, classic
statistical decision theory, neural network techniques and fuzzy logic [43, 69]. RicoRamirez et al. [70] developed a fuzzy logic classifier based on S-band DP radar
measurements to identify the BB and showed that the combination of this classifier
with Kitchen et al.’s [57] algorithm can be used to identify and remove the BB. This
algorithm has also been implemented at operational C-band frequencies and has
shown some skill in identifying and removing the BB [71].
2.5 Variations of the DSD and Radar Rainfall Estimation
Additional errors and uncertainties could be introduced when converting the radar
reflectivity Z into an estimate of precipitation intensity R at ground level [13, 19, 72,
73]. The general form of the Z – R relationship is a power-law given by Z ¼ aR
b ,
where a and b are the parameters that depend on the DSD. The DSD parameters are
obtained empirically by establishing a climatological Z – R relationship or by
244
N. Nanding and M. A. Rico-Ramirez
radar rainfall estimates [11, 13, 15, 53].
A number of techniques have been proposed to deal with the VPR and BB effects
on radar measurements, including climatological corrections depending on seasons
[54, 55] or rain types [56], characterization of VPR by estimating the altitude and
peak of the BB [57], and retrieval of the VPR by filtering the beam-sampling effects
from the comparison of radar reflectivity at different distances and altitudes
[58, 59]. Fabry and Zawadzki [60] studied the structure of the BB using long-term
observations from a vertically pointing radar with high spatial (15 m) and temporal
(2 s) resolutions. Their results highlighted the importance of the shape, density and
fall speed of the ice particles in the existence of a BB. Andrieu and Creutin [58]
proposed an inverse method for retrieving VPR from a two-elevation scanning radar
based on reflectivity ratio (reflectivity at high elevation divided by reflectivity at low
elevation) function. However, this method assumes VPR homogeneity. Further
improvements in the VPR retrieval methods have been developed by using volume
radar scans (scans taken a different elevation angles) [61–65]. A full-volume scan
radar provides an estimate of the VPR, which can be smoothed by the characteristics
of the radar beam. Bellon et al. [66, 67] highlighted the influence of the spatial
heterogeneity of VPRs on the resulting corrections associated with volumetric
sampling strategy. Kitchen et al. [57] developed a method to correct the BB by
using an idealized reflectivity profile convolved with the radar beam power profile.
The current UK operational correction method is based on Kitchen et al.’s [57]
algorithm which relies on forecasts of freezing level heights in addition to a fixed BB
thickness of 700 m. However, this approach does not allow for spatial and thickness
irregularities in the BB which can occur due to atmospheric variability. Smyth and
Illingworth [68] have emphasized that it is important to use a correction procedure
which uses different VPRs for different precipitation types (i.e. stratiform and
convective). Moreover, the advance of polarimetric radar enabled new techniques
to classify hydrometeors for BB correction, such as the decision tree method, classic
statistical decision theory, neural network techniques and fuzzy logic [43, 69]. RicoRamirez et al. [70] developed a fuzzy logic classifier based on S-band DP radar
measurements to identify the BB and showed that the combination of this classifier
with Kitchen et al.’s [57] algorithm can be used to identify and remove the BB. This
algorithm has also been implemented at operational C-band frequencies and has
shown some skill in identifying and removing the BB [71].
2.5 Variations of the DSD and Radar Rainfall Estimation
Additional errors and uncertainties could be introduced when converting the radar
reflectivity Z into an estimate of precipitation intensity R at ground level [13, 19, 72,
73]. The general form of the Z – R relationship is a power-law given by Z ¼ aR
b ,
where a and b are the parameters that depend on the DSD. The DSD parameters are
obtained empirically by establishing a climatological Z – R relationship or by
244
N. Nanding and M. A. Rico-Ramirez