Sources of Error

There are many factors which contribute to rainfall measurements errors using radar, including errors in measuring the reflectivity factor Z, variations in the Z-R relationship and gauge-radar sampling differences.

The errors in measuring the reflectivity factor, Z, include:

  1. Attenuation of the radar beam due to precipitation, gases, cloud, etc., between radar and the precipitation target.
  2. Partial blockage of the transmitted radar beam at low elevations by trees or buildings.
  3. Precipitation targets which do not fill the radar beam.
  4. Anomalous propagation.
  5. Water on the radome which results in attenuation of both the transmitted and received signal.
  6. Random error in the signal estimates. This can be minimised by averaging over a sufficient number of samples.
  7. Incorrect calibration of the radar.

Variations in the reflectivity-rainfall relationship arise from differences in the drop size distribution for the rain event. This distribution can vary significantly for individual storms or rain events, from one day to the next, from season to season and from one location to another. For this reason it is necessary that studies be conducted in the area of concern if accurate rainfall estimates are to be achieved.

Finally there are gauge-radar sampling differences. The sample volume of the radar beam is large in comparison to the rain gauge, particularly at longer ranges, and there can be variations in the rainfall intensity within the sample volume. Consequently, rainfall estimates with radar should be limited to ranges less than 150 km. Furthermore the nominal height of the radar beam is some distance above the rain gauge and the relationship between radar reflectivity and the measured rainfall will vary due to:

  1. Evaporation, coalescence, growth or breakup of rain drops before they reach the surface. Changes in the drop size distribution will result in changes in the radar reflectivity.
  2. Changes in the precipitation type (hail, snow or rain) between the radar sample volume and the gauge. This will depend on the height of the freezing level in addition to the beam height. For example the effective reflectivity factor for snow is quite low while that for water-coated hailstones can be very high.
  3. Horizontal advection of precipitation particles away from the rain gauge due to wind.
  4. Differences in the vertical wind component for individual events.
In practice, for operational rainfall measurement programs the radar should be used in combination with a network of rain gauges with the gauge measurements used to calibrate the radar. This can be done in “real-time” with computer processing of the radar data and the gauge data. With a gauge network density of around 1 per 1500 km2 and a variable radar calibration factor applied across the network it is possible to reduce the average error of radar precipitation estimate to 10-20%.

If a standard Z/R relationship is used for the radar then allowances must be made for situations when the rainfall rate will be overestimated or underestimated.
iDevice icon Reflectivity Exercise
Read the paragraph below and fill in the missing words.

As an operator, you should affected to determine whether estimated rainfall rates for a given area are accurate or inaccurate.

Though some services use a set Z/R relationship, it is likely that each precipitation event has its own Z/R relation. Fill in, for each of the following, if radar would most likely overestimate, underestimate, or accurately estimate effective reflectivity. Assume the radar is operating at specified performance standards.

  • Widespread snow with the air temperature at -5°C is likely to effective reflectivity
  • Large thunderstorm at 180°/25km with many small raindrops and a few very large raindrops is likely to effective reflectivity
  • Large rain shower where all assumptions of the radar equation are met is likely to effective reflectivity
  • Thunderstorm at 40km range with hail and heavy rain is likely to effective reflectivity
  • Small rain shower at 200km range is likely to effective reflectivity
  • Light, continuous rain covers the radome. You are interested in an area 60km from the radar. The radar is likely to underestimate effective reflectivity