Having obtained a water balance for the 1962-1988 time period, one would like to use the 1931-1962 data tabulated by Phillips (1968) to model water levels as did Redmond (1990). The earlier data are of significantly lower quality as water levels were only measured sporadically, at varying locations, and with varying reference levels. In addition, there are no precipitation data in the Park during World War II. The importance of the data from the North Rim gage for obtaining a good match during the 1962-1988 time period shows that the match just using data from the Park Headquarters gage will not be very good, as Redmond (1990) found. This problem does not invalidate Redmond’s conclusion that the first order dependence of water level can be explained by the variation in precipitation with time. Instead of calculating water level, a more appropriate comparison is to use the earlier data to calculate the water balance parameters and then compare them to see if the difference is statistically significant. Several years during the 1931-1962 time period have no water-level measurements, and the water balance parameters have been calculated in two ways. In the first, only water-level changes for years with data are included. In the second, the precipitation and water-level change are averaged over the multi-year period with missing water-level data to generate values for intervening years. The results are given Table 4. The slope B is very similar for the two calculations and to the value obtained for the 1962-1988 period. The precipitation po is 167 cm/y for the first method and 164 cm/y for the second method compared to 169 cm/y for the 1962-1988 time period. The differences are not statistically significant indicating that there has been no change in the water balance between the two time periods. The 2 and 5 cm/y differences in precipitation po may simply reflect measurement uncertainty although it seems likely that differences in measurement technique between the two time periods may bring some bias to the earlier measurements.
| Table 1. Parameter values for yearly water balances for Crater Lake, Oregon. Precipitation po is value at Park Headquarters that will result in no change in level for Crater Lake, coefficient B is ratio of level change to precipitation greater than p0, supply is total water supply to the lake divided by the area of lake. Water loss from the lake is made up of leakage and evaporation, both expressed as volume divided by total lake area. |
 |
| Table 2. Parameter values for yearly water balances for Crater Lake, Oregon, obtained by various correlations of yearly change in water level with precipitation data for the 1962-1988 time period. Uncertainties are standard errors. P-values are from t test (two tailed). |
 |
