3.16. The obvious difference between the assumed cloud models, which describe the approach of a linear cloud front, and true base surge, which probably approaches as a segment ofa circle of finite radius, is negligible for the large surge dimensions observed. As stated above, only a rough general agreement exists between the visual and radiological approach velocities (Tables 3.15 and 3.16). The most favorable comparison is obtained for the visual approach velocity determined at minus 500 feet. Lack of better agreement is probably due to variations in the generation and behavior of different segments of total base surge and to uncertain knowledge of local surface winds. Further difficulties are caused by double spikes (Appendix F) in the first major peak and by obvious changes in the slope of the dose rate curve, which are associated with base surge emergence at early times (Section 3.2). R-of-r velocities determined for rises from 1, 5, and 10 percent to 80 and 100 percent of peak in an attempt to circumvent these difficulties do not yield any significant improvement. The 5- to 100-percent determinations are simply presented as representative. Thus, although the comparison between visual and radiological approach velocities gives somewhat better agreement for the thicker 60° cloud models, distinctions made on this basis may be entirely fictitious. The rough general agreement between the visual and radiological approach velocities does suggest that the visible surge and the radioactive aerosol are moved by the same mechanical forces but does not necessarily imply that they are the same body of airborne material. The velocities tabulated in Table 3.16 represent a best estimate of the speed with which the major radiating source approached the detector. They are, therefore, the vector sum of the velocity due to radial expansion and the surface wind velocity. An approximate value for the radial velocity may be deduced from the approach velocity by assuming that movement of the photographic surge center X (Appendix F) actually represents local variations in surface winds. The tnstantaneous radial component of the local wind at the rad-TOA for each station can then be estimated and is presented in Table 3.17. This radial component is small for rad-TOA’s less than 1 minute, because the surge requires about that amount of time to accelerate to surface wind speed. The appropriate approach velocities corrected for the wind component represent radial velocities due to expansion and are also presented in Table 3.17. The negative velocities obtained at DR 9.0 for Wahoo may reflect possible ship retardation (Section 3.3.2) whereas those obtained at the more distant stations for Umbrella probably indicate that local wind variations based on movements of the photographic surge center X do not necessarily correspond to those existing at the surge periphery. For both shots, records from the Eniwetok weather station show enough variation in both surface wind speed and direction to cause errors in the computed radial components as large as -5 knots. These approximate radial velocities are plotted for Wahoo and Umbrella in Fig- ures 3.132 and 3.133. Because expansion of the base surge into an opposing wind would tend to increase the angle of the front, radial velocities derived from the 90° cloud model are used for the upwind stations. Velocities derived from the 60° model are used in all other directions. Cloud thicknesses of 1,600 feet and 1,200 feet are used for Wahoo and Umbrella, respectively. The scatter in the radial velocity data is partially due to uncertainties in the basic assumptions underlying the calculations but may also be due to actual differences in the initial velocity of expansion along specific radii. Furthermore, local vertical surge development caused by the atoll reef could be reflected as an apparent increase in radiologically determined approach velocity. The high approach velocity reported for Umbrella Station DR 18.6 may represent such a case. Local vertical development over the reef could increase the radiating area without greatly changing horizontal motion. Such action would result in an apparent increase in approach velocity. These approximate radial velocities may be compared with the fluid models of References 97 through 100. In all investigations, fluid columns of a uniform density greater than that of the ambient fluid were released from rest and their collapse studied photographically. A simi- lar collapse has been suggested as the primary mechanism for the formation of base surge. Unfortunately, Reference 97 1s for a solid column, and insufficient information is available in the published work to make an exact conversion to Wahoo and Umbreila conditions. The data 234