Agricultural and Forest Meteorology, 64 (1993) 237-256 237

Elsevier Science Publishers B.V., A m s t e r d a m

Data-sparse estimat{on of lake evaporation, using a

simplified Penman equation

Edward T. Linacre
School of Resource and Environmental Management, Australian National University, Canberra,
A.C.T. 2601, Australia
(Received 20 July 1992; revision accepted 10 November 1992)

ABSTRACT

Linacre, E.T., 1993. Data-sparse estimation of lake evaporation, using a simplified Penman equation.
Agric. For. Meteorol., 64: 237-256.

At m a n y places the only reliable climate measurements available are those of daily m i n i m u m screen
temperature, daily m a x i m u m , rainfall and windspeed. However, even the temperatures alone are sufficient
for estimating monthly mean lake evaporation E0, using the following simplified version of Penman's
formula:
E 0 = ( 0 . 0 1 5 + 0 . 0 0 0 4 2 T + 10-6z)[0.8Rs - 4 0 + 2 . 5 F u ( T - Td) ] (mm day 1),
where T is the daily mean temperature (i.e. the average of the extremes), z is the elevation (m), R~ is the
solar irradiance of the lake's surface, F s t a n d s for (1.0 - 8.7 x 10 5 z), u is the windspeed at 2 m, and Td
is the dewpoint temperature. Two new methods of estimating R s are described, the more accurate being
based on rainfall data, as proxy for the cloud which reduces extraterrestrial radiation. Td can be estimated
from the daily extreme temperatures. The windspeed can be gauged from measurements nearby or at other
times, since accuracy is not important. Errors of estimating monthly mean evaporation are consequently
around 0.3 m m d a y - ~ at Copenhagen. For annual evaporation at three lakes in the USA, estimates differ
from measurements by 0.1, 0.4 and 0.1 m m day -~ , respectively, but the errors of the measurements are
uncertain. Estimates for a reservoir in Western Australia are closer to measurements by the Bowen-ratio
method, than are water-balance measurements. On average, monthly estimates and Bowen-ratio measure-
ments there differ by about 0.8 m m day -~ , whilst averages over 2 years differ by 0.3 m m day ~.
The annual range of monthly mean temperatures at a place is proportional to the latitude and to d °2,
where d (km) is the distance downwind of the o c e a n .

INTRODUCTION

This paper is about the application of the evaporation formula of Penman

(1948) in places where the data available are insufficient for its direct use. In
such circumstances, it is common to take the short-cut of using some empirical

Correspondence to: E.T. Linacre, School o f R e s o u r c e a n d E n v i r o n m e n t a l M a n a g e m e n t , A u s t r a l i a n

N a t i o n a l University, C a n b e r r a , A.C.T., Australia.

0168-1923/93/$06.00 © 1993 Elsevier Science Publishers B.V. All rights reserved.

238 E.T. LINACRE

connection between evaporation rate and temperature, for instance. However,

that is a dead-end, of uncertain relevance elsewhere (Linacre, 1963). On the
other hand, Penman's formula derives from the physics of the process, so that
it is universally applicable. It has proved relatively successful in estimating the
water loss from natural surfaces in several places (e.g. Brutsaert, 1965; Barry,
1969; Hylckama, 1974; Jensen, 1975; Yao, 1981; Young, 1987).
The formula is as follows, in terms of the energy required to evaporate
water at a rate E0p (kg m -2 s -t)
2gop = (sR n +pcO/ra)/(s+7) (Wm 2) (1)
where 2 is the latent heat of evaporation of water (J kg ~), s is the change of
saturation water-vapour pressure with temperature (hPa K -t) between the
dry-bulb and dewpoint temperatures, Rn is the net irradiance of the water
surface (W m-2), p is the density of air (kg m -3) at the ambient pressure and
temperature T (°C), c is the air's specific heat (about 1.0 kJ kg-l K J), D is
the saturation deficit of the air (hPa), r a is the diffusion resistance between the
water's surface and the atmosphere at about screen height (s m-t), and 7 is
the 'psychrometric constant', discussed later.
It is often convenient to replace 2E0p by E0p , which is the equivalent daily
lowering of the water level (mm day- 1). A rate of 1 mm day- 1 represents an
energy flux (2E0p) of about 28.4 W m 2.
Four points can be made about the equation. First, the net irradiance R,
should be reduced by the rate of heating of the water G. However, daytime
heating is often almost balanced by night-time cooling, so that G is usually
ignored for periods of more than a day or two. As a result, Penman (1956)
recommended that his formula (without G) be used only for periods of 1 week
or more, though it has been found satisfactory even for 24 h periods (Tanner
and Pelton, 1960; Van Bavel, 1966).
Secondly, use of Penman's formula for periods longer than 24 h overlooks
the fact that practically all the evaporation occurs during the daytime, when
temperatures are higher than the 24 h mean customarily used in the equation.
Some 64-92% of the evaporation in Australia's Northern Territory takes
place between 06:30 and 18:30 h (Chapman and Kininmonth, 1972). Indeed,
the reverse of evaporation, dewfall, may occur at night. So estimates of
evaporation based on daily mean temperatures are underestimates. This
applies especially if the daily range is large, which occurs in cloudless (i.e. arid)
conditions, where the dewpoint is relatively low (Linacre, 1992). If one
assumes daylight from 06:00 until 18:00 h, a minimum temperature T,, at
06:00 h, a maximum Tx at 15:00 h, and linear changes between them, then it
is easy to derive the daytime mean temperature Ta graphically, as follows
T~ = (0.6T x +0.4Tn) (2)
Thus, Td exceeds the 24 h mean by about a tenth of the daily range Tr (i.e. a
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 239

tenth of (Tx - Tn)). The range can be derived from the mean and dewpoint
temperatures, using a diagram given by Linacre (1992).
Thirdly, a further slight underestimation is due to the product u T (where
u is the daily mean windspeed), shown later to be implied in the term D r a 1
in eqn. (1). The reason is that the daily swing of temperature occurs in parallel
with the day-to-night variation of windspeed, and so the 'product of the
averages' of wind and temperature is less than the 'average of the products'
that is needed in eqn. (1) (Mahrt and Ek, 1984; Linacre, 1992). Such parallel-
ism occurs during 24 h, as higher daytime temperatures coincide with the
stronger winds of daytime. Linacre (1992) illustrated cases where daytime
winds average about 4/3 of the daily mean u, whilst nocturnal winds are about
2/3 of it, so there are simultaneous day-to-night fluctuations by 0.67u.
Likewise, it is observed that the daytime mean temperature differs from the
nocturnal mean by roughly 0.2 R d, i.e. a fifth of the difference between the
daily extremes. It is readily shown (Linacre, 1992) that the underestimate is a
quarter of the product of the fluctuations, i.e. uRd/30. However, this is
insignificant.
Fourthly, Penman omitted the effect of atmospheric stability, which alters
the importance of the advective term (Thom and Oliver, 1977). On the one
hand, instability enhances the evaporation at low windspeeds, in clear daytime
conditions at low latitudes, whilst, on the other, stability on a clear night
reduces evaporation. Fortunately, the complication is normally unimportant,
as regards evaporation over 24 h or more (Mahrt and Ek, 1984). Most
evaporation occurs in the daytime, when the wind is usually great enough for
neutral stability, i.e. it is over 2 m s-~ (E.K. Webb, personal communication,
1985).
Penman's equation (eqn. (1)) shows that the evaporation rate depends
mainly on two factors: the net irradiance and the humidity (Meyer et al.,
1989). Errors in determining these produce about four times more difference
to the evaporation estimate than similar errors in the temperature or wind
data (Venkataraman and Krishnamurthy, 1973; Meyer et al., 1989). For
instance, a 30% error in the wind value creates an evaporation error of only
5% (Woodhead, 1972). So the windspeed need not be known accurately.

ACCURACY OF THE PENMAN FORMULA

Table 1 shows the accuracy of the Penman formula in practice, for periods
of 1 month or 1 year. The median ratio Eo/Eop is 0.99, and the standard
deviation is 0.12, i.e. the probable error is about 8% (Linacre, 1992). There
would be larger errors for shorter periods (Linacre, 1992).
The scatter of ratios in Table 1 is due partly to the various aspects of the
Penman equation just considered, to the use of average data (instead of hourly
measurements, for instance, which are rarely available) and to a lag in the
240 E.T. LINACRE

TABLE I

Comparison of measurements of the evaporation from water E0, with estimates from the formula of
Penman (1948), i.e. Eop, over a month or year

Place Eo/Eop Reference

Lake Hefner 0.97 Penman (1956)

12 m 2 tank in Copenhagen 0.98 in April, Aslyng (1960)
1.19 in July,
1.28 in September
Holland 1.0 Rijtema (1966)
Dutch lake 0.99" Keijman and Koopmans (1973)
Pune, India 1.06 Venkataraman and Krishnamurthy (1973)
Three USSR reservoirs 1.19, 0.93 and 0.94 b Vikutina (1973)

"The average of seven monthly values ranging from 0.83 to 1.18.

b For periods of several months. Over single months, 15 values of the ratio had a mean of 0.85 and standard
deviation of 0.37.

change of lake temperatures (Hounam, 1973). These are also errors involved
in measuring lake evaporation. Spot measurements beside a lake may not
be representative of the whole lake, in view of variations of windspeed,
irradiance and temperature across the surface (Hoy and Stephens, 1977),
depending on the direction of the wind. As a result, calculations based on
upwind measurements would overestimate the evaporation (Morton, 1986).
As regards water-balance estimates of lake evaporation (subtracting other
outputs from the total input), they are vulnerable to errors owing to unknown
seepage.

THE TRANSFER COEFFICIENT INVOLVED IN LAKE EVAPORATION

The value of ra in eqn. (1) can be derived in three ways (Brutsaert, 1982).
Here we review numerous published values, using the terminology of the rest
of the paper.
First, from the 'Dalton equation' for the evaporation rate (Humphreys,
1964; Webb, 1966; Linacre and Hobbs, 1977)
Eo = ku(ew-e) (3)
Hence
2E 0 = k ' u ( e w - e ) (4)
where ew and e are expressed in hectopascals (ew for the saturation vapour
pressure at the temperature of the water surface, and e for the water-vapour
pressure of the atmosphere), and E 0 is in units of kg m 2 s -~. The factors k and
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 241

k' are empirical, and ( k ' u ) constitutes a 'latent-heat transfer-coefficient'. The

dimensions o f the other parts of the equation (Linacre, 1992) mean that k' in
eqn. (4) has the advantage of being dimensionless, unlike k in eqn. (3).
Numerically, k' equals 28.4 k at 20°C, and it is related to r a in eqn. (1) by the
following alternative definitions of 7, the 'psychrometric constant' (HMSO,
1956)
7 = h/(k'u) = (pc/ra)/(k'u) (5)
where h is the 'sensible-heat transfer-coefficient' (Linacre, 1964), and
= pc/2E (6)
where p, c and 2 are as in eqn. (1), p is the atmospheric pressure (hPa) and E
is the ratio of the respective densities of air and water vapour (i.e. 0.622). The
dependence of 7 on p and 2 (i.e. on temperature) means that it decreases with
elevation z (m), equalling (0.67 - 7.2 × 10 -5 z) hPa K -1
The combination of eqns. (5) and (6) gives
ur, = p 2 ~ / k ' p (7)
Now, the latent heat o f evaporation 2 is 2500 kJ kg-1 at 0°C, 2450 at 20°C,
and 2410 at 40°C. Also, the density of sea-level air p is 1.28 kg m -3 at 0°C,
1.19 at 20°C, and 1.11 kg m -3 at 40°C. The density decreases in proportion
to the pressure p, so that the product ur a in eqn. (7) is unaffected by elevation.
It has the following value at 20°C
ur, = 1.19 × 2450 × 103 × 0.622/28.4k× 1000 = 6 4 / k (8)
The expression is 70/k at 0°C, 67[k at 10°C and 59/k at 40°C.
Values of ur a for water surfaces are given in Table 2. The average is 400,
i.e. ra is 400 u -l .
A second way of deriving r a involves the profile o f windspeeds over the
surface, using the equation below for a flat surface (Monteith, 1963; Grace,
1981). It assumes an atmosphere o f neutral stability, i.e. strong wind and
modest net irradiance.
ur, = l n [ ( z - cl)/zo]2 / k 2 (9)
where d is the 'zero-plane displacement', k is Von Karman's constant (equal
to 0.4), z 0 is a measure of the surface roughness, and u is the windspeed at a
height of z metres. F o r a still water surface, d is zero, z0 is about 0.0003 m
(Linacre, 1992), and the wind is ordinarily measured at 2 m. With these values,
ur~ equals 480.
A third procedure for deriving r, involves a dimensionless 'bulk transfer
coefficient' denoted by Ch, called the 'Stanton number' and defined as follows
Ch = h / p c u = 1/ur, (10)
242 E.T. LINACRE

TABLE 2

Values of the Dalton coefficient k, and the equivalent product u r a, where G is the aerodynamic resistance
over the water surface, assuming that u r~ equals 67/k, i.e. a temperature of 10°C

Place Period of k u ra Reference

averaging ( m m s i day i h P a - t)

Lake Hefner 3h 0.15 a 450 Marciano and Harbeck (1954)

Lake Hefner 1 day 0.28 240 Kohler and Parmele (1967)
Lakes in general 3h 0.12 560 Slatyer and Mcllroy (1961) b
Large sunken tank - 0.24 280 Rijtema (1966)
South England 1 month 0.23-0.40 210 Szeicz et al. (1969)
California 1 month 0.13-0.16 460 Szeicz et al. (1969)
Lake Tiberias 1 month 0.33 200 Stanhill (1969)
Israeli fishpond 10 days 0.29 c 230 Stanhill (1970)
Perch Lake - 0.19 350 Ferguson and Den Hartog (1975)
Six US lakes Various 400 L a k s h m a n (1972)
Lake Flevo 2-7 days 0.13 520 Keijman and K o o p m a n s (1973)
Small reservoirs 1 month 0.23 d 290 Szumiec (1973)
2 0 m 2 floating pen 1 month 0.10 e 670 Vikulina (1973)
Large pan floating 1 day 0.11 f 610 Jarvinen and Huttula (1982)
in Finnish lake
Tropical Atlantic 0.11 610 Kessler (1985)

~Assuming that the velocity at 2 m is 0.66 times that at 8m.

b After adjusting the 4 m windspeed to the equivalent at 2 m.
This value is derived from the 'a' in (a u + b), equivalent to ku in the Dalton equation. The term 'b' w a s
governed by banks sheltering the pond.
d Ignoring 0.089 in (0.089 + 0.225 u): presumably the ignored term refers to slight additional molecular
diffusion of water vapour, rather than convective diffusion.
~Ignoring 0.14 in (0.14 + 0.10 u).
fIgnoring 0.15 in (0.15 + 0.I08 u).

Thus, values of p and u at sea level give h as about 1200 Ch U, and k (which
equals about 64/ura, see eqn. (8)) is numerically equal to about 64 Ch. NOW
the value of Ch is within 0.86-1.55 × 10 -3 for neutral, unstable conditions
and low windspeeds at 10 m (Hicks, 1972; Thorpe et al., 1973; Hicks et al.,
1974; Friehe and Schmitt, 1976; Liu et al., 1979; Brutsaert, 1982; Large and
Pond, 1982; Katsaros et al., 1987). So a mean value of 1.2 x 10 -3 is represen-
tative with respect to winds at 10 m, and therefore about 1.8 × 10 -3 at 2 m.
Hence u ra (i.e. Ch-1) approximates 560.
The average o f the three estimates ofur, (i.e. 400, 480 and 560) is 480, which
implies a heat-transfer coefficient (i.e. p C/ra) o f 2.5 u W m -2 K -l near sea
level. This resembles the 2.6 u and 2.3 u W m -2 K -1 quoted elsewhere
(Linacre, 1992) and approximates 2.6 u W m-2 K - l , the median o f 26 further
values from the literature (see Appendix).
Such values derive from measurements at sea level. At high elevations, the
thinner air leads to less advective heat transfer. So the appropriate coefficient
244 E.T. LINACRE

similar numerical value, i.e. (0.42 + 0.012 T + 3.10 -5 z), where z is the
elevation in metres. (This expression is like that of Rouse et al., 1977.)
Thirdly, replace R, by an empirical relationship to the solar irradiance R s,
adjusted for the difference between the albedos of water and a grass surface.
The value for Rn to any grass surface is (0.63 Rs - 40) W m -2 (Linacre, 1992),
where the number 0.63 chiefly depends on the albedo of grass, i.e. 0.22. As
water's albedo is only 0.07, i.e. 0.15 less, so the 0.63 is increased by that much
for lake evaporation, to about 0.8. Hence, R n for water is given by (0.8 Rs -
40) W m -2. This expression depends on neither site nor season.
Lastly, replace R s in the expression above. New ways of doing this are
described in the following section, unique in requiring few or no measure-
ments more than the temperature data already assumed available for calculat-
ing other aspects of the evaporation formula. Method A involves equating R~
to the sum of: (1) the annual mean at that latitude Ry; (2) the month's
deviation q from the mean. A second way, Method B, requires rainfall data
in addition. Method B is more accurate than Method A, but may be less
suitable for periods of less than 1 month. It will be shown that either way of
estimating R~ gives reasonable accuracy in estimating monthly evaporation.

ESTIMATION OF THE SOLAR IRRADIANCE RS

Several ways of estimating R~ have been described elsewhere (Linacre,

1992), but the following are new and require few measurements.

Method A: the 'average-plus-deviation' method

The monthly mean irradiance may be calculated from: (1) the annual mean
at that latitude Ry, plus (2) an amount q, derived from the deviation of the
monthly mean temperature T from its annual mean Ty. Ry is given approxim-
ately by the following empirical relationship to the latitude A, the outcome of
a survey of measurements at dozens of places around the world (Linacre,
1969)
Ry ~--- 210+ 1.8A-0.06A 2 (13)
If we assume a dependence of temperature primarily on irradiance (rather
than advection), the monthly deviation q is a fraction of the annual range of
irradiance R~, equal to the monthly deviation of T from Ty as a fraction of the
annual temperature range Tr, thus
q/nr = (T-- T y ) / T r (14)
So that
q = R,(T-- Ty)/T, (Wm -2) (15)
Now, Rr in eqn. (15) is shown in Fig. 1 to be approximately dependent on
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 243

is 2.5 u F W m -2 K -~, where F equals the proportional decrease of density

with elevation z metres, i.e. (1 - 8.7 x 10 -5 z).

SIMPLIFICATIONS OF THE PENMAN FORMULA

Several authors have pointed out that a disadvantage of Penman's formula

is the need for climate data which are not always available (e.g. Christiansen,
1968; Sanderson and Ahmed, 1979; Hargreaves, 1983; Bunting, 1987; Abdulai
et al., 1990). One way round this problem is to estimate the missing informa-
tion, as discussed elsewhere (Linacre, 1992); or Penman's formula can be
adapted for use where input data are scarce.
An earlier attempt led tO the following approximation to Penman's formula
(Linacre, 1977)
E o = [700(T+O.OO6z)/(IOO-A)+ 1 5 ( T - Td)]/(80-- 7) (11)
where A is the latitude (degrees). Equation (11) has had some success in
estimating evaporation (Clemence and Schulze, 1982; Anyadike, 1987). (But
note that evaporation-pan measurements should not be used to test the
equation (e.g. Warnake and Pochop, 1988; Cahoon et al., 1991), since it
applies only to lake evaporation rates.) Disadvantages are that the estimation
of net irradiance implicit in eqn. (11) is crude, and the equation does not allow
for variations of windspeed.
A better version was described recently (Linacre, 1992)
E 0 = (0.0015+4× lO-4T+lO-6z)
× [480(T+ 0.006z)/(84- A) - 40 + 2 . 3 u ( T - Td)] (12)
However, this uses a relationship between annual means of temperature and
solar irradiance R s which might not apply to monthly values, because of a lag
of temperatures on changes of irradiation. Also, eqn. (12) does not allow for
the effect of ocean currents from other latitudes on coastal temperatures,
which can invalidate any assumed temperature-radiation relationship. So it is
necessary to consider further the simplification of Penman's equation.
It can be done in four ways. First, replace D by its exact equivalent
[s(T-Td)], where Td is the dewpoint temperature, and s is defined by this
relationship (i.e. s is the overall slope of the psychrometric curve between Td
and either the 24 h mean dry-bulb temperature T, or, better, the daytime mean
Ta, in the case of evaporation over 24 h or more). The slope s of the chord
between Td and Ta is approximated by the tangent at T, which is equivalent
to (0.5 + 0.01 T + 0.0019 T 2) hPa K -1. Thus, if Td is 15°C, Tis 18°C and
T.d is 19.2°C, for example, the expression gives 1.296 hPa K -1, whereas it is
actually 1.237 hPa K -~ between 15 and 19.2°C (List, 1949): the error is only
5%.
The second simplification is to replace s/(s+ 7) by an expression with a
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 245

300
, , '//"~. '/

/ .. " ' 7 "

" -."'
~200 •

°
i-

~100
/./..
/ .

I I I I I
0 10 20 30 40 50 60
latitude A : degrees
Fig. I. Effect o f the latitude on the annual range R, of monthly mean values of the global irradiance R~.
The data shown by dots are taken from Linacre (1969). The continuous line was drawn by eye, and is
represented by eqn. (16). The dashed line shows the range of extraterrestrial solar radiation, averaged
between hemispheres, which is obviously unaffected by clouds and aerosols, etc. (Linacre and Hobbs,
1977).

the latitude A, thus

Rr = 60+4A (16)
The dashed line in Fig. 1 shows that the range of irradiance above the
Earth's atmosphere also varies linearly with latitude, but by more than is
indicated by eqn. (16). So the latter may differ with unusually clear skies, in
an arid climate.
Equations (15) and (16) lead to
q = (60+4A)(T- Ty)/Tr (17)
where A is the latitude, T is the monthly mean temperature and Ty is the
annual mean.
The annual range Tr can be derived in two ways. If a whole year's monthly
mean temperatures are available, it is the difference between the January and
July mean temperatures. Or it can be calculated from the empirical relation-
ship, shown in Fig. 2
Tr = 0.13Ad °2 (C °) (18)
where d (km) is the distance inland, downwind of an ocean.
A feature of Method A is the assumed dependence of temperature on solar
irradiance, ignoring both the influence of winds and the effect of the time
taken to warm the surface. This lag is of the order of a month or so (Prescott,
1943; Prescott and Collins, 1951; Pelton et al., 1960; Chang, 1971; Chang and
246 E.T. LINACRE

I I I I I

....
6

• • :
5 • • oo • •

"10 .o •

1
/ L I I I I I
°o 10 20 30 40 50 60
latitude : degrees

Fig. 2. Effect of latitude on the dividend of Tr (the difference between January and July mean temperatures)
and d °2, where d (km) is the distance of the place inland, along the streamline of the prevailing wind from
the upwind ocean, using data from Linacre (1969). (Distances were re-estimated for five places.)

Root, 1975; Miller, 1981; Wang, 1986). The consequence is that estimates of
evaporation involving Method A may tend to be slightly too high when
temperatures are falling, and vice versa (e.g. Linacre, 1992). This might be
cured by using the next month's temperature when calculating the deviation
q.
The scatter of values on which the expression for Ry is based, and the
additional scatter seen in Fig. 1, imply that Method A is liable to error, as
shown below. Presumably the dispersion is due to local irregularities of
cloudiness and to the effect of heat in ocean currents from other latitudes on
coastal temperatures. However, such difficulties can be avoided by using the
following quite different method of estimating the monthly mean solar irradi-
ance R s.
Method B: the 'attenuation-of-Ra' method
This requires additional information on either cloudiness or rainfall. It is
based on the 'extraterrestrial solar irradiance' R a of an area above the Earth's
atmosphere and parallel to the ground, tabulated for any particular latitude
and m o n t h (List, 1949; Linacre and Hobbs, 1977). The subsequent attenua-
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 247

tion of R a depends chiefly on the amount of cloud, as reviewed elsewhere

(Linacre, 1992), leading to the following approximate relationship
R S = Ra(0.85-0.047C ) (Win -2) (19)
where C is the average number of tenths of the sky's are occupied by cloud
at the times of observation.
The value of C can be inferred from a new empirical relationship based on
numerous Australian data, involving the monthly precipitation Pm
(Jones,1991)
C = 1 +0.Slog(Pm)+[log(Pm)] 2 (20)
Thus, for example, if R a equals 200 W m -2 and Pm is 30 mm, there are 3.9
tenths of cloud on average (eqn. (20)), and therefore a solar irradiance of 133
W m 2 at ground level (eqn. (19)).
A simpler procedure of the same kind consists of a direct relationship
between 1og(Pm) and the ratio R s / R a. Data over 36 months from Melbourne
(McIlroy and Angus, 1964) yield the following equation
RJga = 0.89-0.21og(Pm) (21)
Since the ratio R s / R a does not in practice exceed 0.75, except at high
elevations (Linacre, 1992), the minimum rainfall to which eqn. (21) applies is
5 m m m o n t h ~. When precipitation is less, the ratio would be taken as 0.75.
However, the extent to which eqn. (21) applies outside Melbourne has not
been explored.
The accuracy of the two methods of estimating monthly R~ can be
compared in terms of the annual average Ry and the annual range Rr, by using
data from six places in Australia. Figures in Table 3 include values of Ry and
R r measured in individual years (Linacre, 1969) in comparison with calculated
long-term average values from the various equations above. The table shows
tolerable agreement in the case of Ry, but wider differences with Rr (involved
in Method A). Some of the latter differences are due to interannual changes
of the measured Ry and R~ (Linacre, 1969). For instance, 3 years' measure-
ments in Melbourne yielded values of Ry with a range of 15 W m 2, whilst
those of Rr differed by 49 W m -2 (McIlroy and Angus, 1964). On the whole,
Method B appears better, provided rainfall data are available. Its mean error
of 13 W m-2 represents about 6 % of the average measured irradiance in Table
3.

NEW FORMULA FOR LAKE EVAPORATION E0

The considerations above afford a new simplification of the Penman

formula for lake evaporation, as follows
Eo = (0.015+0.00042T+ 10-62 ") x [0.8R s --40+2.5Fu(T-- Td)] (22)
248 E.T. LINACRE

TABLE 3

Comparison of measured values (Linacre, 1969) and estimates of the annual mean solar irradiance Ry
(W m 2) and the range Rr (Wm 2) of monthly mean values (i.e. the difference between Rs in the months
with least and most, respectively)

Place Lat. Annual mean Ry Annual range Rr

(os)
Meas. Fig. 1 eqn. (19) eqn. (21) Meas. eqn. (16) eqn. (19) eqn. (21)

Darwin 12 241 234 242 212 129 120 107 110

Townsville 19 239 240 200 212 144 137 155 107
Alice Springs 24 248 238 278 256 130 156 145 126
Griffith 34 228 240 234 211 238 197 229 213
Deniliquin 36 230 195 235 213 283 202 235 220
Aspendale 38 190 186 209 183 234 212 231 210
Mean error 12 13 18 31 18 29

Estimates of Ry and eqn. (16) estimates of Rr were used in Method A. Method B estimates involved estimates
of Rs from eqns. (19) and (20), along with long-term average monthly rainfall figures (Climatic Averages
Australia, published by the Australian Bureau of Meteorology in 1975). Ry is watts per square metre.

This becomes even simpler (i.e. 'reduced') for any particular case, if Method
A is used to determine Rs. For instance, at Aspendale (Melbourne) the
latitude is 38 ° (and therefore Ry is 186 W m -2 from eqn. (13)) and z is
negligible (so F is unity), and Ty is 16.0°C. Hence
Eo = (O.O15+O.O0042T)[12.6T-92 + 2 . 5 u ( T - Td)] (23)
where 12.6 comes from the 0.8 in eqn. (22) and from eqn. (17), i.e. 0.8 (60 +
4 × 38)/13.5, and 92 from (12.6 x 16.0 + 40 - 0.8 x 186).
Equation (22) implies that evaporation decreases with increasing elevation,
because of a lower value of the factor F and a lower temperature T. This
agrees with observations by Longacre and Blaney (1962) and Kessler (1985).

TESTS OF EQUATION (22)

Three American lakes

Equation (22) can be tested by means of the data in Table 4. The equation
gives a better approximation to the measured lake evaporation E0 than the
earlier eqn. (11), so the latter is now superseded. The average error ofeqn. (22)
is about 0.2 m m day - l , or 5% of the mean measurement. The latter is less
than the figure for the original Penman formula (Table 1). However, the
absolute error is more significant in practice, since evaporation data are
usually used in comparison with rainfall figures, by subtraction. This con-
sideration also indicates that there is no virtue in the accuracy of the evapora-
tion figure being much greater than that of rainfall measurement (Linacre,
1992).
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 249

TABLE 4

Comparison of the annual averge measured evaporation rates of three American lakes (Kohler et al., 1955)
with estimates from eqns. (11) and (22)

Place Latitude Altitude wind Mean Dew- Meas. Estim. evap.

A (m) u temp. point evap. (mm day- ~)
(°N) (ms ~) T(°C) Ta(°C) (mmday ~)
eqn. (I1) eqn. (22)

Lake Hefner 36 365 2.6 15.2 7.4 3.7 4.7 3.6

Lake Elsinore 34 384 ~ 1.9 18.4 9.8 4.1 5.7 3.7
Lake Okeechobee 27 4a 2.0 22.0 15.9 4.3 a 5.2 4.2

a From Morton (1986).

Large tank at Copenhagen

When monthly rainfall data are available (even long-term averages for the
respective months will do), one can use Method B for estimating solar radia-
tion. This has been done for climate data from Copenhagen, at the higher
latitude o f 56°N and an elevation of only 28 m, in order to estimate evapora-
tion rates there. Table 5 shows the agreement between monthly estimates and
measurements, the mean error being 0.25 mm day -1 . The annual totals were
508 mm (measured) and 493 mm (estimated), the error being only 0.04 mm
d a y -1 '

R e s e r v o i r near Perth

A third test of eqn. (22) involves measurements o f evaporation from

Mundaring reservoir near Perth in Western Australia (Hoy and Stephens,
1977). Its area is around 6 km 2. Figure 3 shows the variations o f estimated
evaporation rates and o f measurements made in two ways, over periods of
about a month. The Bowen-ratio method o f determining the evaporation rate
involved measuring the net irradiance and the temperature and humidity
conditions at the water surface and at about 4 m above it. The water-balance
method consists of comparing all the inflows o f water into the reservoir with

TABLE 5

Comparison of measured evaporation rates (mm d a y - I ) from a 12 m 2 pan at Copenhagen in 1958 (Aslyng,
1960; Aslyng, 1966), with estimates from eqn. (22), when R~ was derived by Method B, from rainfall data

Evaporation rate (mm day -I)

Jan. Feb. Mar. April May June July Aug. Sep. Oct. Nov. Dec.

Measured 0.0 0.2 0.5 1.5 2.8 2.9 3.9 2. I 1.5 0.8 0.3 0.2
Estimate 0.0 0.2 1.0 1.6 2.2 3.0 3.1 2.4 1.6 0.9 0. I 0.0
250 E.T. LINACRE

10
MUNDARING /~

IA lI |t
.. ,\/
/ " ih
, / 'xl ~" \-_.",~
-o 7-
E
: / ',\ lA I
BR ~/ estimated ,
o 6-
LIJ ,l t / x. ',X ,/I ~\~
A-A-, ~ I X XX ,/II Xt
x,, /;/ ~X , // Xt,
~'~', /t/ ~ ',.\ / // XI,,
~ ~ \ III A',X / II Xt
~.4- X \ \WB il ~ "~ I /I \l
=- \X:';
X x,
I/i/it
\ \ \ ~,k t, i
; / ~
\
1/ \,,,\/! /,'/
',~. ,"/r~ ; "~ IX , ~1 /

i-? "

1971 1972 1973

Fig. 3. Comparison of evaporation rates at Mundaring reservoir (near Perth, Western Australia - - see Hoy
and Stephens, 1977) measured either by the Bowen-ratio method (BR) or from the water balance (WB),
with values estimated from eqn. (22), using Method A for deriving the solar irradiance.

the outflows, including the evaporation. On the other hand, the estimates are
based solely on figures for the mean temperature, windspeed and atmospheric
water-vapour pressure in each period. Method A was used to derive the solar
irradiance.
The overall mean evaporation rate is estimated as 3.59 m m day-~ and was
measured as 3.92 m m day-~ by the Bowen-ratio method and 4.28 m m day-t
by the water-balance method. The first two rates differ by 0.33 m m day ~,
which is comparable with the annual-evaporation errors of eqn. (22) for the
three lakes mentioned in Table 4. If the Bowen-ratio method is assumed the
more accurate measurement (since possible varying seepage is an unknown
factor in the water balance), errors of the estimates for each approximately
monthly period averaged 0.83 m m day -I , and for the water-balance measure-
ments they averaged 1.02 m m day- ~. So, the average error of the estimates is
less than the difference between measurements.
The chief sources of error in the estimates are the changes of heat stored
within the water, and of heat advected in streams, rain and outflow, which
were allowed for in the Bowen-ratio method of measurement.
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 251

DISCUSSION

Equation (22) requires only temperature, wind and dewpoint data, apart
from the latitude and elevation. Various methods of estimating the dewpoint
(Td) are reviewed elsewhere, and it can be estimated graphically from the daily
extreme temperatures (Linacre, 1992). Often it equals the daily minimum
temperature, except in arid climates, where Td is lower. Any approximate
value for the windspeed is sufficient, possibly estimated from measurements
nearby or at other times.

CONCLUSIONS

A simplified version of Penman's evaporation formula (i.e. eqn. (22)) gives

estimates within about 0.25 mm day-~ of monthly measurements at Copen-
hagen, and of annual measurements at three lakes in the USA. Assessing an
estimate's error is complicated by practical uncertainties in measuring lake
evaporation.
The evaporation at a reservoir in Western Australia was measured in two
ways, and the average difference between the results is greater than the
apparent error of estimates from eqn. (22). The latter is about 0.8 mm day -t
for monthly values.
The coefficient of heat-transfer between water and the atmosphere is about
2.5 u F W m -2 K -t , where u is the windspeed measured at a height of 2 m, and
F is the fractional reduction of air density with elevation.
Two methods are described for estimating solar irradiance (which largely
controls evaporation), using either, (A) empirical annual mean and annual
range values, in terms of consequent changes of temperature, or (B) the
attenuation of extraterrestrial radiation by clouds, whose amount is deduced
from rainfall figures. The latter yields more accurate estimates of evaporation.
The annual range of monthly mean temperature Rr (involved in Method A
of estimating solar radiation) equals 1.3 A d °2 °C, where A is the degrees of
latitude, and d the distance (km) downwind of the ocean. Thus, what is often
called 'continentality' is proportional to d °2.

ACKNOWLEDGEMENTS

Thanks are due to Jaci Pratt and Elinor Hurst of the Country Fire Service
of South Australia for prompting me to write this paper, and for helpful
comments. A Macquarie University Research Grant facilitated the literature
search for heat-transfer coefficients. The diagrams were drawn by Judy Davis
(of the School of Earth Sciences, Macquarie University) and Kevin Cowan (of
the Geography Department, Australian National University).
252 E.T. LINACRE

REFERENCES

Abdulai, B.I., Stigter, C.J., Ibrahim, A.A., Adeeb, A.M. and Adam, H.S., 1990. Evaporation
calculations for Lake Sennar. Neth. J. Agric. Sci., 38: 725-730.
AMS, 1981. Proc. Int. Conf. on Climate and Offshore Energy Resources, 1980 London. American
Meteorological Society, Boston, MA, 253 pp.
Anyadike, R.N.C., 1987. The Linacre evaporation formula tested and compared to others in various
climates over West Africa. Agric. Meteorol., 39:111-1 t9.
Aslyng, H.C., 1960. Evaporation and radiation heat balance at the soil surface. Archiv. Meteorol.,
Geophys. Bioklimatol., Ser. B, 10: 359-375.
Aslyng, H.C., 1966. Weather, water balance and plant production at Copenhagen 1955-1964.
Arsskr. K.Vet-Landbohojsk., 1-21.
Barry, R.G., 1969. Evaporation and transpiration. In: R.J. Chorley (Editor), Water, Earth and Man.
Methuen, pp. 169-184.
Bruce, J.P. and Clark, R.H., 1966. Introduction to Hydrometeorology. Pergamon, Oxford, 324 pp.
Brutsaert, W., 1965. Evaluation of some practical methods of estimating evapotranspiration in arid
climates at low latitudes. Water Resour. Res., 1: 187-191.
Brutsaert, W., 1982. Evaporation into the Atmosphere. Reidel, Dordrecht, 299 pp.
Brutsaert, W. and Yu, S.L., 1968. Mass transfer aspects of pan evaporation. J. Appl. Meteorol., 7:
563-566.
Bunting, A.H. (Editor), 1987. Agricultural Environments. Commonwealth Agricultural Bureau
International, 335 pp. Or Agro-ecological Zonation. Cambridge University, 335 pp.
Businger, J.A., 1973. Turbulent transfer in the atmospheric surface layer. In: D.A. Haugen (Editor),
Workshop on Micrometeorology. American Meteorological Society, Boston, MA, pp. 67-100.
Cahoon, J.E., Costello, T.A. and Ferguson, J.A., 1991. Estimating pan evaporation using limited
meteorological observations. Agric. Forest Meteorol., 55: 181-190.
Chang, J.H., 1971. Problems and Methods in Agricultural Meteorology. Oriental, Taiwan, 96 pp.
Chang, J. H. and Root, B., 1975. On the relationship between mean monthly global reduction and
air temperature. Archiv. Meteorol., Geophys. Bioklimatol., Vienna, Ser. B, 23." 13-30.
Chapman, A.L. and Kininmonth, W.R., 1972. A water-balance model for rain-grown lowland rice
in northern Australia. Agric. Meteorol., 10: 65-82.
Chorley, R.J. (Editor), 1969. Water, Earth and Man. Methuen, London, 588 pp.
Christiansen, J.E., 1968. Pan evaporation and transpiration from climatic data. J. Irrig. Div., ASCE,
94: 243-265.
Clemence, B.S.E. and Schulze, R.E., 1982. An assessment of temperature-based equations for
estimating daily crop water loss to the atmosphere in South Africa. Crop Prod., 11: 21-25.
Esbenson, S.K. and Reynolds, R.W., 1980. Estimating monthly-averaged air-sea transfers of heat
and momentum using the bulk aerodynamic method. Rep. 16, Climate Research Institute,
Oregon State University, 31 pp.
Ferguson, H.L. and den Hartog, G., 1975. Meteorological studies of evaporation from Perch Lake.
A final summmary of the Perch Lake Evaporation Study 1965-1974. AECL-5041 / 1, Chalk River
Nuclear Laboratories, Chalk River, Canada.
Friehe, C.A. and Schmitt, K.F., 1976. Parameterization of air-sea interface fluxes of sensible heat
and moisture by the bulk aerodynamic formulas. J. Phys. Oceanogr., 6: 801-809.
Gloyne, R.W. and Lomas, J., 1980. Lecture notes for training Class II and Class III agricultural
meteorological personnel.WMO PuN. 551, WMO, Geneva, 260 pp.
Grace, J., 1981. Some effects of winds on plants. In: J. Grace et al. (Eds), Plants and their
Atmospheric Environment. Blackwell Scientific, Oxford, pp. 31-56.
Harbeck, G.E., 1958. Lake Hefner waterloss investigations. Publ. No. 45, IASH, Wallingford,
pp. 437-443.
DATA-SPARSEESTIMATIONOF LAKEEVAPORATION 253

Hargreaves, G.H., 1983. Practical agroclimate information systems. In: D.F. Cusack (Ed), Agrocli-
mate Information for Development: Reviving the Green Revolution. Westview, Boulder, CO, pp.
113-127.
Hicks, B.B., 1972. Some evaluations of drag and bulk transfer coefficients over water bodies of
different sizes. Boundary-Layer Meteorol., 3: 201-213.
Hicks, B.B., Drinkrow, R.L. and Grauze, G., 1974. Drag and bulk transfer coefficients associated
with a shallow water surface. Boundary-Layer Meteorol., 6: 287-297.
HMSO, 1956. Handbook of Meteorological Instruments, Part 1. HMSO, London, 458 pp.
Hounam, C.E., 1973. Comparison between pan and lake evaporation. WMO Tech. Note 126, WMO,
52 pp.
Hoy, R.D. and Stephens, S.K., 1977. Field study of lake evaporation: analysis of data from
Eucumbene, Cataract, Manton and Mundaring. Tech. Paper 21, Australian Water Resources
Council, Canberra, 195 pp.
Humphreys, W.J., 1964. Physics of the Air. Dover, New York, 676 pp.
Hylckama, T.E.A., 1974. Estimating evapotranspiration by homoclimates. In: Symp. on Design of
Water Resources Projects with Inadequate Data, Madrid, 1973, Vol. 1. WMO, pp. 73-88.
Inoue, E., Mihara, Y. and Tsuboi, Y., 1965. Agrometeorological studies on rice growth in Japan.
Agric. Meteorol., 2: 85-107.
Jacobs, W.C., 1942. On the energy exchange between sea and atmosphere. J. Mar. Res., 5: 37-66.
Jarvinen, J. and Huttula, T., 1982. Estimation of lake evaporation by using different aerodynamic
equations. Geophysica, 19: 87-100.
Jensen, D. (Ed), 1975. Consumptive Use of Water and Irrigation Water Requirements. ASCE, New
York, 215 pp.
Jobson, H.E., 1972. Effect of using averaged data on the computed evaporation. Water Resour. Res.,
8: 513-518.
Jones, P.A., 1991. Historical records of cloud cover and climate for Australia. Aust. Meteorol. Mag.,
39: 181-189. I
Katsaros, K.B., Smith, S.D. and Oost, W.A., 1987. HEXOS-humidity exchange over the sea: a
program for research on water-vapor and droplet fluxes from sea to air at moderate to high wind
speeds. Bull. Am. Meteorol. Sot., 68: 466-476.
Keijman, J.Q. and Koopmans, R.W.R., 1973. A comparison of several methods of estimating the
evaporation of Lake Flevo. In: Symp. on Hydrology of Lakes, 1973, Helsinki. IAHS, pp.
225-232.
Kessler, A., 1985. Heat Balance Climatology. World Survey of Climatology, Vol.IA. Elsevier,
Amsterdam, 224 pp.
Kohler, M.A. and Parmele, L.H., 1967. Generalized estimates of free water evaporation. Water
Resour. Res., 3: 997-1005.
Kohler, M.A., Nordenson, T.J. and Fox, W.E., 1955. Evaporation from pans and lakes. Res. Paper
38, US Department of Commerce, Weather Bureau, Washington, DC, 21 pp.
Lacy, R.E., 1977. Climate and Building in Britain. Building Research Establishment, UK,
185 pp.
Lakshman, G., 1972. An aerodynamic formula to compute evaporation from open water surfaces.
J. Hydrol., 15: 209-225.
Landsberg, H.E., 1969. Physical Climatology. Gray, Dubois, PA, 445 pp.
Large, W.G. and Pond, S., 1982. Sensible and latent heat flux measurements over the ocean. J. Phys.
Oceanogr., 12: 464-482.
Linacre, E.T., 1963. Determining evapotranspiration rates. J. Aust. Inst. Agric. Sci., 29: 165-177.
Linacre, E.T., 1964. Calculations of the transpiration rate and temperature of a leaf. Archiv.
Meteorol., Geophys. Bioklimatol., Ser. B, 13: 391-399.
Linacre, E.T., 1969. Empirical relationships involving the global radiation intensity arid ambient
temperature at various latitudes and altitudes. Archiv. Meteorol., Geophys. Bioklimatol., Ser. B,
12: 1-20.
254 E.T. LINACRE

Linacre, E.T., 1977. A simple formula for estimating evaporation rates in various climates using
temperature data alone. Agric. Meteorol., 18: 409-424.
Linacre, E.T., 1992. Climate Data and Resources. Routledge, 366 pp.
Linacre, E.T. and Hobbs, J.E., 1977. The Australian Climatic Environment. Jacaranda Wiley,
Brisbane, 354 pp.
Linsley, R.K., Kohler, M.A. and Paulhus, J.L.H., 1982. Hydrology for Engineers, 3rd edn. McGraw-
Hill, New York, 508 pp.
List, R.J., 1949. Smithsonian Meteorological Tables. Smithsonian Institution, Washington, DC, 527
PP.
Liu, W.T., Katsaros, K.B. and Businger, J.A., 1979. Bulk parameterization of air-sea exchanges of
heat and water vapour including the molecular constraints at the interface. J. Atmos. Sci., 36:
1722-1735.
Longacre, L.L. and Blaney, H.F., 1962. Evaporation at high elevations in California. J. Irrig. Drain.
Div., ASCE, 88: 33-54.
Mahrt, L. and Ek, M., 1984. The influence of atmospheric stability on potential evaporation. J. Clim.
Appl. Meteorol., 23: 222-234.
Marciano, J.J. and Harbeck, G.E., 1954. Mass-transfer investigations--Lake Hefner studies. Geol.
Surv., Prof. Pap. 269, US Department of the Interior, pp. 46-70.
Mcllroy, I.C. and Angus, D.E., 1964. Grass, water and soil evaporation at Aspendale. Agric.
Meteorol., 1: 201-224.
Meyer, S.J., Hubard, K.G. and Wilhite, D.A., 1989. Estimating potential evapotranspiration: the
effect of random and systematic errors. Agric. For. Meteorol., 46: 285-296.
Miller, D.H., 1981. Energy at the Surface of the Earth. Academic, San Diego, CA, 516 pp.
Monteith, J.L., 1963. Gas exchange in plant communmities. In: L.T. Evans (Ed), Environmental
Control of Plant Growth. Academic, pp. 95-112.
Morton, F.I., 1986. Practical estimates of lake evaporation. J. Clim. Appl. Meteorol., 25: 371-389.
Neuwirth, F., 1973. Experiences with evaporation pans at a shallow steppe-lake in Austria. In: Symp.
on Hydrology of Lakes, Helsinki. IAHS, pp. 290-297.
Olaniran, O.J., 1981. Empirical methods of computing potential maximum evapotranspiration.
Archiv. Meteorol., Geophys. Bioklimatol., Ser. A, 30: 369-381.
Pelton, W.L., King, K.M. and Tanner, C.B., 1960. An evaluation of the Thornthwaite and mean
temperature methods for determining potential evapotranspiration. Agron. J., 52: 387-395.
Pelton, W.L., King, K.M. and Tanner, C.B., 1960. An evaluation of the Thornthwaite and mean
temperature methods for determining potential evapotranspiration. Agron. J., 52: 387-395.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass.Proc. R. Soc.
London, Ser. A, 193: 120-145.
Penman, H.L., 1956. Evaporation: an introductory survey. Neth. J. Agric. Sci., 4: 9-29.
Pittock, A.B., Frakes, L.A., Jensen, D., Peterson, J.A. and Zillman, J.W. (Eds), 1978. Climatic
Change and Variability. Cambridge University, 455 pp.
Prescott, J.A., 1943. A relationship between evaporation and temperature. Trans. R. Soc. South
Aust., 67: 1-6.
Prescott, J.A. and Collins, J.A., 1951. The lag of temperature behind solar radiation. Q. J. R.
Meteorol. Soc., 77: 121-126.
Rijtema, P.E., 1966. Transpiration and production of crops in relation to climate and irrigation.
Tech. Bull. 44, Institute of Land and Water Management Research, Wageningen, pp. 49-74.
Rohwer, C., 1934. Evaporation from different types of pans. Trans Am. Soc. Civ. Eng., 99: 673-677.
Rouse, W.R., Mills, P.F. and Stewart, R.M., 1977. Evaporation in high latitudes. Water Resour.
Res., 13: 909-914.
Sanderson, M. and Ahmed, R., 1979. Potential evapotranspiration and water deficit in Bangladesh
using Garnier's modification of the Thornthwaite water balance. Climatol. Bull. 25, McGill
University, Montreal, pp. 13-24.
DATA-SPARSE ESTIMATION OF LAKE EVAPORATION 255

Slatyer, R.O. and Mcllroy, I.C., 1961. Practical Microclimatology. Australia, Commonwealth
Scientific and Industrial Research Organization, Canberra, 324 pp.
mass transfer approach. Isr. J. Earth Sci., 18: 101-108.
Stanhill, G., 1970. The heat and water balance of a fish pond in the Upper Galilee of Israel. Isr. J.
Agric. Res., 20: 21-39.
Stigter, C.J., 1980. Assessment of the quality of generalized wind functions in Penman's equations.
J. Hydrol., 45: 321-331.
Sweers, H.E., 1976. A nomogram to estimate the heat-exchange coefficient at the air-water interface
as a function of wind speed and temperature. J. Hydrol., 30: 375-401.
Szeicz, G., Enrodi, G. and Tajchman, S.J., 1969. Aerodynamic and surface factors in evaporation.
Water Resour. Res., 5: 380-394.
Szumiec, M., 1973. Evaporation from small reservoirs. In: Symp. on Hydrology of Lakes, Helsinki.
IAHS Publ. No. 109, IAHS, pp. 284-289.
Tanner, C.B. and Pelton, W.L., 1960. Potential evapotranspiration estimates by the approximate
energy balance method of Penman. J. Geophys. Res., 65: 3391-3413.
Thom, A.S. and Oliver, H.R., 1977. On Penman's equation for estimating regional evaporation. Q.
J. R. Meteorol. Soc., 193: 345-357.
Thorpe, M.R., Banke, E.G. and Smith, S.D., 1973. Eddy correlation measurements of evaporation
and sensible heat flux over Arctic sea ice. J. Geophys. Res., 78: 3573-3584.
Tucker, G.B., 1962. Tropospheric heating over the North Atlantic. Meteorol. Mag., 91: 184-190.
Van Bavel, C.H.M., 1966. Potential evaporation: the combination concept and its experimental
verification. Water Resour. Res., 2: 455-467.
Venkataraman, S. and Krishnamurthy, V., 1973. On the estimation of potential evaporation by the
combination approach. Archiv. Meteorol., Geophys. Bioklimatol., Ser. B, 21: 1-9.
Vikulina, Z.A., 1973. Evaluation of the Penman method for the computation of evaporation from
water bodies using observed data. In: Symp. on Hydrology of Lakes, Helsinki. IAHS, pp.
263-268.
Wang, D., 1986. Characteristics of the annual march of air temperature and the continentality in
China. Acta Geogr. Sin., 41: 70-76.
Warnaka, K. and Pochop, L., 1988. Analyses of equations for free water evaporation estimates.
Water Resour. Res., 24: 979-984.
Webb, E.K., 1960. An investigation of the evaporation from Lake Eucumbene. Tech. Aust., CSIRO,
Div. Meteorol. Phys., Tech. Pap. 10, 75 pp.
Webb, E.K., 1966. A pan-lake evaporation relationship. J. Hydrol., 4:1-11.
WMO, 1974. Guide to Hydrological Practices. WMO Publ. 168, WMO.
Woodhead, T., 1972. Mapping potential evaporation for tropical East Africa: the accuracy of
Penman estimates derived from indirect assessments of radiation and windspeed. In: Proc. Symp.
on World Water Balance, Reading, 1970. WMO, pp. 232-241.
Yao, A.Y.M., 1981. Agricultural climatology. In: H.E. Landsberg (Ed), General Climatology, World
Survey of Climatology, Vol. 3. Elsevier, Amsterdam, pp. 189-298.
Young, A., 1987. Methods developed outside the internationalagricultural research system. In: A.H.
Bunting (Ed), Agricultural Environments. Commonwealth Agricultural Bureaux International,
pp. 43-63. Or Agro-ecological Zonation. Cambridge University, pp. 43-63.

APPENDIX 1

Further papers yielding values of the heat-transfer coefficient for water

surfaces.
In each case, the inferred coefficient (W m -z K -1) is shown in brackets,
256 E.T. LINACRE

where u is the windspeed has been corrected (Linacre, 1992) to that at about
2 m height.
Rohwer (1934), quoted by Olaniran (1981.) (8 + 2.9u), Jacobs (1942) (2.7u)~
Penman (1948), quoted by Stigter (1980) (5 + 2.6u), Penman (1956) (3 +
2.6u), Harbeck (1958) (2.5u), Webb (1960) (2.0u), Tucker (1962) (1.9u), Inoue
et al. (1965) (4 + 3.3u), Bruce and Clark (1966) (2.5u), Kohler and Parmele
(1967) (3+ 2.8u), Brutsaert and Yu (1968) (5 + 3.3u), Landsberg (1969)
(2.7u), Jobson (1972) (2.7u), Lakshman (1972) (Y2u), Hicks (1972) (1.8u),
Businger (1973) (1.4u), Neuwirth (1973) (1 + Y8u), Vikulina (1973) (33 +
1.9u), Thorpe et al. (1973) (1.4u), Hicks et al. (1974) (1.Tu), WMO (1974)
(2.6u), Friehe and Schmitt (1976) (1.7u), Sweers (1976) (5 + 1.2u), Lacy (1977)
(4.1u), Pittock et al. (1978) (1.8u), Liu et al. (1979) (1.6u), Esbenson and
Reynolds (1980) (1.6u), Gloyne and Lomas (1980) (6 + 2.1u), Stigter (1980)
(3.3u), AMS (1981) (2.5u), Olaniran (1981) (8 + 1.3u), Brutsaert (1982) (1.2u),
Jarvinen and Huttula (1982) (3 + 2.1 u), Large and Pond (1982) (1.4u), Linsley
et al. (1982) (2u).
The part independent of windspeed, in the results from Penman, Inoue et
al., Brutsaert and Yu, Neuwirth, Sweers, Gloyne and Lomas, and Olaniran,
are in the range 1 - 8 W m -2 K -l, i.e. around 5 units. This might be due to
infrared radiation fluxes between the water and surfaces at air temperature,
which correspond to a net coefficient of about 5 W m -2 K -1 (Linacre, 1992).
The median of the 26 values above may be found by calculating the
coefficient in each case, for winds of 1 m s- ~, 2 m s- 1 and 3 m s- 1, and then
finding the median for each windspeed, i.e. 2.7 W m -2 K -~ , 5.0 W m - 2 K - I
and 7.7 W m - 2 K - I , respectively. These imply a coefficient of about 2.6 u W
m-2 K - l .