| OCR Text |
Show Fairly light wind gusts can deposit (remove) a considerable amount O f snow on (from) the sampling sheet. The height of the sampling box is a compromise, based on experience, between getting too close to the surface and collecting large amounts of blowing snow, and getting above the sur f ace and getting into a serious collection problem, i. e. , having snow blown past or off of the collection sheet. All other errors in the system are small. A second, occasionally significant source of error is that of time coordination between the two samplings. During high rates, a small time discrepancy in the times of the samples being compared can introduce considerable error. This source of error, though, is usually small compared to the win! caused error. For the purpose of calibration the mass of snow collected on a plastic sheet has been converted to thousandths of an inch of precipitation. Likewise, the integral of the output signal of the optical snow rate sensor has been converted to the same units. This has been done using a linear conversion between the sensor output and precipitation rate derived from earlier indepena ent calibration data. r r where si 5t' goo( ll•mi niqU M for ere] orr trat poiI 2.1 FOl tior scr doc me Cr: 0. 3d = rrhu: 1 is precipitation rate - in/hr Kr1 2.: d is recorder deflection - milliamperes The correlation coefficient between over 400 collected short period samples and optical sensor indications, spanning a very wide range of snowfall conditions, is O. 80 (Stinson 1968). Table 4 gives the correlation coefficients for these data, stratified by temperature and wind speed. Ro of la1 19 pu CE TABLE 4 SI Correlation Coefficients Between Optical Sensor Indications and Collected Snow Totals Wind Speed (MPH) Correlation Coefficient 2 0 o. 93 4 0. 91 0. 87 10 8 6 o. 96 8. 82 12 o. 85 14 0. 53 16 0. 72 bi d 18 0. 64 0. 7~ n- d1 Temp. °F Corr. Coef. 5 10 0.10 15 0.82 20 0.70 32 25 0.74 30 0.88 35 0.9 4 0.27 tt a b: |
