Evaporation from Pretty Lake has been computed for a 2%- year period between 1963 and 1965 by the use of an energy budget, mass-transfer parameters, a water budget, a class-A pan, and a computed pan evaporation technique. The seasonal totals for the different methods are within 8 percent of their mean and are within 11 percent of the rate of 79 centimeters (31 inches) per year determined from published maps that are based on evaporation-pan data. Period-by-period differences among the methods are larger than the annual differences, but there is a general agreement among the evaporation hydrographs produced by the different computation methods.

The energy budget was an excellent means of computing unbiased evaporation data for periods of a month or longer from June through September. It is not reliable in the springtime, when Bowen ratios are large and when the large changes in stored energy may be hard to measure accurately owing to errors in the capacity table. The need for sophisticated equipment, frequent temperature surveys, and complex computations makes the energy budget the most expensive of the several methods used. Effective use was made of the Koberg method in estimating long-wave radiation when accurate instrument records were not available. Effects of sediment heating and cooling were computed to have influenced evaporation as much as 0.03 cm day" 1 (centimeters per day) just after the autumnal overturn. The change is significant during the fall, when the evaporation for Pretty Lake is low, and would be more significant in a shallow lake, where the heat storage by the sediment would be large in proportion to the storage by the water.

The corrected fall in stage computed by the water-budget method agreed well with the evaporation rates computed by other methods during the dryer seasons. Decreased rates of fall in stage during the wet seasons indicated net inflow seepage that was estimated to be equivalent to a stage change of more than 0.2 cm day" 1 at some times.

Evaporation data based upon class-A pan records and computed pan evaporation were too large early in the season and too small late in the season. The differences were caused by energy storage, which affected the lake evaporation as energy was stored in the spring and released late in the season. Energy-storage effects can be corrected, but the corrections require some of the same expensive data that were used in the energy budget. The mass-transfer system proved to be an effective low-cost means of computing evaporation, a means that is well suited to low evaporation rates.

The mass-transfer coefficient was determined to be 0.00560 cm hr day- 1 mile" 1 mb" 1 (centimeter per day per millibar per mile/hour), the relative standard error of the energy-budget calibration being about 6 percent. Springtime and autumn evaporation rates computed by the mass-transfer method were slightly higher than rates computed by other methods, and summer rates from mass-transfer computations were slightly lower than rates computed by other methods. Anemometer stalling is believed to have caused unreliable mass-transfer evaporation data during two periods having very low wind velocities.

Assuming that Pretty Lake is typical of the many small natural lakes in its region, it is concluded that in most cases the evaporation information needed for hydrologic studies can be provided with satisfactory accuracy by a combination of the mass-transfer method and one or two other methods, without the expense of a complex energy-budget study.

The different methods, although poor, agree that evaporation when there is ice cover is generally small (less than 0.1 cm day" 1 ), but the evaporation rates during the few days just before freezeup or just after ice breakup are significant