Aldo Bellon

High resolution rainfall measurements by radar for very small basins: the sampling problem reexamined

Abstract The dependence of the accuracy of short-period radar rainfall accumulations on periodic sampling of the rainfield is investigated. It is shown that errors due to sampling can be greater than all the other errors combined if accumulations are improperly computed. However, if the movement and the development of the storms are taken into account, these errors can be reduced dramatically. Using an accumulation method taking storm movement and evolution into account, the space and time resolution of radar maps resulting in the most accurate 5 min accumulations over areas smaller than 1 km 2 was determined. It is shown that the best accumulations are obtained with very high time resolution data. Furthermore, for a given time resolution, there is an optimum spatial resolution that minimizes rainfall accumulation errors.

Error Statistics of VPR Corrections in Stratiform Precipitation

Abstract Errors in surface rainfall estimates that are caused by ignoring the vertical profile of reflectivity (VPR) and range effects have been assessed by simulating how fine-resolution 3D reflectivity measurements at close ranges are sampled by the radar at various ranges and heights. Uncorrected and corrected accumulations from 33 events of mainly stratiform precipitation, with a recognizable melting layer for over 250 h, have been generated using two basic procedures: (a) the “near range” or “inner” VPR and (b) the intensity-dependent “climatological” VPR. The root-mean-square (rms) error structure has been derived as a function of height and range, for accumulations ranging from 5 min to 2 h, for various brightband heights and verification areas. However, it is the errors along the lowest default height that are most relevant. The stratification of the results by the height of the bright band is essential to understand the influence of the bright band with range. The largest errors (>100% at near ra...

Forecasting of hourly accumulations of precipitation by optimal extrapolation of radar maps

Abstract The forecast of radar rainfall rate maps is optimized when the maps to be forecast are averaged over an area A = L2 such that L = kTλ where L is in kilometres, T is the forecast period in minutes, and k and λ are empirical coefficients that are usually restricted to 1.0 ⩽ k ⩽ 1.3 and 0.7 ⩽ λ ⩽ 0.8. Application of this relationship to the generation of forecast 1 h rainfall accumulation maps has consistently reduced r.m.s. errors by at least 10% when compared with those derived with the originally unsmoothed data.

The accuracy of short-term radar rainfall forecasts

Abstract A total of 37 weather sequences which passed over the City of Montreal, Quebec, Canada, and which were observed by radar, have been analysed. The object was to determine the accuracy of simple forecasting schemes in giving estimates of the amount of rain an hour or two ahead. Verification was achieved using data from a network of telemetred gages. It was found that the radar-measured accumulations have an inherent error of the order of 25%, 0.5-hr. forecasts have an error of ∼ 50% and 3-hr. forecasts have an error of ∼ 60%.

Real-Time Comparisons of VPR-Corrected Daily Rainfall Estimates with a Gauge Mesonet

Abstract The relative skill of two vertical-profile-of-reflectivity (VPR) correction techniques for daily accumulations on a selected dataset and a real-time dataset has been verified. The first technique (C1) adjusts the 1-h rainfall amounts already derived on a Cartesian CAPPI map at an altitude of 1.5 km in a “one step” procedure using the range-dependent space–time-averaged VPR over the 1-h interval. The C2 technique corrects the nonconvective polar reflectivity measurements of each 5-min radar cycle that are also centered at a height of 1.5 km according to a VPR that is similarly derived but over a shorter time interval. The results emphasize the importance of applying a VPR correction scheme—in particular, in a climatic regime in which most of the liquid precipitation falls from stratiform echoes. The crucial importance of the choice of datasets is also underlined, causing differences in the final assessment that may be greater than those between the various algorithms. Both techniques perform well ...

Measurements of melting layer attenuation at X‐band frequencies

The attenuation at X band caused by the melting layer of stratiform precipitation is derived by comparing long-term average reflectivity profiles from collocated X-band and UHF vertically pointing radars. The total attenuation caused by the bright band is equivalent to the attenuation across approximately 10 km of rain that is present below it. This bright band attenuation is still 3–5 times the “rain equivalent” attenuation, the latter being obtained by integrating the specific attenuation deduced from the entire reflectivity profile across the bright band as if the reflectivity were due to rain. This suggests that in most stratiform rain cases, the total attenuation due to the melting layer exceeds that of the rain below it.

Estimation de lames d'eau spatiales à l'aide de données de pluviomètres et de radar météorologique — Application au pas de temps journalier dans la région de Montréal

Abstract Several areal rainfall estimation methods using rain gage and weather radar data are reviewed including: (1) Thiessens and Kriging methods relying on rain gage measurements only; (2) the classical cumulative procedure after transformation of reflectivity measurements using a standard Z - R relationship for conventional radar measurements alone; and (3) the uniform calibration method (using a constant multiplicative factor or a nonlinear regression) and the simplified cokriging method (previously proposed by the authors) for rain gage-radar combinations. An objective procedure based on the definition of reference rainfall depths at ground level and a set of validation criteria is proposed to compare these methods. The methods have been applied to a sample of eleven daily rainfall events observed in the Montreal area by the McGill weather radar and the Environment Canada rain gage network. The results show that the methods taking into account the spatial variability of rainfall (kriging, simplified cokriging) work much better than more classical approaches. Furthermore, the optimal combination of radar and rain gage information through simplified cokriging leads to better results than each measurement system alone for six out of eleven cases (especially for those presenting high areal variability).

Errors in the Thiessen technique for estimating areal rain amounts using weather radar data

Abstract High-resolution radar data were used to estimate rainfall accumulation patterns for thirteen summer storms. The Thiessen polygon method is used to estimate average rainfall over the Yamaska Basin. Thiessen estimates were compared to estimated radar rainfall averages over the basin. Errors for the thirteen storms analysed gave values between 3% and 69% for the whole basin.

A 9‐year summary of radar characteristics of mesocyclonic storms and of deep convection in southern Québec

Abstract Nine summers of reflectivity and Doppler data, (1993–2001), archived by the Marshall Radar Observatory of McGill University have been processed for the purpose of identifying the location, strength and frequency of occurrence of severe weather events associated with convective activity. A mesocyclone detection and tracking algorithm has identified 329 features with a lifetime of at least 10 minutes. The distribution of their duration, path length, diameter, depth, rotational velocity and shear are provided, as well as their hourly and monthly frequency. Their geographical distribution reveals an increased relative probability of occurrence over the western area of the McGill radar coverage as well as to the north, near Mirabel, and a reduced probability of occurrence near the metropolitan area of Montréal. The analysis of deep convection has been performed in terms of upper level Vertically Integrated Liquid water content maps (UVIL) where the vertical integration of reflectivity begins at a height of 5 km. UVIL maps are preferred among other types of reflectivity‐based radar products because they are the least affected by radar artefacts such as bright band, ground echoes, anomalous propagation and beam blocking. An inherent bias, present in the standard generation of UVIL maps, and due to an oxygen attenuation overcorrection and to the vertical profile of reflectivity of thunderstorms as sampled by a broadening beamwidth, has been identified and corrected. The geographical distribution of UVIL measurements in terms of the time that a UVIL threshold of 10 and 15 kg m−2 is exceeded has been obtained for both the individual years and for the entire 9‐year period. Areas of preferred convection have been identified in the latter, particularly along the flatlands of the St. Lawrence River Basin. A relative minimum is quite noticeable along a broad corridor running north and south of the radar over heavily forested mountainous regions. This combination results in a non‐negligible correlation of the order of ‐0.2 between the height of the topography (m) and deep convection. Curves revealing the relative severity of the nine years in terms of a continuous UVIL threshold have also been derived. The years 1994, 1998 and 1999 emerged as the most active years while 1996 and 2000 were clearly very quiet.

A radar‐based methodology for preparing a severe thunderstorm climatology in central Alberta

Abstract The Radar data Analysis, Processing and Interactive Display (RAPID) system, developed by McGill University researchers, synthesizes spherical coordinate radar data onto Cartesian maps displaying Constant Altitude Plan Position Indicator (CAPPI) reflectivity, Vertically Integrated Liquid water content (VIL), and other radar‐based parameters. In this study, Carvel radar (53.34°N, 114.09°W) data from July 2000 were processed using McGills RAPID software. Specifically, we compared observations of severe convection, as identified by selected radar‐based reflectivity parameters, with surface severe weather reports and atmospheric sounding data. July 2000 was characterized by frequent severe thunderstorm activity over central Alberta; there were seven days with golfball‐sized hail, and two days with confirmed tornadoes. The VIL, upper level VIL (UVIL), and the maximum reflectivity at 7 km (Z7) were employed to quantify the strength and frequency of storms within a 120‐km radial distance of Carvel. For each day, the intensity of the convection was quantified by counting the total number of 1‐km2 pixels in the study area that exceeded the severe thresholds for VIL, UVIL and Z7. The severe thunderstorm algorithms were found to be very effective at correctly identifying the observed severe thunderstorm events. All three radar parameters indicated a diurnal cycle, with severe convection starting after noon and peaking between 16:00 and 18:00 Local Daylight Time. A positive correlation was evident between the observed storm severity and the daily UVIL and Z7 pixel counts. The daily UVIL and Z7 pixel counts were also positively correlated with the Convective Available Potential Energy (CAPE) calculated from proximity soundings.