3. Prediction of Bluff Recession . 



As discussed previously, Seibel (1972) determined a linear relationship 

 between average lake level and bluff recession. His bluff recession rate meas- 

 urements (Table 11) correlated well with the average lake level for each period 

 (correlation coefficient of 0.73), explaining about 50 percent of the variance. 

 Seibel also considered the average number of storms in each period which was 

 fairly constant and did not correlate well. 



Since this study includes more detailed measurements over shorter time 

 periods, these relationships were reexamined using simple and multivariant 

 regression analysis. With bluff recession rate, B, as the dependent vari- 

 able, linear relationships with the following independent variables for each 

 period were examined: 



AL = average of daily lake levels during each period. 



RL = average rate of lake level change computed from monthly average 

 lake levels. 



HL = average of the highest 1/4 daily lake levels. 



W = percent of time that winds were onshore (220° < < 20°) and 



greater than 26 kilometers per hour as measured at the Muskegon, 

 Michigan, airport (the nearest weather station 137 kilometers to 

 the north) . 



The selection of some variables was arbitrary and the results may have possibly 

 been improved, for example, hy increasing or decreasing the cutoff windspeed. 

 However, this was not done since the intent was only to identify the important 

 variables, not to develop the best possible prediction equation. 



The data were refined by assuming that all bluff recession occurred during 

 the ice-free periods. Therefore, storms and lake levels during the ice-covered 

 periods were not considered and the value of each variable was computed based 

 on the length of the ice-free periods using the estimated periods of ice cover 

 given in Section 11,5. (Note: all rates of bluff recession discussed previ- 

 ously have considered the entire period regardless of ice cover.) Only data 

 from reach A were considered because of its lack of major structures and the 

 fact that eight recession rate measurements were made in the reach. These 

 data in final form are given in Table 12. 



The reduction in period length due to ice cover caused a significant in- 

 crease in the winter recession rates which accentuated the already high winter 

 values compared to the lower summer rates. This is the inverse of the lake 

 levels which are high in the summer and low in the winter. 



Figure 35 shows the relationship between the variables, including the 

 actual variations in the mean monthly lake level, the mean monthly rate of 

 lake level change, and the number of days per month that onshore windspeeds 



54 



