Selection of the proper method of computing refer- ence crop evapotranspiration depends on:
• Type, accuracy, and duration of available climatic data.
• Natural pattern of evapotranspiration during the year.
• Intended use of the evapotranspiration esti- mates.
The type, quality, and length of record of climatic data greatly affect the selection of an ETo method. Some irrigation management applications require real-time data while design and water right considerations require an assessment of historical water use patterns. Thus, the length of time that various types of data are available may dictate the type of method to use in estimating ETo. In many locations air temperature has been recorded for long periods. Wind speed, relative humidity, and solar radiation data are less available and are more difficult to measure, causing these data to be less reliable. Thus, some locations may require use of the temperature based ETo method while at other locations, other methods would be more appro- priate. The available climatic data should be compiled and evaluated before beginning any computation. The usable methods can be identified once data quality has been determined.
The natural pattern of crop water use can affect the selection of an ETo method. Crop evapotranspiration varies from day to day because of fluctuating climatic conditions and plant growth. The variation can be large in some climates. The daily crop evapotranspira- tion can be averaged over a period, such as 5 days.
This provides the average daily crop evapotranspira- tion for that period. The average daily crop evapo- transpiration for each 5-day period of the summer could be computed for a series of years, producing a set of 5-day average daily crop evapotranspiration values. Of course, the 5-day average daily crop evapo- transpiration data will vary among the sets.
An example of the variation of average daily evapo- transpiration for ryegrass for different lengths of the averaging period is given in figure 2–14. The probabil- ity shown in this figure represents the chances of the average daily evapotranspiration being less than a given amount of evapotranspiration. For example, the average daily evapotranspiration for a 5-day period will be less than 0.225 inches per day for 99 percent of the values in the set of 5-day averages. If we assume that the future will resemble the past, we can expect the 5-day average evapotranspiration for ryegrass at this location to be less than 0.225 inches per day 99 percent of the time in the future.
Figure 2–14 Variation of the average daily ETc as affected
by the length of the averaging period (adapted from Doorenbos and Pruitt 1977)
0.25 0.20 0.15 0.10 0.05 0 0 5 10 15 20 25 30 35 99 95 90 75 50 25 10 5 1
Length of averaging period, days
Average daily evapotranspiration of Ryegrass, inches/d
Probability (%) that average daily crop evapotranspiration will exceed a given amount
Averaging dampens out the fluctuation of daily evapo- transpiration data and decreases the range of average daily evapotranspiration. Therefore, the range of values for the 5-day average daily evapotranspiration is smaller than that of the daily values. This is illus- trated in figure 2–14. Consider the 5-day average daily evapotranspiration for the 1 percent and 99 percent probabilities. The 5-day average daily evapotranspira- tion is smaller than 0.225 inches per day 99 percent of the time and smaller than 0.090 inches per day 1 per- cent of the time. Therefore, the 5-day average is be- tween 0.09 and 0.225 inches per day 98 percent of the time. Compare this to when the data are averaged over 10 days. The 10-day average is between 0.10 and 0.215 inches per day 98 percent of the time.
Errors in daily estimates of reference crop evapotrans- piration also tend to balance out when averaging over a period. On some days, the errors associated with either the climatic data or with the prediction method cause ETo estimates to be excessive. On other days the method might underpredict ETo. These errors compen- sate during the period, thus the accuracy of the ETo estimate generally improves with longer computa- tional periods (Jensen and Wright 1978).
The combined processes of less natural variation in average evapotranspiration for long periods and the error compensation within a period for ETo predic- tions cause the magnitude of potential errors in ETo estimates to decrease with the length of the computa- tion period. Thus, less precise ETo methods may provide adequate accuracy for long-term estimates. However, complex equations are required for short- term (daily) estimates.
Studies have shown that the Penman-Monteith method is more reliable for any length period than methods that use less climatic data (Jensen, et al. 1990). The method works well for daily calculations and for estimating monthly or seasonal water needs. If ad- equate data are available or can be estimated, the Penman-Monteith equation should be considered. The radiation method and the temperature (FAO Blaney-Criddle) method are less precise than the Penman-Monteith method. These methods are accept- able for predicting the average daily water use for a period of days. However, they can produce significant errors for an individual day. Thus, these methods are
The evaporation pan method is less reliable for short- term estimates than other ETo methods and is recom- mended for periods of 10 days or longer. Evaporation pans can be accurate if well maintained and properly located. If the pan has a history of proper use, 10-day periods can be used. Poorly maintained pans and inappropriate siting can lead to severely biased data. If little previous history is available for a pan, caution should be exercised even for computing ETo for longer periods.
The purpose for computing ETo may determine the calculation method. Three examples will illustrate the variation in ETo needs. Irrigation scheduling requires local real-time data. Irrigation system design considers a historical record to evaluate the expected maximum capacity for water supply and delivery systems. Reser- voir design or water right determination may only require monthly water use estimates.
Daily ETo estimates are not necessary for some irriga- tion scheduling applications. If a field is irrigated every 10 days, scheduling using the radiation or temperature based method, or an evaporation pan, may produce essentially the same schedule as that using the Pen- man-Monteith method. If high-value, shallow-rooted crops are grown on coarse textured soils, daily ETo estimates may be necessary for accurate scheduling. In such cases the Penman-Monteith method would be best suited.
The selection of an ETo method for designing an irrigation system depends on the required irrigation frequency. If crops are irrigated frequently because of a shallow root zone, coarse textured soils, or mainte- nance of large soil-water depletions, the required water supply rate will be larger than that for infre- quent irrigation. Results in figure 2–14 illustrate this concept. To design a system you might want to meet the average daily evapotranspiration at least 90 per- cent of the time. If you irrigated daily, the ETo for
design would be 0.21 inches per day in figure 2–14. The design ETo drops to 0.195 inches per day for a 5- day irrigation frequency and to 0.190 inches per day for a 10-day period (fig. 2–14). The Penman-Monteith method is needed to adequately design for the daily irrigation frequency. Either the radiation method, the temperature method, or the Penman-Monteith method will suffice for the 5- or 10-day irrigation frequency. For design, climatic data must be available for a num- ber of years to develop the probabilities as shown in figure 2–14. A less precise ETo method with a longer history may be preferable to a precise method where a limited length of climatic data are available.
To design and operate a reservoir, or to establish water rights, the short-term estimate of ETo is less valuable than the monthly or annual water use pattern. Often these uses require consideration of several crops and numerous fields where exact information is not available for each parcel. Thus, average ETo values for biweekly, monthly, or annual periods may be adequate. For these applications, all the ETo methods are acceptable, and the quality of the available climatic or evaporation pan data may be the deciding factor.