The behaviour of various formulas for evapotranspiration of grass in nonres
tricted soil water conditions is considered. These are the expressions base
d on the Penman formula, i.e. "old" Penman, Penman-Monteith, Thom-Oliver an
d the Version recommended more recently by the FAO. Moreover, the Priestley
-Taylor and the Makkink formulas are considered, which are radiation-based.
Comparisons are made between daily mean values estimated with these formul
as and direct measurements. The latter were collected over grass in the per
iod 1979-1982 in the catchment area of the Hupselse Beek (The Netherlands).
It was found that if all required input data were measured, the Priestley-
Taylor and the "old" Penman formula yielded the best results. The assumptio
n that soil heat flux can be neglected introduces a systematic and a random
error of roughly 5%. The empirical estimates for net radiation from sunshi
ne duration, temperature and humidity appear to perform rather poorly. Thes
e estimates improved significantly if solar radiation was measured directly
. The empirical expression proposed by Slob (unpublished) that requires inc
oming solar radiation only as input, provided better results than the other
more complicated expressions. Moreover, this study reveals that evaporatio
n of unstressed grass is primarily determined by the available energy, i.e.
good evaporation estimates can be obtained by using simply lambda E = 0.86
(R-n - G). The Makkink method appears to be attractive for practical applic
ations. These findings support the use of Makkink's formula for routine cal
culations of crop-reference evapotranspiration as has been done by the Roya
l Netherlands Meteorological Institute since 1987.