Campbell Scientific PWS100 Present Weather Sensor - Manual (Page 94)

Section 8. Functional Description
size/velocity value of 0 for drizzle, 0.4 for rain, 0.3 for snow flakes, 0.4 for
graupel, 0.1 for ice pellet etc. The fuzzy value assigned is always between 0
and 1. Since certain particle types are mutually exclusive then some types will
be assigned 0 based on pure logic (e.g., a particle > 0.5 mm diameter will have
a rain value that can be >0 whereas the drizzle value for this type will always
be 0).
Once each possible particle type has been assigned a fuzzy logic value for all
of the parameters the values are multiplied for each particle to give 9 particle
type scores between 0 and 1. The highest value is more likely to be the correct
particle and so this type is assigned to the individual particle. If two values are
the same then the particle is of unknown type and is assigned a type as such.
Over the measurement period there are likely to be a number of particles falling
through the detection volume. To define a precipitation type the individual
particles are binned into 11 classes. 9 specific types, 1 unknown and 1 error.
The unknown classification is used for particles which appeared to have been
measured correctly but either have fuzzy logic classifications that are not
unique OR fall outside the realms of possibility for natural particles (e.g., they
are measuring as travelling too fast for their size, or perhaps a particle appears
to be classified as a snow flake at very high temperatures). Measurements
where two particles have overlapped in the optical path are also classified as
unknown. Unknown particles’ size and speeds measurements are excluded
from statistical calculations, but as at least one particle has definitely been
detected the rainfall intensity and total measurements are corrected by adding
one extra particle with the characteristics of the average of the good particles
falling at that time.
The error classification is used for particles where various quality checks
determine the measurement to be suspect (e.g., a particle is moving upwards,
the signal to noise ratio is too large, the particle has fallen through the edge of
the measurement zone or the measurements are off scale in some way). In
general particles with the erroneous classification are not likely to be real or
should not be counted (upwards moving for instance) so these are not included
in any statistics nor is any correction made to rainfall rates or totals.
During normal measurements a small fraction of unknown or error particles
will be counted. There will always be a small fraction of error particles for
instance which fall through the edge of the optical path. The proportions of
these particles can increase with the type of precipitation and weather
conditions (e.g., under heavy rainfall conditions both more unknown and error
particles will be counted). This is due to a larger proportion of particles
clipping the edge of the optical path (as they are larger) and an increased risk
that two particles will pass through the optical path at the same time. Similarly
where there are very windy conditions more particles will not fall through all
four light sheets so cannot be properly measured.
8.6.2 Precipitation Intensity
Precipitation intensity calculations are possible because of the instruments
accurate analysis of particle size. Over the defined period of time, the volume
of all particles is summed and precipitation rate output is then given as an
equivalent mm h
-1 measurement.
8-8