.

The objective of this study is to validate diurnal ET estimates by some models using the in-situ measurements obtained from
the locally constructed lysimeter evapotranspiration model A classical evapotranspirometer (Lysimeter) for
evapotranspiration (ET) measurement was designed, constructed and installed at University observatory at the Federal
University of Technology, Akure (Longitude 5o 18’E, Latitude 7o 8’N). This was done so as to have the required input
parameters for the evaluation of the models from the observatory where the constructed lysimeter was kept. The lysimeter
was kept running for six months on diurnal timescale. In addition, estimation of ET was carried out by the evaluation of some
models for the same period. The methods used include: Advection-Aridity (AA), Priestley–Taylor, Thornthwaite, Makkink,
Granger and Gray (GG) models, and Penman combination. The ET data obtained from the Lysimeter were used to validate
the results of the evaluated models for the same station. The simulated results were compared with the ET data from the insitu
measurement obtained from the locally constructed lysimeter. The coefficient of determination for all the models were
significantly high (0.61-0.98) except Thornthwaite that had the lowest R2 (0.04). The level of accuracy was determined by
calculating the root mean square error (RMSE); the lowest RMSE was found for Priestly-Taylor (4.10) while the rest ranged
from 22.13 for Makkink to 61.13 for GG method. The mean values of ET estimates ranged from 138.21mm/day to 189.19 for
Penman Combination; Priestly Taylor had 138.21mm/day, which was the nearest mean value to the in-situ value found to be
137.14mm/day. The lysimeter values were underestimated consistently by AA method by 19.25% while it was largely
overestimated by Penman Combination and GG methods by 75% and 45% respectively. From the above, it was found that
Taylor has the highest determination coefficient of 0.98 and the least error of 4.10, hence, it can be concluded that the ET
values by the Lysimeter can be well predicted by the Priestley-Taylor model with good accuracy, taking into consideration
the lysimeter sources of error.
Key words: Validation, Potential evapotranspiration, lysimeter, models, construction.