calculate the value of sensible heat flux *H*.

Table 4 clearly shows the variations of w′ *T *′ with u2m over a period of one year. The slope

varies from 0.0298 for the measurements made from May to August to 0.014 for the period from

December to the next April. In the MayAugust period, the sensible heat flux is about three times

greater than the value calculated for the DecemberApril period for a given Reynolds number *Re*.

covered ground than over grass-covered ground. This is evidenced in the variation of *w*′ *T *′ with

the ratio of u2m / *u** (Fig. 8). This trend can also be seen from Table 4 in which the slopes of the

relation between *u** vs. u2m , σw′ vs. u2m , and σ u′ vs. u2m are slightly lowered as the test period

expanded from May to August, December and then to April. The slopes of the relation between σu

vs. *u** (Table 4) increases slightly as the test period progressed as was expected, i.e., to have the

same value of *u** during the winter period (in this case December to April). It needs a higher value

of mean wind speed u2m and thereby creates a higher value of standard deviation of fluctuating

velocity in the horizontal direction.

Though the study reported here is preliminary in nature, it has clearly established the fact that

the sensible heat flux determined during the winter months or during snow- or ice-covered ground

is about only one third of the value obtained during spring and summer months, i.e., over grass-

covered ground. Therefore it is speculated that the use of expressions involving temperature and

mean velocity gradients evaluated at measurement height *z *to estimate sensible heat flux from

equations such as eqs 21 or 31 will definitely introduce a significant error because these equations

are derived without the consideration of snow surface as a lower-surface boundary.

For sensible heat flux determined under stable conditions (which exists mostly at night under

calm wind conditions), the data are much more scarce and showed a much greater dispersion,

discussed in section VII, Comparison of Sensible Heat Fluxes. For the case of stable atmospheric

conditions, Hicks and Martin (1972) reported an average value of 9 W/m2 (for a mean wind speed

of 2.7 m/s and ∆*T *of 3C), which is about one third of the value obtained from this study under

approximately the same test conditions (i.e., 27 W/m2) (but for a period of one year, the heat flux

reduces to 13 W/m2, which is close to the value of 14 W/m2 reported by Dyer (1961) as well as of

16 W/m2 reported by Bernier and Edwards (1989).

For unstable atmospheric conditions, Dyer (1961) conducted sensible heat flux measurements

over level pasture land with a mean wind speed of 4.2 m/s and r eported a heat flux value of 155

W/m2, which is nearly identical to the value of 180 W/m2 calculated from eq 84 based on the

present study. This close agreement may be accidental, because in another study Dyer (1967)

order of 3C, the value of *H *will be on the order of ~1100 W/m2, which is about seven times greater

than the value reported by Dyer (1961) as well as the value calculated based on eq 84.

In summary, based on several past studies as well as the present one, it can be concluded that not

only does sensible heat flux depend on a number of factors, but it is difficult to obtain consistent

results because of the constant changing of the atmosphere. The fact is that even during an

averaging period of 10 minutes the wind speed and its direction vary greatly. Therefore, the

fundamental assumption of constant, steady, and homogeneous upwind flow never can be realized

in the field measurement to satisfy the condition of a so-called constant flux surface layer. It is

expected that the results obtained with the eddy-correlation technique will provide a much more

reasonable estimation for the sensible heat flux than the other methods discussed in section III,

Computation of Turbulent Fluxes. Due to the constantly changing atmospheric dynamics, the

steady, homogeneous surface layer flow will never be realized in field measurements to ensure the

existence of the constant flux layer. Localized or regional measurements, which are used in this

study, may still provide a much more reliable method to find the sensible heat flux needed for

surface energy balance calculations.

39