The atmometers underestimated daily ETr by an average of 21% across the state. The average root-mean-square-error (RMSE) of daily ETg from all stations was 1.14 mm. Correlation between ETg and ETr was highest in the Piedmont and lowest in the Coastal stations. Data analyses indicated that on rain-free days the RMSE (0.96 mm) was significantly smaller than rainy days (1.45 mm) and most of the severe underestimations occurred on rainy days. Hourly time-series revealed that evaporation ceased during and following the rainfall, which suggested rainwater accumulation on the evaporating surface of the atmometer despite its convex shape. The amount and duration of rainfall were shown to affect the amount of reduction in ETg. In all cases the interception was significantly different than zero, which suggested the need for linear regression models to adjust ETg to ETr instead of single factors.
Regression equations to convert daily ETg to ETr were developed for each of the three regions (i.e. the sandy-loam Coastal Plain, the clay-soils inland, and the sandy-soils Sandhills) and with different rainfall amounts. The corrected ETg data resulted in an average RMSE of under 0.80 mm across the state in any conditions. Using the regression equations the average RMSE of ETg were reduced by 19%, 45%, and 49% on rain-free days, days with less than 10 mm of rainfall and days with at least 10 mm of rainfall, respectively.
The ETg - ETr relation was also compared in different climate conditions was found to be most sensitive to wind speed and then to humidity and radiation and was generally insensitive to temperature. The modified index of agreement declined sharply from 0.75 in the lightest-wind class to 0.29 in the strongest-wind class; and the average ratio of ETg/ETr decreased from near 0.9 to 0.36In the strongest-wind conditions the RMSE was greater than 2 mm d-1. ETg was closer to ETr in relatively dry and strong-radiation conditions. Temperature had little influence on the ETg - ETr relation and even in the highest quartile of temperature, the average ratio of ETg/ETr was still small than 1. This suggested that the problem of raised evaporation due to heat storage in the container was observed in this type of atmometer.
This study showed that this type of atmometer tended to underestimate ETr, but with proper regression equations developed from local data it could be used to estimate reference ET where meteorological data is scarce to apply the Penman-Monteith equation. Aggregation of the ETg measurements in the weekly or monthly interval would reduce the percent estimation errors.