Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Journal o f Hydrology, 150 (1993) 189-216 189

0022-1694/93/$06.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved

[31

History of forest hydrology

James S.G. McCulloch*'a, Mark Robinson b


aBurcot Tower, Burcot, Abingdon, OX14 3DJ, UK
bInstitute of Hydrology, Wallingford, OXIO 8BB, UK
(Received 1 June 1993; revision accepted 15 June 1993)

Abstract
Hydrology as a science and a technology is examined, as are some of the myths on the role of forests in
hydrology and water resources. The history of catchment area research is traced, in Europe, in the USA and
in East Africa, with particular reference to forest hydrology and, in the earlier years, to water quantity
rather than water quality. The importance of associating physical process studies with hydrological systems'
investigations, to enhance understanding of why particular catchments behave as they do, is stressed.
Recent advances in hydrochemistry have been exploited to elucidate water flow paths within experimental
catchments. Stimulated by requirements for research into acidification of surface waters, research catch-
ments have proved to be valuable outdoor laboratories from which a much improved understanding of the
flow processes has been achieved. Conflicting claims about the impacts of forestry are described and
discussed.

Introductory history of hydrology

Hydrology is the science of water. It deals with the hydrological cycle of


water, from precipitation through the various phases of water on the land
surface, interception by vegetation, infiltration into the soil, drainage to
ground water and runoff, either from surface or ground water. Eventually
(and, in the case of ground water, 'eventually' may be a long, long time), the
water which has not evaporated from the vegetation or from the land surface
will gravitate to the sea; from there it will evaporate and ultimately fall again
as rainfall. Water is blessed by being a renewable resource, howsoever man-
kind may waste it or pollute it. The concept of the hydrological cycle now
seems obvious and simple; it is incredible that it was only some 300 years ago
that Perrault's (1674) study of the basin of the River Seine led to this under-
standing. With all the requirement for water in all the early civilisations, just

* C o r r e s p o n d i n g author.
190 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189.216

how did mankind survive until then in ignorance of the hydrological cycle, the
most basic of the principles of the science of water? Until the late seventeenth
century, what was understood (or believed) about water?
Plato (428-348B.C.) accepted the ancient Greek (Empedocles of Agri-
gentum) theories about the primary elements of matter: air, water, earth
and fire; he added a fifth element, which Aristotle (385 322B.C.) sub-
sequently explained as ' heaven'. The karst geology of Greece, coupled with
the Greek interest in argument, rather than measurement, may well excuse
Plato for his acceptance of the Tartarus (huge subterranean reservoir)
explanation for the origin of rivers and springs; Aristotle derided this concept
and favoured the idea of conversion of air into water in the ground, by
analogy with the way cooling changes air into water above the earth. The
first correct measurement of streamflow as a function of velocity was by Hero
of Alexandria, and the first Briton known to write on any hydrometeoro-
logical matter was the Venerable Bede (A.D. 673 735). However, until
theory and description gave way to hydrological observation and experimen-
tation, progress in hydrology, as opposed to hydraulics, was very limited. The
British contributions to hydrology as a science from the time of Perraut
focused rather on physical processes in hydrology and on development of
instruments. Thus Halley (1656-1742) and Dalton (1766-1844) are prin-
cipally associated with evaporation, and Wren and Hooke with rainfall
measurement (Biswas, 1970). Recently, the US National Research Council,
in a publication entitled 'Opportunities in the Hydrologic Sciences', included
some 20 brief biographies of eminent hydrologists. In a US publication, it is
obvious that American hydrologists would predominate; the two Britons
included were Penman (1909-1984), whose equation for estimating potential
evaporation is perhaps the single most significant recent contribution to the
science of hydrology, although he laid no claim to be a hydrologist, and Hurst
(1880-1978), who spent his lifetime studying the hydrology of the Nile and
discovered a major empirical law concerning long-run dependence in
geophysics.
A study of ancient civilisations, however, indicates that mankind was well
versed in the use and management of water; the technology preceded the
science of hydrology by thousands of years. Malpas and Watson (1991)
defined technology as: 'the result of the systematic harnessing of all knowl-
edge and experience to produce something practical and commercially useful
- - a product, a manufacturing process, a system, a service, a methodology.'
They commented also that: 'Technology often runs ahead of science', and
quoted Lord Porter: 'Thermodynamics owes more to the steam engine than
the steam engine owes to science'.
The basic engineering skills which led to such wonders of the ancient world
J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216 191

as the Marib dam in Yemen, near the city of Marib, the capital of the Queen of
Sheba, were entirely adequate for planning and harnessing flows of natural
water for the diverse purposes of mankind, until some natural, but extreme,
event outside previous living memory. Such a perturbation, in the guise of an
extreme flood, was the cause of the catastrophic failure of the Marib dam,
after hundreds of years of safe and successful operation. China's Hwang Ho,
the river of sorrows, testifies by its name to the difficulty of prediction of
catastrophic events. Yet, even today, the conventional hydrological approach
is to seek at least a 30-year record of, say, annual maximum flows; it then is a
straightforward statistical exercise to predict (as opposed to forecast!) future
flows for a given return period of say 1 in 30 years or 1 in 100 years, always on
the assumption that the data provided are samples from a continuous distri-
bution, unaffected by any perturbations. However, the impact of mankind on
the hydrological cycle may well lead to such perturbations; in Africa, for
instance, entire upland catchment areas, covered in protected forest, have
been cleared within a few years to provide land for agriculture for previously
landless peasants. The series of annual flows far downstream of such major
changes in land use may well show (and indeed did, in the example mentioned
above) that a perturbation has invalidated the assumption of continuity on
which hydrological planning is generally based.
The distinction drawn by hydrologists between prediction and forecasting is
purely a matter of convention. A forecast indicates, say, the flow at a given
point on a given day in the future, whereas a prediction states a probability of
occurrence of a flow of a given magnitude. Despite recent improvements in
meterological forecasting of rainfall, forecasting the flow at a given point next
week or on the first of next month is still more an art than a science, unless one
is dealing with a large river basin with a long response time to incident rainfall.
Hydrology, as a technology, is really a study of extreme values in terms
of practical application, of floods and droughts; hydrology, as a science, is
concerned with understanding the various components of the hydrological
cycle and of the ways in which water moves between its various phases,
solid, liquid and gas. Until the late twentieth century, hydrology was prac-
tised more as a technology than as a science. Once an overall scientific under-
standing of a system or a process has been achieved, what was previously
research must be reclassified as simply routine application of a technology.
Water quality was, until recently, given scant consideration in matters of
forest hydrology. In Chow's (1964) 'Handbook of Applied Hydrology', only
37 pages out of 1500 were devoted to water quality. In the Sopper and Lull
(1967) 'Proceedings of the Forest Hydrology Seminar at Penn State Univer-
sity', no consideration was given to water quality aspects of forestry, except in
terms of soil stabilisation.
192 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189-216

Myths about the role of forests in hydrology and water resources

Forests cover about 4300 million ha, about one-third of the earth's land
surface (Food and Agriculture Organisation (FAO), 1998a, b), and comprise
about 2500 million ha of closed forest and 1800 million ha of open woodland
(consisting of savanna, shrub and degraded forest). The total area is more
than three times the area under crops, and an understanding of the
environmental role of forests is of great importance to water resources
internationally.
It is inevitable that, in the absence of a truly scientific basis for hydrology,
myths and legends will be promulgated and, with time, will become
accepted with all the authority of folklore. Forest 'hydro-logic' suggests
that:
(1) forests 'make' rain. This notion was first attributed to Christopher
Columbus; he contrasted the plentiful rainfall in the heavily forested West
Indies with that of the deforested Azores and Canary islands (Kitteridge,
1948; Rakhmanov, 1962). The idea gained support from findings that rain-
gauges in small forest clearings often catch more precipitation than those
outside the forest. Certainly, the belief that forests induce higher rainfall
and hence increase infiltration of water into the soil so that total streamflow
is increased was strongly held by East African foresters in the 1950s and 1960s.
Although there may be increased rainfall losses owing to interception of
rainfall by the forest canopy, it was believed that these will be offset by
reduced transpiration, as both depend on the same energy (Penman, 1948,
1963; Costin and Dooge, 1973).
(2) Natural forests reduce floods - - and erosion. It is c o m m o n knowledge
that rainfall infiltration into the soil is more effective through the forest
canopy and the underlying forest litter than through grassland or agricultural
crops. The leaf litter and the more porous soil beneath forests encourage
infiltration and water storage, rather than rapid overland flow. The resultant
reduction in surface runoff and in erosion (through the binding action of tree
roots) after rainfall in a well-forested catchment is likely to facilitate recharge
to underlying strata and groundwater. This will ensure the continuation of
baseflow well into and perhaps throughout the dry season.
(3) Forests augment low flows. As a result of the greater infiltration under
forest, Zon (1927), Kitteridge (1948) and Rakhmanov (1962) considered that
there was more opportunity for recharge to underlying strata. The effect is
likely to be enhanced in areas where snowfall and snowpacks form a
significant proportion of the precipitation input to a catchment. Also, snow
melts more slowly under forests, so runoff is less variable than for shorter
vegetation.
J.S.G. McCulloch, M. Robinson / Journal ~f Hydrology 150 (1993) 189-216 193

Early catchment experiments in Europe

Fears of increased erosion, more severe flooding and adverse effects on


climate after destruction of natural forests in the European Alps led to forest
protection measures for avalanche control in Switzerland dating back to the
fourteenth century; local decrees against clearing forests were made as early as
the sixteenth century in Austria, France and Italy to prevent increased torrent
flows (Kitteridge, 1948). The Swiss Federal Constitution of 1874 (para. 24)
provided for support for the correction and embankment of torrents as well as
for the afforestation of their catchments. Yet, in a series of disasters in the
Alps in the late nineteenth century, villages were flooded, landslides occurred
and agricultural land was damaged by deposition of sediment; this was
attributed to the extensive clearance of forests for pasture land and led to
the first true catchment study in 1900, in the Bernese Emmental region of
Switzerland (Engler, 1919) comparing the hydrological response to precipi-
tation of two 0.6 km 2 catchments, the Sperbelgraben (99% forested) and the
Rappengraben (69% pasture and 31% forest). From the Sperbelgraben catch-
ment, flood flows and annual water yields were lower, baseflows were higher
(Burger, 1943) and erosion, measured as bedload, was half that from the
Rappengraben pasture catchment; furthermore fewer landslides occurred.
Before 1916, measurements were not made in three or four of the winter
months. The pasture catchment was slightly larger in size, so the measured
area was reduced in 1927 to be more similar to that of the forest catchment,
and the type of flow structures was changed; a marked change in the water
balance for both catchments resulted. The experiment came to an end, tem-
porarily, in 1957, although measurements have since been restarted. The
period 1927 1956 is therefore the most suitable for comparison between the
catchments (Casparis, 1959).
Over this period, the mean annual precipitation was 1641mm for the
forested catchment (48% runoff) as compared with 1719mm (59% runoff)
on the mainly grassed catchment, i.e. the forested catchment had about 11%
lower annual streamflow. Peak flows (both snowmelt and storm rainfall) were
also lower from the forest, and this was attributed to higher measured infil-
tration capacities under the forest than in the grassland, as well as the slower
rate of snowmelt in the forest. In an analysis of streamflow peaks for the
period, the 50 highest rainfall peaks averaged 0.75m3s-lkm -2 from the
mainly pasture catchment compared with only 0.37m3s-lkm 2 from the
forested catchment. Similarly, comparison between dry weather flows from
the two catchments also revealed clear differences. A study of 83 periods of
flow recessions indicated that although flows after rainfall were initially higher
from the mainly pasture catchment, they declined at a faster rate than from
194 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189-216

the forested area, by about 10 days after rainfall they became lower and this
difference increased with time. The forested catchment therefore supported
the higher dry weather baseflows.
Penman (1959, 1963), on the basis of water balance considerations coupled
with his estimate of evaporation, cast doubts on the accuracy of measurement
of both precipitation and streamflow at Emmental. He stressed the likelihood
of differential leakage from the catchments, and suggested that the apparently
greater evaporative loss from forest was not proven. There has been increasing
recognition of the limitations of such inter-catchment comparisons when,
inevitably, other factors that may be of hydrological importance differ too.
There has, for example, been some suggestion that the pastures of the
Rappengraben catchment were unforested owing to its particularly shallow
soils, so that differences between the catchments' outflows might reflect soil
differences as well as vegetational ones. Nevertheless, the concept of the
benefits of forest protection and rehabilitation in mountainous areas is still
widely accepted (Keller, 1988).

Early catchment studies in the USA

Although there have been many catchment studies into forest hydrology
throughout the world (Bosch and Hewlett, 1982; Bruijnzeel, 1990), by far the
majority of such experiments have been carried out in the USA. Internation-
ally, some of the most noteworthy contributions to catchment area research
have been made in the USA, where it is called watershed research.
Forests cover approximately 30% of the conterminous USA. It is a country
with a rapid rate of development as a result of both its own population growth
and large-scale immigration; in consequence, extensive forest clearances
occurred in the latter part of the nineteenth century (Cox et al., 1985). How-
ever, the USA has greater resources to devote to environmental research and
conservation than many developing countries.
Interest in forest influences on the hydrology of the USA followed closely
that in Europe. Some of the early settlers from Europe had been concerned
that forest clearance, especially by burning, increased the severity of flooding.
Following a 'period of propaganda' (Kitteridge, 1948) in the late nineteenth
century between conservers of forests and of other natural resources and
special interest groups involved in cutting down the forests for timber and
for agriculture, the value of forests for the protection of streamflow became
accepted. In 1891, the National Forest Service was established to conserve
these natural resources, and the Weeks Act of 1911 authorised the acquisition
of lands to form national forests in the eastern USA to conserve the catch-
J.S.G. McCulloch, M. Robinson /Journal of Hydrology 150 (1993) 189-216 195

ments of navigable rivers. Yet, at the time, there were still no studies into the
effect of forestry on streamflow, although it was recognised that much was still
to be learned regarding the hydrological effects of forests; the case for their
conservation had yet to be proven. The first such study was in the winter of
1911 - 1912 in the White Mountains of New Hampshire; the flood crest from a
forested catchment was observed to be lower than that from a felled area
(Hoyt and Langbein, 1955).
The Wagon Wheel Gap experiment in southern Colorado, USA, was the
first true experimental (i.e. involving deliberate application of a treatment,
such as forest planting or clearing) catchment study (Bates and Henry, 1928).
Two small adjoining forested catchments (less than 1 km 2) were selected and
gauged in a steep mountainous area with about 530 m m y e a r -1 precipitation
(half as snow). The area has deep, well-drained soil composed mainly of
coarse sand, although rock fragments sometimes make up 50% of the top
1 m; the steady pattern of streamflow is dominated by a long snowmelt peak.
The forest was typically light and open, dominated by aspen, spruce and

0.30

"(2) Oo
gE 0.25

• •
I11
E
¢-

g 0.20
E
-r-

X
kU

0.15 i i i
O. 15 0.20 0.25 0.30
Control catchment flow (mm / day)

Fig. 1. Comparison of summer low flows (exceeded 95% of time in period 1 June 31 October) for
individual years before (squares) and after (circles) deforestation of the experimental basin at Wagon
Wheel Gap. Removal of the forest cover increased baseflows.
196 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189 216

Douglas fir. Measurements commenced in 1911 and, after an 8 year calibra-


tion period (to allow for systematic differences between the catchment out-
flows to be quantified), the forest was removed on one catchment in 1919-
1920. Studies were continued for a further 7 years, over which period the
cutting increased the annual streamflow (relative to the unchanged 'control'
catchment) by 30 mm, mostly because of an increase in the spring snowmelt.
However, there was also a small increase in summer low flows (Fig. 1),
attributed not only to lower summer transpiration (or evaporation) losses
from short vegetation as compared with forest, but mainly to increased
groundwater recharge in winter as a result of reduced winter evaporation
(Bates and Henry, 1928).

Classification of catchment experiments

Catchment studies may be grouped broadly into three main types:


(1) correlation studies in which the streamflow is compared between differ-
ent catchments. The simplest example selects two catchments to be as similar
as possible in all respects other than vegetation. This was the approach taken
in the Swiss catchment comparison at Emmental, described above. It has the
great advantage of providing results immediately. However, no two catch-
ments are ever completely identical, so some of the observed differences (but
how much is not known) may well be due to factors other than vegetative
cover. A more complex approach, which makes some allowance for the effects
of other variables, is a multiple catchment study using data from many
catchments representing a range of different conditions (forest cover, soils,
climate, etc.). The variability of flow variables (such as mean annual flow, or
flood frequency) is then related, statistically, to catchment parameters in a
multiple regression equation. In practice, this yields an empirical equation
with often about four 'independent' (but usually hopelessly interrelated)
variables. This provides a crude prediction of response (typically 60-80%
of the variability between catchments), but little understanding; potentially,
it can actually lead to misunderstanding! Thus, it was widely reported that
studies of streamflow from very large catchments in the former Soviet Union
'prove' that forestry increased river flows (Rakhmanov, 1962).
(2) Single catchment studies offer a more direct approach in which the
streamflow behaviour of a catchment is related statistically to climatic vari-
ables before the land cover is changed. This calibration then predicts the
pattern of flows that would have occurred in the absence of the land cover
change, for comparison with the flows actually observed. This solves the
problem of differences in other physical factors, as the same catchment is
J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216 197

studied under different vegetation covers. There is, however, the problem of
climatic variability to be addressed. The weather in the calibration period
before the vegetation change may be wetter or drier than in the period after-
wards.
(3) Paired catchment experimental studies are, thus, often preferred in
practice, and combine the correlation and single catchment experimental
approaches. This approach is based on two similar catchments which are
studied for a calibration period; then one catchment, the 'experimental' catch-
ment, is subjected to a change, and the other, the 'control', remains
unchanged. Comparison of the outputs of the two catchments before and
after the vegetation change allows the effects of climatic variability to be
identified. This was the approach adopted at the Wagon Wheel study
described above.
The early investigations, up to the 1960s, were concerned only with the
volume and distribution of water from the catchment areas, the water from
which was assumed to be pure and unaffected by any form of pollution, other
than sediment in the streams. Except for such studies as that on the impact of
cattle grazing in an experimental catchment at Coweeta (Johnson, 1952), few
questions of water quality, other than that of sediment, were considered
relevant.

Concept of physical process studies to support system studies

Until the second half of the twentieth century, tackling hydrological and
water resources problems involved less than consideration of the whole of the
hydrological cycle, or even of all the local hydrological system. Hydrology was
dominated either by the engineer, with an immediate problem of culvert,
bridge or dam spillway to design, or by the geographer, concerned more
about synoptic views of whole regions. Only two component phases of the
cycle were capable of direct measurement - - the input, precipitation, and the
output, runoff. Little consideration was given to the evaporation and storage
phases of the hydrological cycle, other than to calculation of evaporation 'by
difference'. Additions to soil moisture storage and ground water were cal-
culated, also 'by difference', after assuming a value for evaporation derived
from some empirical equation. Even nowadays, yields from catchments are
often quoted, particularly by engineers, as percentages of incoming rainfall.
Because, in a catchment, rainfall over the years is much more variable than the
more conservative potential evaporation, the tendency to quote a percentage
yield of a catchment may well reflect times when evaporation and soil moist-
198 J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216

ure were unknown and incapable of measurement and it seemed logical to


express the catchment yield in terms of the input on the assumption that the
terms in the water balance equation representing evaporation and changes in
soil moisture and groundwater storage were negligible!
However, to understand the results from such catchment area studies as
have been completed, to know their limitations and their applicability to other
areas, a knowledge and understanding of the physical (and chemical)
processes within the catchments is essential. Without this, the results are
difficult to extrapolate, and the observations and conclusions may well be
applicable only to situations where an identical set of processes is operating.
Every catchment is unique, so that if its problems must be solved on the spot,
without import of information and understanding from studies elsewhere,
then catchment area experimentation must be wholly uneconomic. To
encourage the quest for physical understanding of hydrological results,
Penman, in 1967, asked: 'Are we going to put all our energy in just measuring
what happens, or shall we put a little more effort in research to find out why
things happen?' One of the earliest process studies was into the interception of
rainfall by the tree canopy, by Horton (1919); this represents a loss of pre-
cipitation that would otherwise be available to the soil. From measurements
of rainfall from a set of large raingauges (43 cm diameter) under various tree
species, and also of stemflow, he produced curves (in general, hyperbolic in
form) of total interception (net rainfall corrected for stemflow) versus storm
rainfall. However, the concept of linking the energy balance with the water
balance implied that not all such intercepted rainfall was lost to the water
balance. Penman (1948) considered that 'whilst energy was being used up to
get rid of intercepted water, the same energy could not be used to get rid of the
transpired water', and, in 1963 noted that 'the same energy cannot be used
twice, and while the intercepted water is being evaporated, the drain on soil
water is checked'. The recognition that interception losses could be much
greater than evaporation estimates, and that only a part of this loss might
be balanced by reduced transpiration (Rutter, 1967) was the key to under-
standing how forests increased water losses and why different catchment
studies had produced a range of apparently conflicting results.
In the East African catchment experiments, begun in 1957-1958, the
water balance studies were enhanced by the best measurements then available
of the energy balance. The deep porous soils, coupled with routine measure-
ments of soil moisture, allowed 10 day running balances of the water budget
in the experimental catchments and hence checked the internal consistency
of the measurements overall (Pereira, 1961). Initially, all the changes in land
use affected the aerodynamic transport processes; however, for the most
economically beneficial alternative land uses the change was minimal in the
J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189 216 199

long term; thus pines replacing bamboo, and tea gardens replacing montane
forest, resulted in complete ground coverage of rather similar aerodynamic
characteristics.
The British Institute of Hydrology catchment area experiments involved
physical process studies from the outset, sometimes but not necessarily within
the experimental catchments. Soil physical, including nuclear, studies of soil
moisture content and fluxes, micro-meteorological measurements of atmos-
pheric water fluxes using Bowen ratio and eddy correlation techniques, lysi-
metric studies and nuclear measurements of forest canopy wetness, fluvial
geomorphological and soil chemical investigations - - all these physical pro-
cess investigations have contributed to a continuing gain in understanding of
the various processes occurring within the experimental catchments.

East African catchment area research experiments

In the East African countries of Kenya, Uganda and Tanzania, the prin-
cipal environmental variable is water. Given water, almost every known crop
can be grown. Without it, none! The awareness in the Old and in the New
World of concern about the hydrological consequences of changes in land use
in the upland catchments important for water supplies extended to the
situation in Africa. On the basis of the beliefs in the effectiveness of forest
in protecting the upland environment, substantial areas in East Africa had
been designated as forest reserves. Under colonial rule, these high-altitude
forests had been 'closed' to settlement or grazing by the indigenous popula-
tions, not because of the commercial value of the (largely cut-over) remnants
of montane forest but as a kind of land bank to protect the soils from erosion
and to preserve the dry weather streamflow on which the settlements down-
stream were entirely dependent. As growing local populations increased the
pressure on existing agricultural land, a series of catchment studies was
initiated, to seek answers to social and political questions on the hydrological
consequences of changes in land use in a variety of upland catchment areas.
Practical feasibility required that the experimental projects were restricted in
size to headwater catchments, only hundreds of hectares in size; within them
the entire catchment was considered as a hydrological system. The land use
options were realistic; they were laid down by the relevant authorities in the
different countries at the time. Depending on the outcome of the experiments,
political decisions would be taken to alter the land use over the entire area of
which the experimental catchments were considered representative. The land
use options were:
(1) for the Aberdare Mountains to the NE of Nairobi, in Kenya, in the
200 J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216

catchments which provided Nairobi's water supply, replacement of uneco-


nomic indigenous bamboo forest and occasional outliers of montane rain-
forest of, for example, Podocarpus sp., by commercially viable softwood
plantations or pasture for sheep grazing; (2) for an area of the Mau forest
reserve in north-western Kenya, replacement of a heavily cut-over regrowth of
montane rainforest by a tea estate; (3) for an area of overgrazed savannah
deteriorated to bush, in northern Uganda, clearing and subsequent controlled
grazing. Each hydrological experiment envisaged a number of catchments,
chosen in the belief that all were watertight and from which one would be
chosen, at random, for 'treatment', in respect of a change in land use. A
'calibration' period would be available to compare the hydrological perfor-
mance of the different catchments, before the land use was changed.
Whatever the limitations of the technology at the time, the East African
countries offered tremendous opportunities for a physicist undertaking
research in hydrology. The rainfall was seasonal and the soils in the impor-
tant catchment areas were deep and porous; so deep and porous indeed that a
whole season's streamflow could be stored within the soil profile, without ever
penetrating to the water tables, which tended to be at great depths. With no
more sophisticated instrumentation than a soil sampling tool, a weighing
balance and a drying oven, soil moisture storage, a previously unmeasured
component of the hydrological cycle, could be determined. In an era before
reliable soil moisture measurements by neutron gauges or techniques based on
measurements of dielectric constant were available, in East Africa routine
gravimetric soil sampling every 10 days at three sites within each catchment
was entirely feasible, largely because of the availability of adequately trained
local staff. Many years after the initiation of the project, the availability of soil
moisture neutron probes minimised the effort involved in determining soil
moisture; it also reduced the impact on the catchments of extensive soil
sampling.
All these experiments involved clear-cut (in both senses!) comparisons of
one treatment with another in complete, albeit relatively small catchments.
Elsewhere in the world, in more developed and more heavily populated
countries, such clearly defined comparisons over complete catchments
would be almost impossible to achieve.
In effecting these catchment experiments over the years, many problems
arose and were resolved, more or less successfully. The most difficult of
solution was (and always is) the question of the water tightness of the catch-
ments selected. Only over periods of measurement lasting for many years can
one be reasonably certain that no measurable input or output has been over-
looked. Inevitably, therefore, catchment research is not a short-term
operation. Another problem is that an experiment envisaged to run for 20
J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216 201

1400

1200

1000

g
E
a) 800

600

400

2(1(I

I958-60 1961-66 196 7- 73


Period (years)

[~ Water use of pines Extra water use by bamboo

Fig. 2. Increase in water use in a catchment covered with a growing pine plantation at Kimakia in Kenya,
comprising young seedlings (1958-1960), growing to canopy closure (1961 1966), and at canopy closure
(1967-1973), relative to a catchment covered with mature bamboo forest.

years, may outlive the socio-economic considerations which gave rise to it in


the first place. In fact, most of the East African projects were sufficiently well
conceived for their results to be of real value to the countries, even after the
changes which followed independence. Thus, in Kenya, substantial areas of
the Aberdares are now afforested with pines, which grow quickly and are a
national asset, and numerous areas of tea gardens, producing high-value
crops, have replaced indigenous montane forest of low productivity. In each
of these changes in land use, the alteration in the hydrological response to
rainfall and in the evaporation and transpiration from the catchment has
proven to be minimal (Fig. 2). The Uganda project, at Moroto, sited in an
area of savannah reduced to bush by over-grazing, showed that clearing the
bush and controlling the grazing restored the livestock-carrying capacity of
the land, a result of considerable relevance to the Ugandan government.
The East African Catchment Area Research project was begun in 1957
1958. Initial results were published in 1961 (Pereira), a volume of data in 1976
202 J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189 216

(Edwards et al.) and full results in 1979 (Blackie et al.). Finally, in 1981, the
results were summarised by Edwards and Blackie.

Later US catchment area experiments and development of Coweeta as outdoor


laboratory

Increasing flood damage in the two decades after the establishment of the
Wagon Wheel Gap experiment kept up public interest in forests and flooding,
and in 1936 the Omnibus Flood Control Act gave the US Forest Service
responsibility for flood surveys of forested catchments. As it is easier and
quicker to start a catchment area experiment by cutting down (after calibra-
tion) an already forested area, than to select a non-forested area and grow a
mature forest, most experiments are of the latter 'negative' type. Eighty per
cent of the catchment studies listed in the Bosch and Hewlett (1982) review of
the effects of forestry on streamflow were concerned with deforestation (by
logging, fire or insect attack).
At Coweeta in the eastern USA, approximately 60 years of data have been
collected on the relationship between land treatment and water resources
(Swank and Crossley, 1988). In an area of 18 km 2 of steep and rugged forest
landscape in the southern Appalachian Mountains, some 25 small catchments
have been studied, including the effects of afforestation and deforestation. The
area ranges in altitude from 680 to 1800m and has soils derived from
Precambrian gneiss, together with deep colluvial deposits along the stream
channels. The area is covered with hardwood forest. The study, commenced in
1933, has become the oldest continuously operating catchment research area
in the world. After a 6 year calibration period, certain of the sub-catchments
were subjected to various conversion treatments of the forest cover. There-
after, from the late 1950s, there was increased emphasis on studies of factors
controlling water movement, to develop sound principles for management to
improve water yield and to protect the environment. Through the 1960s, the
scope of the investigations broadened to include more environmental studies
and, in the 1970s, water quality and acid rain were included. Hewlett and
Helvey (1970) described the results of one of the sub-catchment experiments
on deep sandy loam soils with a very high infiltration capacity. A paired
catchment study in which one of two forested catchments was clearfelled
with minimal ground disturbance (to avoid soil compaction and erosion)
showed, clearly, an increase in peak flows and in the volume of storm runoff.
The Coweeta forested catchments inspired ecosystem research at many
other sites including Hubbard Brook. The Hubbard Brook catchment study
(Likens et al., 1977; Bormann and Likens, 1979) covers an upland area of
J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189-216 203

31 km 2 in the northern Appalachian Highlands of New Hampshire, USA. It


was set up in 1955, with a number of small catchments (0.1-1 km2). The area
ranges in altitude between 230 and 1000 m and has well-drained podzolic soils
derived from glacial till. The annual precipitation is about 1200mm, with
snow lying from December to April. Although the precipitation is spread
evenly through the year, the streamflow is not, with over one-third occurring
in April, largely as a result of melting of the winter snowpack. There is a
snowmelt flood threat in almost every year, and minimal baseflow in late
summer. The central aim of the study was to determine optimal methods of
management for controlling extremes of streamflow. Since the 1960s, it has
also become the site of intensive water quality studies (Pierce et al., 1970).
Many other forest catchment studies in the USA were mentioned in Bosch
and Hewlett's (1982) review. Small plot studies have also been conducted; for
example, Wilm and Dunford (1948), using a number of small (5ha) plots,
found that partial felling of lodgepole pine increased streamflow by 30%,
which they attributed to reduced interception losses.

British catchment area research experiments

Before 1960, the official British policy was to afforest the land of reservoired
catchments. However, Law (1956) demonstrated that the annual increase in
water demand by the conifers planted in the Stocks reservoir catchment of the
Fylde Water Board resulted in such a loss of the water resource that it was
necessary to purchase water from alternative sources to make up the shortfall.
Perhaps for the first time the economic consequences of the consumptive use
of a catchment's forest cover had been quantified in monetary terms! Social
and political awareness of the requirement of a fuller understanding of the
hydrological consequences of afforestation of the high-altitude water catch-
ments resulted in the creation, in 1962, of the British Hydrological Research
Unit (HRU), to undertake catchment area research. The H R U developed into
the Institute of Hydrology (IH), with much wider terms of reference, and
catchment experiments were established principally at Plynlimon, in mid-
Wales (Fig. 3), and at Balquhidder in Scotland. The IH also took over
responsibility for the East African catchment experiments, funded by the
UK Overseas Development Administration. However, the slow growth
rates of trees in the UK made catchment area research involving sequential
complete changes in land use unrealistic treatments in any reasonable time-
scale. Furthermore, as all the available catchments extended beyond the tree
line, 100% forest cover was never a possibility. In Wales, at Plynlimon, one
204 J.S.G. McCulloch, M. Robinson/Journal ~?/Hydrology 150 (1993) 189-216

800

700
E
0

E 600

0
'.,= 500
._o_

400'

300 -

200
l ] I I
1970 ] 975 1980 1985 1990
Year

Fig. 3. Annual water use of the Plynlimon catchments, showing the greater water use of the forested basin,
and the sensitivity of the absolute values to climatic variability, indicating the necessity for catchment
experiments to be of a long term nature.

sub-catchment was clearfelled after a lengthy calibration period, and the


hydrological behaviour was compared with a largely forested catchment
and with a catchment occupied only by sheep pasture (Kirby et al., 1991).
Figure 3 shows the sensitivity of catchment experiments to climatic variability
and hence the necessity for long term catchment studies (as opposed to the
'calibrate, cut and publish' philosophy sometimes observed). In Scotland, one
catchment at Balquhidder was forested and one was grassed; after calibration,
part of the grassed catchment was planted with trees and the forested catch-
ment was felled (Whitehead and Calder, 1993).
Although detailed systems' studies of experimental catchment areas can
produce definitive results for the particular area under study, the results
seldom merit the expense of time and effort in the investigation, unless they
are capable of extrapolation to other areas in similar environments. The
means of justifying such extrapolation are physical process studies; one
facet at a time, the mystery of results inconsistent with researchers' previous
J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189 216 205

understanding may be explored. The significance of interception of rainfall by


vegetation in a frequently wetted stand of forest was appreciated by Rutter
(1967), but was understood fully only after detailed micro-meteorological
examination of measurements over flat(!) uniform forest at Thetford,
Cambridgeshire (Stewart and Thorn, 1973). The results of these investi-
gations, too complex and expensive to be repeated in individual catchment
areas, were backed by extensive studies of throughfall and stemflow in lysi-
metric blocks of forest within the Plynlimon catchment. Similarly detailed
studies using physical techniques resolved the mystery of reduced evapora-
tion from the catchments in Highland Scotland (Calder, 1993).
Over the years, the British catchment projects and their associated process
studies enhanced understanding of the comparative roles of trees and grass in
the evaporative phase of the hydrological cycle; evaporation of the rain
intercepted by the leaves of the trees doubled the evapotranspiration from
the wet and windy catchments at Plynlimon (Kirby et al., 1991). Clearly, such
a result should have been of particular relevance to Highland Scotland, but
the initial results of the Balquhidder catchment study as a whole appeared to
contradict this expectation. Without physical process investigations, an
impasse would have been reached; in fact, the process studies revealed that,
at Balquhidder, the enhanced evaporation resulting from interception of rain-
fall by the forest canopy was matched by the surprisingly high water use of the
extensive heather in the grassed catchment and the minimal water use of grass
at the low temperatures experienced for much of the year above the tree line in
Scotland (Whitehead and Calder, 1993).

Some other experimental studies

Liebscher (1972) compared two small catchments (about 0.75 km 2) in the


Upper Harz mountains in Germany. Instrumented in 1948, the Wintertal was
covered by spruce forest and the Lange Bramke catchment had recently been
clearfelled and had a grass cover. Since then, long-term studies of the water
balance and of water chemistry and flow processes have been carried out.
Forest regrowth on the Lange Bramke has reduced streamflow, whereas
clearfelling on the Wintertal has led to an increase in annual flows.
The long-term Krofdorf forest experiment in Germany (Brechtel and
Fuehrer, 1991) is a sophisticated multiple paired catchment study. Four
small forested catchments (10 20ha) were instrumented in 1971; after a 10
year calibration period, two of the basins were subjected to forest cutting,
and the other two were left unaltered. Changes in runoff and water quality
are being assessed by comparison of subsequent flows with those predicted
206 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189 216

using the statistical relationships between the catchments derived from the
calibration period. Surprisingly, perhaps, the use of two control basins yielded
somewhat different predictions. Furthermore, these predictions proved to be
sensitive to climatic variations between the calibration and experimental
phases, despite the relatively long calibration period.
The greatest change in forest area is undoubtedly in tropical regions, which
contain almost half of the world's closed forest. Bruijnzeel (1990) reviewed the
impacts of conversion of tropical forests, and emphasised the following:
(1) the effect of forest conversion in the short term is dependent on the
intensity and manner in which the clearance is carried out. Also, there is
often an enormous disparity between the techniques of 'good management'
that are generally adopted in research experiments and the methods actually
used in practice.
(2) Over the longer term, the effect is very dependent on the character of the
new land use cover and its management. For example, the longer-term effects
of forest clearance will be less if the land is well managed subsequently for a
rubber plantation, for a tea estate or for softwood forestry than for, say,
pastureland or arable cropping.

Considerations of water quality

For many years, it was widely accepted that forestry was a natural pro-
tection for the chemical and biological purity of streams. Although catchment
area research experiments involving complete changes in land use have been
invoked to answer particular questions on the hydrological consequences for
water supplies of alterations in the land use of upland catchment areas,
emphasis until recently was placed entirely on water quantity, with scant
regard for considerations of water quality. The principal impact of changes
in land use was thought to be in the evaporation and transpiration from the
catchments, although changes in land use might also alter the degree of
erosion. However, there was concern about the build-up of nitrates in soil
and ground water. Enthusiasm developed for studies of the principal hydro-
logical processes controlling catchment behaviour, in place of the 'black box
models' previously favoured. Also, the possibilities of exploiting advances in
hydrochemistry in tracing flow processes within the catchments was appre-
ciated. After Rosenqvist's (1978) demonstration of the importance of within-
catchment processes and hence of the requirement for catchment chemistry
balances, Sklash and Farvolden (1979) showed how the chemical character-
istics of rainfall, even in rapid response catchment situations, did not pre-
dominate; using stable isotope data, they proved that the bulk of the water
J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216 207

in stream flood hydrographs was not the storm rainfall itself but water
already in the catchment, as soil water and ground water. However, the
major impetus in applying hydrochemistry to catchment research was the
'acid rain' debate and the possible acidic implications of afforestation and
deforestation (Christophersen and Neal, 1990). This represents a growing
recognition of the interaction between 'natural' internal processes and
external effects (anthropogenic effects such as sulphate and nitrate oxide
deposition).
It is now widely accepted that forests are associated with increased acidifi-
cation of streams in areas subject to significant atmospheric pollution. The
low aerodynamic resistance of the forest, which leads to increased evaporative
losses of the water intercepted by the forest canopy (as compared with those
from shorter vegetation), also results in more deposition of pollutants, owing
to the greater aerodynamic roughness and larger leaf area. This is true for
both wet deposition of cloud and mist droplets, which contain significantly
higher ionic concentrations than larger raindrops, and for dry deposition by
impaction of particles and by reactive gases.
The chemical outputs from forested and agricultural activities are affected
by differences in surface and sub-surface flow paths as well as by such manage-
ment practices as ploughing, drainage and fertiliser applications.

Resolution of some myths about the effects of forests

What, therefore, can be stated about the results of nearly a century of


catchment areas research, with particular reference to forest hydrology?
Have the results of extensive studies of forested catchments justified the
expense in time and effort over the years? Is there now sufficient understand-
ing of the impact on the hydrological cycle of planting new forests or remov-
ing indigenous forests in upland catchments areas? Where an effect can be
demonstrated, is it caused simply by the presence of trees, rather than of an
alternative vegetation? To what extent does the effect reflect the diverse
management practices adopted by different national forestry organisations?
These include pre-planting soil drainage and cultivation, application of arti-
ficial fertilisers and disturbance on thinning and felling. This assessment must
take into account not only those aspects centred on the physical volume of
water harvested (which was the objective of the original studies of complete
catchments), but also qualitative considerations such as interception of
chemical aerosols, release of nitrates on harvesting, etc., all of which are
important in determining the environmental impact of the establishment or
removal of forest.
208 J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216

Do forests 'make' (i.e. attract) rain?

Apart from the well-known case of interception of occult precipitation on


forest canopies from mist and cloud, there is no evidence that planting or
cutting of forest affects the rainfall experienced in an area in any way.
Indigenous forests tend to occur at higher altitudes, which often have a higher
rainfall than nearby lowlands. Furthermore, it is difficult to measure,
accurately, rainfall in a forest, other than with a raingauge mounted at
canopy level; to comply with the meteorologically recommended maximum
screening angle of a raingauge sited at ground level, a raingauge clearing in
30m tall forest would need to be large, of the order of I ha, which would
generally be unacceptable to foresters. Raingauges in clearings of smaller size
may be over-sheltered and so may not record the true rainfall, a fact which
may well contribute to the myths about forests and rainfall. As experiments in
cloud seeding prove, it is difficult to substantiate any trend or changes in
rainfall data to any acceptable level of statistical 'proof'. This includes total
amount of rainfall (the basic data to determine which may themselves be
subject to differential bias owing to possible measurement errors in such
markedly different vegetations as forest and short crops), rainfall duration
(testing claims of shorter storms in the absence of forest as compared with
longer falls), and rainfall intensity (heavier falls in the absence of forest).
However, in the Sahel, deforestation might alter the radiation balance such
that incoming, possibly convective, rainfall might be reduced (Charney, 1975).

Do forests reduce floods and hence erosion?

The idea that natural forests reduce floods and hence, soil erosion, is
soundly based in the appreciation that rainfall infiltration into the soil is
more effective through the forest canopy and the underlying forest litter
than through agricultural crops or grassland; the resultant reduction in sur-
face runoff and in erosion (through the binding action of tree roots) after
rainfall in a well-forested catchment is likely to facilitate recharge to under-
lying strata and ensure the continuation of baseflow well into and perhaps
throughout the dry season. However, the interception of precipitation by the
tree canopy is not necessarily balanced by a corresponding decrease in trans-
piration losses as was thought (Penman, 1948, 1963; Costin and Dooge, 1973),
and so may represent a partial loss of water to the soil and to streamflow.
Forests may reduce small floods but, generally, not extreme events. The
effect may depend upon the age of the trees and on the particular forest
management; in the UK, for example, the conventional pre-planting ground
treatment, before afforestation, such as the installation of roads and extensive
J.S.G. McCulloch, M. Robinson / Journal of ttydrology 150 (1993) 189-.216 209

drainage, tends to increase flood peaks. The impact of forestry on flood flows
at a catchment outlet also depends upon the relative timing of flows from the
component sub-catchments.
Forests may reduce erosion, especially in comparison with clearfelled land
on steep slopes, but under certain managerial practices, e.g. where extensive
drainage has been installed across the contours, erosion may be increased. It is
very important to distinguish clearly between the effects of logging and land
cover change in forest clearance studies. Many of the observed effects of
logging are more the result of soil compaction and logging roads than the
removal of the forest biomass. Thus, deforestation studies may exaggerate the
true soil-protective role of forestry against erosion by increasing erosion
through the disruptive logging methods used. In addition, forest fires may
change soil properties; an extreme case is that of hydromorphic soils which
resist rewetting after burning.

Do forests augment low flows?

By maximising infiltration into the soil and hence to ground water, as


opposed to surface runoff, forest undoubtedly augments low flows in some
circumstances (Pereira, 1992); on occasion, these flows are maintained right
through the dry season. However, rooting depths of trees are generally greater
than those of short crops and the effect on low flows of greater dry season root
abstraction may outweigh the benefits of higher infiltration rates. Calder et al.,
(1992) working in India, have shown that, because of increased rooting depth,
the annual water use of indigenous trees and of eucalypts without access to
ground water was twice that of an annual agricultural crop. In East Africa,
low pressure washing out of root profiles indicated that perennial stolonifer-
ous grasses, such as Pennisetum clandestinum had rooting systems similar in
extent to montane forest trees and eucalypts. Robinson et al., (1991) noted a
reduction in annual flows and a much greater proportional reduction in low
flows following the afforestation of grassland by conifers. This may be attrib-
uted to higher interception and transpiration losses in summer. Artificial
drainage of forest plantations may also increase dry weather flows, at least
in the early stages of a forest rotation (Robinsom 1986), by removing water
from deeper layers of the soil profile.

Do forests evaporate more water than alternative crops?

Although forests increase catchment evaporation owing to interception


(Calder and Newson, 1979), there may be exceptions, particularly in tropical
areas, where rainstorms are heavy but infrequent. Transpiration from forest
210 J.S.G. McCulloch, M. Robinson /Journal of Hydrology 150 (1993) 189-216

may be reduced because of the poor state of the trees, which may be over-
mature or severely damaged by pollution. Partial cutting of forest may well
enhance evaporation from the remaining vegetation and understorey, so
largely counteracting any reduction in interception losses (Kitteridge, 1948;
Hornbeck, 1975; Greenwood et al., 1985).
As no continuously growing vegetation is known to evaporate and transpire
more water than do trees, forests probably reduce streamflow overall. Despite
the higher infiltration and water storage capacities of forest soils, the net
rainfall under tree canopies is less than in open areas owing to interception
by the canopy, and much of the infiltrated water is transpired by the trees. In
interpreting the results of catchment investigations on low flows, it is very
important to define, rigorously, the status of the non-forested land in any
comparisons: land which has been deforested may behave very differently
from land under short natural vegetation, as a result of soil compaction,
erosion, etc. Local geology and climatic regime are also important.
As forest albedo is generally less than that for short crops or grasslands,
more direct solar radiation is absorbed by the forest, over a substantially
greater depth of penetration. While this effect may contribute to reduced
temperature and humidity gradients in forests, the better aerodynamic mix-
ing is the principal reason for gradients being of the order of tenths of a degree
as compared with those over short crops - - a fact which complicates their
measurement in physical process studies.

Is the scale of experimentation adequate?

Above all, the interrelationship between forests and hydrology depends on


scale. Although an upland catchment is more characteristic of an extensive
region than a lysimeter, on what scale must experimental catchments be
operated for the results to be capable of extrapolation? Whereas measure-
ments of incoming radiation would appear to set an upper bound to the
energy available for evaporation and transpiration, the micro-meteorological
Bowen ratio experiments at Thetford (Stewart and Thorn, 1972) and the
Plynlimon catchment area measurements both indicate that substantial
amounts of energy may be advected to an experimental catchment (Shuttle-
worth and Calder, 1979). As regards rainfall, Shuttleworth (1988) suggested
that extensive deforestation of the Amazon basin might result in significant
reductions in rainfall in the catchment, as about half of the incoming rain
originates from forest evaporation. Likewise, in interpreting the results of
basin studies for the effect of forestry on flood peaks in particular, it is
necessary to consider the scale. Experimental feasibility requires that the
size of many catchment studies is less than 80ha (Bosch and Hewlett,
J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189-216 211

1982). Changes in peak flows at this scale are not in themselves sufficient
evidence to prove a change at a larger scale. Hewlett and Helvey (1970)
argued that changes in flood volumes are of more significance, as their effect
downstream is more additive than simple peak flows, owing to flood wave
attenuation; in addition, their timing is crucial.

Is the experimental treatment realistic?

Finally, it still needs to be emphasised that there may be a great difference


between the manner of forest treatments in experimental catchments and in
actual forestry practice.

Towards the future

Over the years, crises of public confidence in environmental matters have


stimulated, or discouraged, advances in the hydrological sciences. Twenty
years ago, a new ice age was thought to be approaching, and the Club of
Rome caused equally great concern with its publication of 'Limits to Growth'.
In the UK, there were fatalities through river flooding; hence funding was
made available for the UK Flood Studies Report (Natural Environment
Research Council (NERC), 1975). Subsequently, a 1 in 200 year drought in
1975 1976 caused inconvenience but no fatalities; thereafter, investment in
water resources research was cut! All such national and international panics
are now subsumed into the present concern about global warming. Although
such crises may serve a useful purpose in leading to improvements in measure-
ment techniques and networks, under no circumstances should they be used as
a form of scientific blackmail, to compel governments to spend more and
more on collecting data for data's sake; there is a danger of a backlash of
public opinion and a growing public scepticism which is harmful to science in
the long run. The cost of ever-increasing networks coupled with greater
frequency of observation from improved, automated, instrumentation may
well prove prohibitive. Cost and benefit must be balanced. The marginal cost
of each percentage point in the quest for greater precision may far outweigh
the benefit.
At the outset of the international scientific programmes for hydrology, such
as UNESCO's International Hydrological Decade (IHD) and subsequent
Programmes (IHP) and WMO's Operational Hydrology Programme, the
long-term objectives were thought to be the measurement of hydrological
variables on all the catchments all over the world. However, it was eventually
realised that the idea of ground-based observations in other than a select
212 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189 216

number of representative catchments was unrealistic and hence, unachievable.


However, advances in remote sensing of data, even of flow data, have made
international monitoring of water resources more realisable. This, coupled
with the considerable improvements in global circulation models (GCMs)
has enhanced the contribution of hydrology to the assessment of the potential
impacts of global warming. Even now, hydrological inputs to GCMs facilitate
statistical estimation of the role of deforestation in Amazonia, Africa and
Indonesia (Polcher and Laval, 1993).
Nevertheless, during the IHD and IHP, numerous benchmark, represen-
tative and experimental basins were established. In proportion to the expense
and effort of installation and operation, the results seldom justified transfer to
ungauged areas; neither did they increase the understanding of the hydro-
logical behaviour of the regions. Not until the establishment of the
F R E N D (Flow Regimes from Experimental and Network Data) project
(Gustard et al., 1989; Roald et al., 1989) was any major concerted inter-
national attempt made to extrapolate research results beyond individual
catchments. Catchment data from over 40 European agencies were brought
together in an extensive database which was used for statistical analysis and
modelling to generalise the results. In monitoring natural and man-made
changes in hydrological regimes, the (mapped) proportion of forest cover
was obtained for many of the catchments studied. However, despite the
sometimes dramatic changes which may follow changes in forest cover, no
statistically significant relationship was found between the proportion of
forest cover and flow parameters such as mean annual flood and the
95 percentile exceedence low flow (Gustard et al., 1989). Part of the reason
may be the lack of statistical independence between forestry and other catch-
ment characteristics. The tendency for forests to be associated with the wetter
areas, less suitable for agriculture, may account for the finding of Rakhmanov
(1962) that annual stream flows were higher from forested than from
non-forested catchments.
In the many parts of the world where irrigation is practised, problems of
rising salinity must be considered. In Western Australia, such salinity prob-
lems owing to rising water tables have been exacerbated by the clearing of
(mostly) eucalypt forest. In view of the tradition that eucalypts 'mine' water,
this effect of rising water tables after clearance appears to 'prove' the con-
siderable difference in water use of different vegetations. However, the subject
is well addressed in several of the papers presented at this meeting, and so is
mentioned here only for completeness.
The foresight of the planners of well-conceived catchment area research has
resulted in a range of exemplary outdoor ecological laboratories in several
countries. Within these, the opportunities for multi-disciplinary research far
J.S.G. McCulloch, M. Robinson /Journal of Hydrology 150 (1993) 189--216 213

outweigh the vision of the pioneers; they, nevertheless, would be delighted at


the improvements in instrumentation now available for measuring many of
the components of the water balance of the hydrological system within an
experimental catchment. Research into forest hydrology has, indeed, devel-
oped out of all recognition over the last 20 years, largely because of the
introduction of novel techniques to measure facets of the hydrological cycle
not previously measurable. These include neutron and capacitance probes for
non-destructive determination of soil moisture and zero flux plane displace-
ment techniques for measuring water fluxes in the soil; automatic weather
stations capable of relaying their observations of all the relevant meteor-
ological variable by satellite, if necessary from the other hemisphere; remote
sensing of most of the hydrological variables, even of flow data, from
inaccessible sites. Above all, the hydrochemist has joined forces with the
physicist in probing the processes in operation within experimental catch-
ments. Sophisticated micro-meteorological equipment utilising Bowen ratio
and eddy correlation techniques provides spot determinations of evaporative
fluxes within the boundary layer. Stable isotope techniques and chemically
inert tracers provide chemical fingerprints which allow water fluxes within
catchments to be determined and the processes unravelled. Increasingly, all
these facilities are being assembled in major inter-disciplinary, international
investigations, which have already led to considerable improvements in the
early generations of global circulation models.

References

Bates, C.G. and Henry, A. J., 1928. Forest and streamflow experiment at Wagon Wheel Gap,
Colorado. Mon. Weather Rev. Suppl., 30: 1-79.
Biswas, A., 1970. History of Hydrology. North-Holland, Amsterdam, 336 pp.
Blackie, J.R., Edwards, K.A. and Clarke, R.T. (Editors), 1979. E. Aft. Agric. For. J., Special
Issue, 43: 1-413.
Bormann, F.H. and Likens, G.F., 1979. Pattern and Process in a Forested Ecosystem.
Springer-Verlag, New York, 253 pp.
Bosch, J.M. and Hewlett, J.D., 1982. A review of catchment experiments to determine the effect
of vegetation changes on water yield and evapotranspiration. J. Hydrol., 55:3 23.
Brechtel, H.M. and Fuehrer, H.-W., 1991. Water yield control in beech forest a paired
watershed study in the Krofdorf forest research area. IAHS Publ. 204:477 484.
Bruijnzeel, L.A., 1990. Hydrology of Moist Tropical Forests and Effects of Conversion: A State
of Knowledge Review. Free University, Amsterdam, 224 pp.
Burger, H., 1934. Einfluss des Waldes auf den Stand der Gewasser. II. Mitteilung. Der
Wasserhaushalt im Sperbel - und Rappengraben yon 1915/16 bis 1926/27. Mitt. Eidg.
Anst. Forstl. Versuchswes, 18:311 416.
Calder, I.R., 1993. The Balquhidder catchment balance and process experiment results in
context - - What do they reveal? J. Hydrol., 145: 467-480.
214 J.S.G. McCulloch, M. Robinson / Journal of Hydrology 150 (1993) 189 216

Calder, I.R. and Newson, M. D., 1979. Land use and upland water resources in Britain - - a
strategic look. Water Resour. Bull., 16: 1628-1639.
Calder, I.R., Swaminath, M.H., Kariyappa, G.S., Srinivasala, N.V., Srinivasala Murthy, K.V.
and Mumtaz, J., 1992. Deuterium tracing for the estimation of transpiration from trees. J.
Hydrol., 130:37 47.
Casparis, E., 1959. 30 Jahre Wassesmesstationen im Emmental. Mitt. Schweiz. Anst. Forst.
Versuchswes., 35:179 224.
Charney, J. G., 1975. Dynamics of deserts and droughts in the Sahel. Q. J. R. Meterol. Soc.,
101:193 202.
Chow, V.T., 1964. Handbook of Applied Hydrology. McGraw Hill, New York, 1467 pp.
Christophersen, N. and Neal, C., 1990. Linking hydrological, geochemical and soil chemical
processes on the catchment scale: an interplay between modeling and field work. Water
Resour. Res., 26:3077 3086.
Costin, A.B. and Dooge, J.C., 1973. Balancing the effects of Man's actions on the hydrological
cycle. In: Man's Influence on the Hydrological Cycle. Irrigation and Drainage Paper 17.
FAO, Rome, 19-51.
Cox, T.R., Maxwell, R.S., Thomas, P.D. and Malone, J.J., 1985. This Well-Wooded Land:
Americans and their Forests from Colonial Times to the Present. University of Nebraska
Press, Lincoln.
Edwards, K.A. and Blackie, J.R., 1981. In: R. Lal and E.W. Russell (Editors), Tropical
Agricultural Hydrology. Wiley, Chichester, pp. 163-188.
Edwards, K.A., Blackie, J.R., Cooper, S.M., Roberts, G. and Waweru, E.S., 1976. Summary of
Hydrological Data from the EAAFRO Experimental Catchments, East African Com-
munity, Nairobi, 282 pp.
Engler, A., 1919. Untersuchungen fiber den Einfluss des Waldes auf den Stand der Gewasser.
Mitt. Schweiz. Anst. Forst. Versuchswes., 12, 636 pp.
FAO, 1988a. An interim report on the state of forest resources in the developing countries.
FAO FO/MISC/88/7, Rome.
FAO, 1988b. Yearbook of Forest Products. FAO, Rome.
Greenwood, E.A., Klein, L., Beresford, J.D., Watson, G.D. and Wright, K.D., 1985. Evapora-
tion from the understorey of the Jarrah Forest, SW Australia. J. Hydrol., 80: 337-349.
Gustard, A., Roald, L.A., Demuth, S., Lumadjeng, H.S. and Gross, R., 1989. Flow regimes
from experimental and network data (FREND) I and II. Institute of Hydrology,
Wallingford.
Hewlett, J. and Helvey, J., 1970. Effects of forest clear felling on the storm hydrograph. Water
Resour. Res., 6: 768-782.
Hornbeck, J.W., 1975. Streamflow response to forest cutting and vegetation. Water Resour.
Res., 11: 1257-1260.
Horton, R.E., 1919. Rainfall interception. Mon. Weather Rev., 47: 603-623.
Hoyt, W.G. and Langbein, W.B., 1955. Floods. Princeton University Press, Princeton, N J,
469 pp.
Johnson, E.A., 1952. The effect of farm woodland grazing on watershed values in the South
Appalachian mountains. J. For., 50:109 113.
Keller, H.M., 1988. European experiences in long-term forest hydrology research. In: W.T.
Swank and D.A. Crossley (Editors), Forest Hydrology and Ecology at Coweeta, Ecological
Studies No. 66. Springer Verlag, New York, pp. 407-459.
Kirby, C., Newson, M.D. and Gilman, K., 1991. Plynlimon research: the first two decades.
Institute of Hydrology, Wallingford, Rep. 109, 188 pp.
J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189 216 215

Kitteridge, J., 1948. Forest Influences. McGraw-Hill, New York, 394 pp.
Law, F., 1956. The effect of afforestation upon the yield of water catchment areas. J. Br.
Waterworks Assoc., 38: 489-494.
Liebscher, H., 1972. Results of research on some experimental basins in the Upper Harz
Mountains. IAHS Publ., 97:150 162.
Likens, G.E., Bormann, F.H., Pierce, R.S., Eaton, J.S. and Johnson, N.M., 1977. Biogeo-
chemistry of a Forested Ecosystem. Springer-Verlag, New York, 146 pp.
Malpas and Watson, 1991. Technology and Wealth Creation. British Association.
Meadows, D.H., Meadows, D.L., Randers, J. and Behrens, W.W., 1972. The Limits to
Growth. Earth Island, London, 205 pp.
Natural Environment Research Council (NERC), 1975. Flood Studies Rep., 5 vols. NERC,
London.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. Proc. R. Soc.,
London, Ser. A, 193:120 146,
Penman, H.L., 1959. Notes on the water balance of the Sperbelgraben and Rappengraben.
Mitt. Schweiz. Anst. Forst. Versuchswes., 35:99 109.
Penman, H.L., 1963. Vegetation and hydrology. Tech. Comm. 53, Commonwealth Bureau of
Soils, Harpenden, 124 pp.
Penman, H.L., 1967. Discussion to paper by C.L. Wicht - - The validity of conclusions from
South African multiple watershed experiments. In: W.E. Sopper and H.W. Lull (Editors),
International Symposium on Forest Hydrology, Penn State University, 29 August- 10
September 1965. Pergamon Press, Oxford, p. 760.
Pereira, H.C. (Editor), 1961. E. Afr. Agric. For. J., Special Issue, 27:1 131.
Pereira, H.C., 1992. Keynote paper. 10th World Forestry Congress, Paris, 1991, Proc. 3,
pp. 139 150.
Perrault, P., 1674. De l'Origine des Fontaines. Pierre le Petit, Paris, 229 pp.
Pierce, R.S., Hornbeck, J.W., Likens, G.E. and Bormann, F.H., 1970. Effect of elimination of
vegetation on streamwater quantity and quality. IAHS Publ., 96:311-328.
Polcher, J. and Laval, L., 1993. A statistical study of regional impact of deforestation on
climate of the LMD GCM. Clim. Dyn., submitted.
Rakhmanov, V.V., 1962. Role of forests in water conservation. Translated from Russian by A.
Gourevitch and L.M. Hughes. Israel Program for Scientific Translations, Jerusalem, 1966,
192 pp.
Roald, L., Nordseth, K. and Hassel, K.A., 1989. FRIENDS in Hydrology. IAHS Publ., 187:
1 491.
Robinson, M., 1986. Changes in catchment runoff following drainage and afforestation. J.
Hydrol., 86:71 84.
Robinson, M., Gannon, B. and Schuch, M., 1991. A comparison of the hydrology of moorland
under natural conditions, agricultural use and forestry. Hydrol. Sci. J., 36: 565-577.
Rosenqvist, I.Th., 1978. Acid precipitation and other possible sources for acidification of rivers
and lakes. Sci. Total Environ., I0:39 49.
Rutter, A.J., 1967. An analysis of evaporation from a stand of Scots pine. In: W.E. Sopper and
H.W. Lull (Editors), International Symposium on Forest Hydrology, Penn State University,
29 August I0 September 1965. Pergamon Press, Oxford, pp. 403 417.
Shuttleworth, W.J., 1988. Evaporation from Amazonian rainforest. Proc. R. Soc. Lond. Ser. B,
233:321 346.
Shuttleworth, W.J. and Calder, I.R., 1979. Has the Priestley-Taylor equation any relevance to
forest evaporation? J. Appl. Meteorol., 18:639 646.
216 J.S.G. McCulloch, M. Robinson/Journal of Hydrology 150 (1993) 189-216

Sklash, M.G. and Farvolden, R.N., 1979. The role of groundwater in storm runoff. J. Hydrol.,
43: 45-65.
Sopper, W.E. and Lull, H.W. (Editors), 1967. International Symposium on Forest Hydrology,
Penn State University, 29 August 10 September 1965. Pergamon Press, Oxford, pp. 1 813.
Stewart, J.B. and Thorn, A., 1973. Energy budgets in pine forest. Q. J. R. Meteorol. Soc., 99:
154-170.
Swank, W.T. and Crossley, D.A. (Editors), 1988. Forest Hydrology and Ecology at Coweeta.
Ecological Studies 66. Springer-Verlag, New York, 469 pp.
Whitehead, P.H. and Calder, I.R. (Editors), 1993. The Balquhidder Experimental Catchments.
J. Hydrol., 145: 215-480.
Wilm, H.G. and Dunford, E.G., 1948. Effect of timber cutting on water available for stream-
flow from a Lodgepole Pine forest. US Dep. Agric. Tech. Bull., 968, 43 pp.
Zon, R., 1927. Forest and Water in the Light of Scientific Investigation. US Government
Printing Office, Washington, DC, 106 pp.

You might also like