Low Precipitation Observational Bias at Cooperative Climatological Stations1

An upward bias exists in probabilities of 0 or trace weekly total precipitation since small amounts often occur undetected and are recorded as 0 or trace at climatological stations where observations are made only once a day. Exact evaluation and correction of this bias is difficult, but individual estimates of 1-week P(0,T) for substations in the north-central region of the United States can be multiplied by a factor of 0.8 to reduce them to more reasonable values. Although the opportunity for low precipitation bias decreases with increasing length of period, significant bias still persists in the 2- and 3-week estimates in the western part of the north-central region during the winter season. Since the probability of measurable precipitation is 1–P(0,T), the P(0,T) bias is carried into the precipitation probabilities, but compensating biases in the gamma-distribution parameter estimates apparently contain most of the bias in the 0.01 to 0.09-inch interval.