Characteristics of the raindrop size distribution in seven tropical cyclones have been studied through impact-type disdrometer measurements at three different sites during the 2004-06 Atlantic hurricane seasons. One of the cyclones has been observed at two different sites. High concentrations of small and/or midsize drops were observed in the presence or absence of large drops. Even in the presence of large drops, the maximum drop diameter rarely exceeded 4 mm. These characteristics of raindrop size distribution were observed in all stages of tropical cyclones, unless the storm was in the extratropical stage where the tropical cyclone and a midlatitude frontal system had merged. The presence of relatively high concentrations of large drops in extratropical cyclones resembled the size distribution in continental thunderstorms. The integral rain parameters of drop concentration, liquid water content, and rain rate at fixed reflectivity were therefore lower in extratropical cyclones than in tropical cyclones. In tropical cyclones, at a disdrometercalculated reflectivity of 40 dBZ, the number concentration was 700 Ϯ 100 drops m Ϫ3 , while the liquid water content and rain rate were 0.90 Ϯ 0.05 g m Ϫ3 and 18.5 Ϯ 0.5 mm h Ϫ1 , respectively. The mean mass diameter, on the other hand, was 1.67 Ϯ 0.3 mm. The comparison of raindrop size distributions between Atlantic tropical cyclones and storms that occurred in the central tropical Pacific island of Roi-Namur revealed that the number density is slightly shifted toward smaller drops, resulting in higher-integral rain parameters and lower mean mass and maximum drop diameters at the latter site. Considering parameterization of the raindrop size distribution in tropical cyclones, characteristics of the normalized gamma distribution parameters were examined with respect to reflectivity. The mean mass diameter increased rapidly with reflectivity, while the normalized intercept parameter had an increasing trend with reflectivity. The shape parameter, on the other hand, decreased in a reflectivity range from 10 to 20 dBZ and remained steady at higher reflectivities. Considering the repeatability of the characteristics of the raindrop size distribution, a second impact disdrometer that was located 5.3 km away from the primary site in Wallops Island, Virginia, had similar size spectra in selected tropical cyclones.

Error characteristics of rainfall measurements were studied using six collocated Joss–Waldvogel (JW) disdrometers that are located at NASA’s Wallops Flight Facility. The six disdrometer means of rain rate R, reflectivity Z, and differential reflectivity ZDR, for a given minute were considered as a reference. The maximum deviations of R, Z, and ZDR from the mean in a rain event were 0.6 mm h−1, 1.3 dB, and 0.05 dB, respectively. Rainfall statistics were then examined between disdrometer pairs. The root-mean-square (rms) difference of R, Z, and ZDR between paired disdrometers in a rain event were as high as 3.2 mm h−1, 3.7 dB, and 0.3 dB, respectively. The rms difference of R and ZDR were even higher when the disdrometer observations were stratified based on reflectivity intervals. The differences in disdrometer rainfall measurements have a potential impact when the disdrometers are considered as calibration tools for vertically pointing and scanning radars. The differences between the disdrometer measurements also result in differences in coefficients and exponents of the derived relations between radar parameters and rain rate. Among the four different relations between radar parameters and rain rate, the absolute difference in rain rate |ΔR| from two different JW disdrometers was highest in R(ZH, ZDR) and lowest in R(KDP, ZDR). The other two relations were R(Z) and R(KDP). The |ΔR| increases with increasing horizontally polarized reflectivity ZH, and differential specific phase KDP in both single- and dual-parameter rainfall estimators, while the |ΔR| increases with decreasing ZDR in dual-parameter rainfall estimators. Several sources of JW disdrometer malfunctions were also presented. The hardware problems were the leading cause for the malfunction of the JW disdrometers, as identified by the manufacturer. A single JW disdrometer could have inherent measurement errors that can only be identified in the presence of collocated (preferably two) rain-measuring instruments.

An experimental study of small-scale variability of raindrop size distributions (DSDs) has been carried out at Wallops Island, Virginia. Three Joss-Waldvogel disdrometers were operated at a distance of 0.65, 1.05, and 1.70 km in a nearly straight line. The main purpose of the study was to examine the variability of DSDs and its integral parameters of liquid water content, rainfall, and reflectivity within a 2-km array: a typical size of Cartesian radar pixel. The composite DSD of rain events showed very good agreement among the disdrometers except where there were noticeable differences in midsize and large drops in a few events. For consideration of partial beam filling where the radar pixel was not completely covered by rain, a single disdrometer reported just over 10% more rainy minutes than the rainy minutes when all three disdrometers reported rainfall. Similarly two out of three disdrometers reported 5% more rainy minutes than when all three were reporting rainfall. These percentages were based on a 1-min average, and were less for longer averaging periods. Considering only the minutes when all three disdrometers were reporting rainfall, just over one quarter of the observations showed an increase in the difference in rainfall with distance. This finding was based on a 15-min average and was even less for shorter averaging periods. The probability and cumulative distributions of a gamma-fitted DSD and integral rain parameters between the three disdrometers had a very good agreement and no major variability. This was mainly due to the high percentage of light stratiform rain and to the number of storms that traveled along the track of the disdrometers. At a fixed time step, however, both DSDs and integral rain parameters showed substantial variability. The standard deviation (SD) of rain rate was near 3 mm h 21 , while the SD of reflectivity exceeded 3 dBZ at the longest separation distance. These standard deviations were at 6-min average and were higher at shorter averaging periods. The correlations decreased with increasing separation distance. For rain rate, the correlations were higher than previous gauge-based studies. This was attributed to the differences in data processing and the difference in rainfall characteristics in different climate regions. It was also considered that the gauge sampling errors could be a factor. In this regard, gauge measurements were simulated employing existing disdrometer dataset. While a difference was noticed in cumulative distribution of rain occurrence between the simulated gauge and disdrometer observations, the correlations in simulated gauge measurements did not differ from the disdrometer measurements.

No abstract

A comparative study of daily and monthly rainfall between research and operational gauges was conducted at the mid-Atlantic region. Fifty research tipping-bucket gauges were deployed to 20 sites where each site had dual or triple gauges. The gauges were in place to validate the National Aeronautics and Space Administration’s newly developed polarimetric radar rainfall estimate. For logistic purposes, these research gauges were collocated with operational gauges and were operated over a year at each site. Therefore, this is an experimental study, which involves a mixture of one to five sites of seven operational gauge networks. A very good to excellent agreement between the two collocated research gauges at daily time scale raised the authors’ confidence to consider them as a reference before comparing with the operational gauges. Among operational networks, the National Weather Service’s (NWS) Automated Surface Observing Systems (ASOS) weighing bucket and the Climate Reference Network and Forest Services tipping-bucket gauges demonstrated high performance for both daily and monthly rainfall, while the Federal Aviation Administration’s Automated Weather Observing Systems (AWOS) tipping-bucket gauges performed poorly. Among the other networks, the ASOS tipping-bucket and Cooperative observer program’s stick gauges seemed to be reliable for monthly rainfall, but not always for daily rainfall. The Virginia Agricultural Experimental Station (VAES) tipping-bucket gauges, on the other hand, had a mixture of high and low performance for daily and monthly rainfall. Unlike other gauge networks, VAES gauges were in place for long-term research applications.