was decided. The results of these first two processes are shown in Table 2. 

 Examining Table 2, it can be seen that more events were selected for the lower 

 range of water levels («8 ft NGVD) than were selected for the higher water 

 levels (above 10.5 ft NGVD). This was done for two reasons. First, the prob- 

 ability mass representing the lower part of the NED curve will be much larger 

 than the probability mass representing the higher portion of the curve. This 

 is caused by the logarithmic nature of the frequency of water levels. 

 Experience has shown that frequency curves are more easily constructed when 

 the probability masses assigned to simulated events vary as little as possi- 

 ble. For example, the probability mass per year associated with a 0.1-ft in- 

 crement located at 8.0 ft on the NED curve is 0.14; whereas, the probability 

 mass per year for a 0.2-ft increment at 11.2 ft is 0.00035. Therefore, more 

 events were selected at the lower return periods to divide this large prob- 

 ability mass into smaller segments. Secondly, especially when considering 

 overtopping, events formed from many more combinations are possible at the 

 lower stages (large surge plus low tide plus medium waves, small surge plus 

 medium tide plus large waves, etc). At the higher stages fewer combinations 

 are possible (large surge plus large tide plus large waves). Consequently, 

 the higher end of the curve can be represented by fewer events than can the 

 lower end. 



20. Choosing the stage increment sizes and the number of events 

 selected for simulation from each increment is a subjective decision. This 

 decision is based on the range of stages to be represented, the largest 

 differences in probability tolerable for accurate curve generation, and the 

 financial constraints on the number of events that can be simulated. Unfor- 

 tunately, the only sure way to determine if the decisions are correct is to 

 view the results. Therefore selections are made, and the goodness of these 

 decisions is reflected in the error bands presented in Part VIII. 



21. The third part of the selection process is the actual selection of 

 events. The 13,500 possible events, created by combining the storm surge with 

 tide, were ranked by the maximum water level that occurred during the surge 

 plus tide time-history. At each of the stage increments shown in Table 2, 

 events were randomly selected from the portion of the 13,500 events with max- 

 imum water level equal to that height increment. This was done independently 

 for each of the three sets of 50 events. Although these maximum water levels 

 are for the Boston N0S tide gage, events selected for simulation in the study 



17 



