Weighing gauge algorithms in common use by meteorological services and others traditionally apply a fixed threshold to the difference between the most recent and previous measurements to determine whether precip has occurred. The threshold, intended to eliminate noise primarily caused by wind pumping, will vary from gauge to gauge and region to region.

Some algorithms allow the measured weight to decrease as well as increase where others permit increases only. Where decreases are permitted, evaporation becomes a factor unless prevented with a layer of oil. Where decreases are not permitted, the diode effect will allow any noise which exceeds the threshold to appear as increases in weght. Depending on the characteristics of an algorithm, light precip which does not exceed the threshold may be lost completely or may eventually be counted much later than the actual event.

Four new algorithms have been developed, three at the Meteorological Service of Canada. The fourth was developed earlier by the Swedish Meteorological Institute. These algorithms require minutely data, which is normally an average of five second samples.

Although this effort was made necessary with the incorporation of the very sensitive Geonor weighing gauge into the Environment Canada surface weather/RCS network, the results may be applied to other gauges.

Method 1 is a filter method where the minutely averages are further averaged in a succession of boxcar filters. Methods 2 and 3 are noise threshold methods, using minutely means as input, which take some combination of maxima, minima and a noise related term over some length of time to test against a previous accumulation to determine if accumulation should be updated. Method 4 is a filter and noise threshold combination method. Here, input to the algorithm is from a 9 minute boxcar whose input is the minutely means.