SR50AT-Series Sonic Ranging Sensors
20
values, a 0.0 reading is usually output. For snow depth outputs, the error
indicator value is –999. An invalid temperature reading is also indicated by a
–999 reading. For snow depth applications, these can be easily filtered out
when analyzing the data.
Consideration should be taken in a control type application to deal with invalid
readings. For example, if the sensor is used to initiate a water level alarm,
multiple readings should be used to ensure that a single invalid reading does
not trigger the alarm condition.
9.2.2 Data Filtering
There are scenarios where the SR50AT can produce values with higher than
expected errors. For example, in very low density snow, very little echo is
returned back to the sensor. The increase in echo quality numbers is an
indication of the weak signals. Under these circumstances, an SR50AT can
under, or over, estimate snow depth. If the signal is too weak, the sensor will
also output a value of 0 for the distance to target. When the echoes are weak,
the sensor also automatically increases sensitivity. This makes the sensor more
prone to the occasional erroneous reading from flying debris, drifting snow, or
mounting hardware just outside the beam angle.
The reason not to average values is that occasionally a number with a very high
error value is produced, skewing the average. The value should be ignored and
not averaged. Based on experience, the best technique to eliminate errors and
filter out high error readings is to take the median value. This technique also
helps to automatically filter out zero readings that can occasionally be
produced.
For example, for a given station, a reading is taken every 5 seconds for 1
minute and the median value is taken from the readings. All the programming
examples in this manual use this method for data filtering.
If 11 consecutive values are as
follows for snow depth
After being sorted from
low to high
0.33
–1.1
0.34
0.10
0.35
0.28
–1.1 (erroneous reading)
0.32
2.0 (erroneous reading)
0.33
0.37
0.33
0.28
0.34
0.36
0.35
0.10 (high error value)
0.36
0.33
0.37
0.32
2.0
The best course of action would be to ignore the five lowest values and take the
6th value (0.33).