The use of dual-frequency microwave links for the detection of sleet

The use of dual-frequency microwave links for
the detection of sleet
Robert Cummings, Graham Upton, & Anthony Holt
University of Essex [Email: [email protected]; Website:];
Background to the research
The effect of sleet on a link
A NERC-funded research project centred on Bolton in the NorthWest of England, showed that it was possible to use dual-frequency
microwave links to obtain reliable estimates of path-averaged
rainfall rates under most weather conditions. The findings have
been confirmed through the subsequent EU-funded MANTISSA
project, for which additional links have been installed in Germany
and Italy.
A snow-flake with its surface covered by a thin film of water is “seen” by a
microwave as though it were an enormous water droplet. The result is an
attenuation that is disproportionate to the amount of water present. We
now present a method for detecting sleet, without knowledge of
temperature. We verify the results by comparison with temperature.
Microwave links are well suited to urban areas, and (straight) valleys
and are thereby complementary to raingauges and radar which face
difficulties in providing reliable estimates in such areas. Raingauges
face the problems of vandalism and urban chasms, radar has
problems of clutter and obscuration.
Our method uses the single frequency estimates of rainfall derived from the
non-linear K = aRb relations. For a given polarization, over the range of
frequencies of interest, the value of b is always greater at the lower
frequency. In the presence of apparently massive raindrops this greater
value for b translates into a greater overestimate from the lower frequency.
Let H and L denote the single-frequency rainfall estimates from the
higher and lower frequencies, respectively. Studying the time series of
gauge-based and link-based estimates (without reference to temperature) we
suggest that (for this pair of attenuations) minutes for which
Rainfall Measurement using the d-f link
A dual-frequency link consists of a pair of microwave transmitters
and receivers, usually located up masts. Fading will occur if the
signal from the transmitter is attenuated by scattering from particles
along its path. The most severe fading is caused by precipitation.
Specific attenuation, K (dB/km), is related to rain rate, R (mm/hr),
by the non-linear formula K = aRb. The parameters a and b depend
critically on frequency and are sensitive to the (unknown) drop
shape, the dropsize distribution, and the temperature.
However, there exist pairs of frequencies for which the difference
in attenuations is nearly linearly related to rain rate and is relatively
insensitive to the unknown rain characteristics. Thus accurate pathaveraged rainfall estimates can be obtained.
Sleet detection using the link
L /H> 1.3
H > 6 (mm/hr)
should be judged as being affected by sleet.
A typical sleet event
The first diagram in Figure 2 shows the time series for an event on
February 17th-18th, 2000. The estimated path-averaged rain rate based on
five nearby gauges (black line) diverges markedly from the corresponding
red/green line (showing the corresponding estimates based on attenuation
difference) during the period 18:00 to 21:00. The part indicated in green is
that estimated by the algorithm as being sleet-affected. The temperatures at
the lower end of the link are indicated in the second diagram. The sleetaffected period corresponds to a dip below 3.5°C.
Typical results
Rainfall for the link path is estimated using a weighted average
from gauges under the link. This is compared with the estimate
obtained using the link method. For each event and for each
frequency, estimates of the baseline attenuation level are made
before and after the event using a standard computer procedure.
Figure 1 summarises the results for 157 events, for a 23km link in
the Bolton region for which the frequencies are 17.6GHz and
12.8GHz (both vertically polarized). The link has its transmitter at
400m above sea level, with the receiver at an altitude of about
130m. In the figure the events subsequently identified as containing
more than 5 minutes of sleet are indicated with blue crosses. The
generally excellent results are marred by occasional over-estimates,
mostly due to sleet.
The relation with temperature
Over the 21 months to which Figure 1 relates, there were 722 minutes with
H/L > 1.3 and H > 6 mm/hr . Of these (see Figure 3) about 95% correspond
to minutes for which the temperature is believed to have been less than 5°C
at the lower end of the link: a reassuring validation of the procedure.
It is a pleasure to acknowledge the support of the European Union through contract EVK1-CT-200060 . We are also very grateful for the suggestions and
support provided by our colleagues on the MANTISSA and Bolton projects.