528 SAUCIER, HALL, AND NELSON humidity during the spring season, Weather radars are operated at the Oklahoma City airport, Tinker Air Force Base southeast of Okla- homa City, NSSL in Norman, Wichita Falls, and Fort Worth. At most of these, photographic scope records are made for later research use. An M-33 radar system at Norman is also employed by this project for current storm surveillance and for obtaining wind soundings in the thunderstorm surroundings. During the 1963 storm season, we maintained rainfall collectors situated over the northwest one-third of the network. These were pans of 25 sq ft area which emptied into 10-liter plastic jugs. With this surface area of the pan, a rainfall of 0.15 in. gave a 10-liter sample, which was believed to be more than sufficient for the radiochemical analyses planned. Under each pan, a dry filter paper, which was exposed to the air but sheltered from the rain, was placed horizontally to collect a sample of the dry fallout. This was to serve as an index to the dry fallout collected by the pan which would contaminate the sub- sequent rain sample. Each time the pan was washed, by either hand or rainfall, the filter paper was changed. The radioactivity of the filter paper was found to be below threshold in each case. Therefore the procedure for observing dry fallout was discontinued in 1964. Analysis of the 1963 data revealed that the radioactivity, as well as the rainfall, varied considerably from one collector to the next, even though the sample collected by any pan was a composite of the entire storm over that location. This was attributed to the existence of significant horizontal gradients in radioactivity within the storm produced by variations in the storm circulation of dimensions equal to or smaller than the collector network. Also, propagation of the storm would pro- duce local time variations in the amount of radioactivity collected. To obtain meaningful results required a sampling technique that would measure the time variations at each station. These time variations measured at each in a network of collectors, compared with the in- tensive meteorological records for the storms, would lead to a more complete picture of the circulation, propagation, and development of the storm. The rain-sample collector pans were modified so that each would obtain as many as ten 1-gal sequential samples of rainfall per storm. A 1-gal sample represents about 0.05 in. of rain. The tenth sample at each collector integrated all rainfall from the storm beyond about 0.50 in. Some loss occurred between samples due to strong winds. The automatic sample collector was designed by S. J. Hall from the pattern devised earlier by the Illinois State Water Survey group. Figure 4 is a photograph of the original collector pan on the left and of the automatic sequential sample collector on the right. As each jug in sequence is filled to the desired level, it drops by its weight, and the next jug is driven into position by torque on the axial drum sup- plied by a pulley and weight system. The weights are common build-

Select target paragraph3