PRECIPITATION STRUCTURE FROM GROUND AND SPACE-BASED RADAR OBSERVATIONS

R.V.Calheiros(1), C.A.Morales(2), E.N.Anagnostou(2)

  1. IPMet/FET, University of the State of SPaulo (UNESP), 17033-360, Bauru, SP, Brazil.
  2. Department of Civil and Environmental Engineering, University of Connecticut, Storrs, Ct, USA

 

 

 

 

 

Abstract. The knowledge of the internal structure of storms has gained another powerful tool with the deployment of precipitation radars on board of satellites. This is particularly true for resolving the complex structure of tropical rainfalls. This paper deals with selected storms occurring in central São Paulo, Brazil, as observed by the TRMM satellite precipitation radar (PR) and the Bauru S-Band Doppler radar, and presents an approach to the combined use of both observation tools. The consistently larger degree of vertical layering for the PR, increasing with distance from the Bauru radar, is evident. On the other hand, the greater sensitivity of the ground radar is shown to significantly enhance the composite picture of the storms, providing the outer shell (light rain) of the observed structure. Such detection of the lighter rain is particularly important in areas like the Amazon where it represents a substantial portion of the precipitation (Greco et al, 1994).


 

1.Introduction

With the launch of the Tropical Rainfall Measuring Mission (TRMM) Satellite in 1997 the first space-based weather radar capability became available (Meneghini et al, 1999). The advantage of the simultaneous utilization of that capability together with the one provided by ground radars has already been stressed by previous authors (Bolen and Chandrasekar, 1999, Anagnostou et al, 2000). A new perspective for the observation of rainfall has become available by taking advantage of the appropriate characteristics of each radar. This paper considers basic aspects of a different application, i.e., resolving the structure of convective storms with the combined use of the TRMM PR and of the S-band Doppler weather radar operating at Bauru, Brazil, (22°21’28"S, 49°05’37"W). This area, situated in the central region of the State of São Paulo is subject to intense convective storm activity during the rainy period (October-March), showing the typical high gradients of tropical rainfalls.

The advantages gained were the high vertical resolution of the PR, while from the Bauru radar use was made of its relatively good capability to detect weak precipitation, thus yielding a combined picture of storm structure.

Reflectivity gradients for quasi simultaneous observations of the storms were estimated and analyzed.



Correspondence to: R. V. Calheiros

 

 

 

2. The radars and data

The TRMM PR operates at 13.8GHz, with a peak power in excess of 500W. The antenna beamwidth (3dB) is approximately 0,71° giving a horizontal resolution near the surface of approximately 4.3km. Scanning is restricted to near-nadir to minimize surface clutter problems and gather samples of sufficient size. The nominal range resolution is 250m. The noise base level is about – 111dBm, with a corresponding minimum detectable reflectivity (MDZ) of about 17dBz (0dB S/N).

The Bauru radar radiates about 550KW of peak power in the S-band, and the current operational product has a 1 km range resolution. The 3 dB beamwidth is 2°, and MDZ is less than 13dBZ at 200km.

A set of eight thunderstorms during the periods October to March 1998/99 and 1999/2000 were initially selected taking into account their intensity and electrical activity. CAPPIs at 2km above mean sea level were generated from the PR and the Bauru radar. The time difference between both observations was within a few minutes for all events. Vertical sections along the ground radar radials were chosen to cover the largest extension of rainfall while crossing through high-reflectivity cores.

For the PR, the cross sections were composed from layers of horizontal reflectivity at 250m height intervals. For the Bauru radar, a preliminary estimate of average gradients was made, using a B-scan type cross section for three of the storms with rain areas extending through range intervals smaller than 20km and situated at greater ranges.

3. Processing and results

Statistics of the vertical gradients were computed. Figure 1a presents the curve for the PR; only gradients above 3.0 dBZ/250m were considered, for a total of 270 events. About 60% of the values are concentrated in the interval from > 3 dBZ/250m to 4.5 dBZ/250m still leaving a significant amount of gradients above 4,5 dBZ/250m. For the Bauru radar the resulting curve is presented in Figure 1b, indicating that about 50% of the gradients are smaller than about 7 dBZ/2km. The largest values verified for the horizontal gradients were in the order of 25 dBZ/5km for the PR and 14 dBZ/km for the Bauru radar.

Figures 2 and 3 illustrate an alternative combination of the PR and the Bauru radar observations for the storm of 20 February, 1999. They are composed by the PR image as a core and the boundaries of the Bauru radar, as an outer shell. The cross sections were displaced with respect to each other along the range axis to minimize mismatch problems at the boundary between them. The cross section was chosen to cover as much precipitation as possible, and is not radial.

a) TRMM PR

b) BAURU RADAR

Fig. 1. Cumulative percentage distribution of gradients for the PR (a) and the Bauru radar (b).

 

 

4. Comments and conclusion

Proper caution should be exercised when comparing the gradients shown in the observations of each radar, one major factor being the substantial differences embedded in the geometries of the measurements. In particular, the presentation of the observations from the Bauru radar, while providing enough information for the scope of this work, gives good indications on the gradients. In any case, attempting to keep as much as possible the full resolution of data is always a problem when comparing measurements from different sensors.

Fig. 2. CAPPIs at 2km height of the storm on 20 October 1999 at 23:06 UTC from the Bauru radar (top, left) and the PR (top, right) and combined image with the PR core and the Bauru radar’s outer shell (bottom). Range ring is approximately 150 km.

 

Fig. 3. Cross sections for the storm of 20 October 1999 at 23:06 UTC. The top left image refers to the Bauru radar and the top right one to the PR. The lower part of the Figure shows the combination of the two images with the core being the full PR image and the outer shell superimposed from the Bauru radar.

The issue of the calibration of the radar is an important factor when comparing radars and was approached by Anagnostou et al. (2000). An offset between the PR and the Bauru radar of measured reflectivities is apparent in the data, but did not interfere significantly with the objectives of this work.

The potential for the application of the combined observations in storm structure analysis has been demonstrated. One example of the importance of the availability of high-resolution reflectivity profiles is in the assimilation of radar data in mesoscale forecasting in São Paulo. The numerical model utilized has vertical levels starting with a 100m separation layer, which increases by a 1:2 factor within the boundary layer. Future work should include image fusion algorithms.

 

Acknowledgments. Thanks are due to Maria Andrea Lima for the help in data handling and processing.

 

References

  1. Greco, S., J. Scala, J. Halverson, H.L. Massie, Jr., Wei-Kuo Tao and Michael Garstand: "Amazon Coastal Squall Lines. Part II: Heat and Moisture Transports", Mon. Wea. Rev., 122, 1994, p.625-635.
  2. Meneghini, R., T.Iguchi, T.Kazu, T.Kawanishi, H.Kuroiwa, K.Okamoto and D.Atlas; "The TRMM Precipitation Radar: Opportunities and Challenges", 29th International Conference on Radar Meteorology, Montréal, Canada, 1999, Preprints, p. 621-624.
  3. Bolen, S.M. and V.Chandrasekar; "Comparison of Satellite-based and Ground-based Radar Observations of Precipitation", 29th international Conference on Radar Meteorology, Montréal, Canada, 1999, Preprints, p. 751-753.
  4. Anagnostou, E., C.A.Morales and T.Dinku, "On the Use of TRMM Precipitation Radar Observations in Determining Ground Radar Calibration Biases", 2000, submitted to the Journal of Oceanic and Atmospheric Technology.