Analysis of storm hydrograph and flow pathways using a three-component hydrograph separation model

Journal of Hydrology, 142 (! 993) 71-88

71

Elsevier Science Publishers B.V., Amsterdam

[ll

Analysis of storm hydrograph and flow pathways using a three-component hydrograph separation model O.O. O g u n k o y a a a n d A. J e n k i n s b

~Department of Geography, Obafemi Awolowo University, Ile-lfe, Nigeria hlnstitute of Hydrology, l~'allingford, OXIO 8BB, UK (Received 6 April 1992; revision accepted i July 1992)

ABSTRACT Ogunkoya, O.O. and Jenkins, A., 1993. Analysis of storm hydrograph and flow pathways using a three-component hydrograph separation model. J. Hydrol., 142: 71-88. A three-component hydrograph separation model involving full chemical-isotopic mass balance equations is presented and tested using storm runoff and catchment hydrological and hydrochemical data from Allt a Mharcaidh catch,ment in Scotland. A key issue in the use of the model is the rational and objective selection of representative end-members given spatio-temporal variation in catchment hydrochemistry. In addition, a highly involved catchment sampling strategy should be adopted so that no key runoff source is omitted. End-member mixing analysis was used in determining the set of storm runoffend-members used in the model. Results show that incident precipitation, soil water and ground water contributed 15%, 19%, and 66%, respectively,of the June 1989 storm runoff. Maximum instantaneous contributions were 26%, 40%, and 78%, respectively.

INTRODUCTION

Recent attempts at identifying sources of storm runoff have involved hydrograph separation using the chemical and isotopic mass balance equation:

QrCr = OiCI + Q2C2 + . . .

+ QmCn

(1)

where Cr, Ct, C2, . . . . Cnare the concentrations of solutes, electrical conductivity, pH, environmental stable and radio-isotopes of total storm runoff its assumed components. Qr, 01, 02, • • • , O, are the corresponding discharge rates. Correspondence to: O. Ogunkoya, Department of Geography, Obafemi Awolowo University, lle-lfe, Nigeria.

0022-1694/93/$06.00

© 1993 - - Elsevier Science Publishers B.V. All rights reserved

72

O.O. OGUNKOYA AND A. JENKINS

These studies, e.g. Sklash and Farvolden (1979), Rodhe (1987) and Bonell et al. (1990), have conceived storm runoff as having two distinct sources: 'old water' and 'new water', where old water is ground water and soil water that have been in catchment storage over time, while new water is the incident precipitation and includes direct channel precipitation, surface runoff and rapid response subsurface flow. The latter is grouped together with the others in new water because it was believed that only rapid throughflow which moved through macropores, and would therefore reflect incident precipitation chemistry contribute to storm runoff (Pilgrim et al., 1979). The aim of these studies was to separate the storm runoff hydrograph into its assumed components and from this, draw inferences on sources and hydrological pathways of the runoff. A number of conditions have to be satisfied in the use of this method, and these include the following: (1) The chemistry (solute and isotopic) of the waters from the various sources are distinguishable, i.e. each source has its own chemical identity; (2) the chemical identity is maintained in transit from source "reservoir' to stream channel, and is only changed during mixing in the channel, i.e. there is spatial and temporal uniformity in the identity of each 'end-member'; (3) the ground water and soil water, which together constitute old water, are chemically equivalent, or the soil water contribution to stream flow is insignificant. Most results obtained from the two-component hydrograph separation model indicate that old water is the significant component of storm runoff and that new water contributes less than 20%, even of instantaneous peak flow (e.g. Hooper and Shoemaker, 1986; Obradovic and Sklash, 1986; Pearce et al., 1986; Bishop and Richards, 1988). Some results, however, indicate that up to 60% of storm runoff could consist of new water (e.g. Bottomley et al.. 1984; Bonell et al., 1990). These results have been met with scepticism due to the extreme simplification of catchment prc~cesses by the assumptions of the model, and in particular, the limitation on the number of possible storm runoff endmembers. For instance, it has been shown that considerable spatio-temporal variations occur in the chemical-isotopic concentrations of both old and new water, especially the latter (Rodhe, 1987; McDonnell et al., 1990). In addition, the two sub-components of old water, ground water and soil water, may have different chemical identities. These limit accuracy of results and precision of identification and explanation of hydrological pathways, and therefore detract from the relevance of the model in explaining storm-runoff processes (Ogunkoya and Jenkins, 1991). Opinions have been expressed that an

ANALYSIS OF STORM HYDROGRAPH AND FLOW PATHWAYS

73

expansion of the components in the model to at least three, involving principally, distinctive groundwater and soil water components, would improve its capability (e.g. Bishop and Richards, 1988). A three-component model has been used by Dewalle et al. (1988) and Swistock et al. (1989), with the components being channel precipitation (i.e. rainfall intercepted directly by the wetted surface of the stream channel), soil water (i.e. near channel unsaturated zone water in soil macropores) and ground water (i.e. antecedent streamflow) presumably dominated by saturated zone water. Overland flow was assumed negligible due to the high infiltration capacities of the forested study catchment, and/or assumed to be of similar identity to incident precipitation. However, a full chemical-isotopic mass balance approach was not adopted as only one tracer 180 was used. Rather, the volume of direct precipitation in total streamflow was determined by measurement of throughfall intensity and simultaneous stream channel area, and the use of a least-squares procedure for analyzing direct precipitation accretion and routing through a channel. This assumes a certain source area expansion model which may be difficult to substantiate, given the nonuniformity of rainfall over a catchment, and the non-systematic variation of channel width with distance downstream due to variation in topography and geology. The results of Swistock et al. (1989) show that whereas ground water was still the dominant component of runoff, and accounted for 74-90% of total storm runoff volume and 62-97% of instantaneous flow, soil water was a major component for most storm runoff events, accounting for 6-25% of total storm runoff and up to 52% of instantaneous flow. Channel precipitation was a minor component, accounting for between 1 and 4% of total storm runoff and up to 14% of instantaneous flow. The present study elucidates a three-component hydrograph separation model involving full chemical-isotopic mass balance equations. The paper uses a storm runoff in AUt a' Mharcaidh catchment, Cairngorm Mountains, Scotland to demonstrate the operations of the model, and its capabilities in facilitating the drawing of inferences on hydrological pathways. The problems associated with the identification of storm runoff end-members to be used in the model are also reviewed. The assumptions underlying the use of this model are that: (1) Storm runoff has three end-members: (a) incident precipitation (p) which has at least two subcomponents, channel precipitation and overland flow, both having similar chemical identity; (b) soil water (sw) in the unsaturated zone of source areas; (c) ground water (gw). Channel precipitation is rain falling directly onto areas saturated to the surface, while overland flow is water flowing over such a surface or some other impermeable areas.

74

O.O. OGUNKOYA AND A. JENKINS

(2) Each of the end-members has a distinctive chemical-isotopic identity, especially in terms of the tracers used. (3) Temporal variations may occur in the concentration of end-members from beginning to end of storm runoff. This may be due to the Rayleigh distillation or rain-out effects in the case of precipitation, or entry of an end-member into the 'reservoir' of another, e.g. precipitation mixing with soil water, or ground water return flow mixing with soil water. The temporal variation at the sampling site is known. This assumption may introduce difficulties due to problems of determining the time synchronization of mixing in the reservoir and entry of the mixture into the channel. Ratio of mixture can only be estimated from streamflow chemistry. (4) Streamflow at inception of storm runoff, i.e. baseflow, could be a mixture of soil water and ground water, the ratio of m~xing depending on antecedent catchment wetness. STUDY AREA

The study catchment, Allt a' Mharcaidh, has an area of 9.98 km 2 and lies on the western edge of the Cairngorm Mountains, north-eastern Scotland. It consists of three distinct topographic units: a gently rising upland plateau which lies above 700m; steep valley sides lying between 550 and 700m; a gently sloping valley floor (below 550 m). The highest point in the catchment is at 11 i I m, while the outlet is at 320 m. The main catchment, G 1, comprises two smaller headwater catchments., G2 and G3 (Fig. 1). The catchment is underlain by biotite granite, but thick deposits of boulder clay derived from local rock cover much of the valley floor. Alpine podsols predominate on the plateau, peaty podsols on the valley sides and blanket peat up to 1 m thick on the valley floor. The peaty podsols have a weakly indurated horizon which promotes occurrence of a perched water table. Downslope drainage from the valley sides promote saturated conditions in the peat soil on the valley floor. The peat is characterized by the presence of pipes and incised channels eroded to the level of the mineral horizon. The vegetation consists of northern blanket bog on the valley floor, lichenrich boreal heather moor on the valley sides, and alpine azalea-lichen heather on the plateau. Some stands of mature native Scots pine (Pinus sylvestris) occupy the lower slopes of the catchment. The average annual precipitation is between 1100 m and 1500 m across the catchment, up to 30% of which may fall as snow. Rainfall-induced storm runoffs are most common in summer and autumn, while snowmelt events mainly occur during early spring, and rain on snow events occur during late spring and winter.

ANALYSIS OF STORM HYDROGRAPH AND FLOW PATHWAYS

75

ALLT a" MHARCAIDH

Geal-charr

0 ,

500 m ~

~

I..

lkm w

=,=,J

"Sgoran Dubh Mor

G Gaugingsites R Raingauges R3.

Fig. I. The Allt a' Mharcaidh catchment showing the location of monitoring stations.

METHODS

Streamflow was monitored at 20rain intervals at the rated outlet of the main catchment (G1) using Druck pressure transducers. Stream pH and conductivity were also monitored at 20-min intervals using a pHox 100 DPM. Rainfall was monitored at a network of five tipping bucket gauges

76

o . o . OGUNKOYA AND A. JENKINS

(R1,...,R5; Fig. 1). Rainfall at R2 closely simulated spatial average catchment rainfall. Storm runoff events were sampled hourly while rainfall was sampled at R I by a sampler which collected a discrete sample for each 1 mm of rainfall, and at R5, by a bulk collector which took rainfall samples over a 2h period. Bulk precipitation was additionally collected at four locations ( R I , . . . , R4) using 40cm diameter funnels. Four boreholes, BH2, BH23, BH24 and BH26 (Fig. 1) were sampled manually every 2 h during the early part of the storm event and less frequently towards the end. BH2 and BH26 were located close to the channel while BH23 and BH24 were located on the valley side. The water table at BH2 and BH26 was less than I m from the soil surface and about 3.5 m from the surface at the other two locations. Throughflow was sampled at two sites using gutter lysimeters. The first site was on the valley floor (PH, i.e. peat) and the second, on the valley side (PP, i.e. peaty podsol). Lysimeters were placed at the depths (50cm, 100cm) at each site, corresponding to PH2, PH4 and PP2, PP4. Samples were analyzed for deuterium (D%o) relative to SMOW at the Institute of Hydrology, Wallingford, using mass spectrometry (accuracy ___2%o), and for a variety of chemical determinants including Na, K, Ca, Mg, AI (total monomeric), NH4, CI, SO4, NO2 and total organic carbon (TOC), at the Freshwater Fisheries Laboratory, Pitlochry. Acid neutralizing capacity (ANC) was estimated using the relationship: ANC(/~Eq. 1-~) = ~ ( C a -~+ + Mg 2+ + Na ÷ + K ÷ + NH~-)

(2)

- ~ (NO3 + Cl- + SO 2-)

The three-component hydrograph separation model used two of the three conservative tracers determined: ANC (see Neal et al., 1991), chloride (see Neal et al., 1988), and deuterium (6D%o). It may be noted that ~n' tracers are needed to partition the hydrograph into (n + l) components. The linear equations used are derived thus: If 0~is the proportion of incident precipitation (p), fl is the proportion of soil water (sw), 7 is the proportion of ground water (gw), chloride and deuterium are the tracers, x is C1 concentration of stream water, y is 6D%o of stream water, A~, A 2, A 3 and Bt, B2, B3, are the corresponding Cl and 6D%o of incident precipitation, soil water and ground water, respectively, then: fl,4 2 + TA 3

(3)

Y = ~Bi + fiB, + 7B3

(4)

X

=

~A I +

B3) + ( y (A2 - A3)(B1 - B3) - ( B 2 - -(x

0~ =

-

A3)(B

2 -

B2)(A

2 -

B3)(AI

-

A3) A 3)

(5)

77

ANALYSIS OF STORM HYDROGRAPH AND FLOW PATHWAYS

= =

(x -

A3)(B,

(A21 -

A3)(B,ot - - fl

-

B3) -

ty -

t~3)(A,

-

B3)-

(B2- B3)(A~-

A3) A3)

(6) (7)

Two approaches were taken in the determination of realistic values of the As and Bs in eqns. (5) and (6), based on the assumption 3. One approach assumes constance in the identity of end-members over time, and the other, a temporal variation. In the former, soil water and ground water C1 and 6D%o immediately prior to the storm event were adopted as the A2, A3 and B2, B3, respectively. These values were taken as constant over the duration of the storm runoff. A~ and Bt, (i.e. precipitation C1 and ~D°/'oo, respectively) were taken as the volume weighted mean tracer concentrations of incident precipitation. A volume weighted mean, however, implies a temporally random variation of tracer concentrations around the weighted mean whereas in reality, the temporal variability may not be random given the Rayleigh distillation and rain out effects (see Ingraham and Taylor, 1986; McDonnell et al., 1990). Furthermore, such a weighted mean allows tracer concentrations of the last raindrops in a storm event to influence separation of hydrograph for the beginning of storm-runoff response, a consequence of which is an underestimation of the ground water and soil water components if the rainfall concentrations trend towards ground water and soil water concentrations. In the latter approach, trend lines (running means) were fitted on the ground water and soil water tracer concentrations from the background values to those at end of the storm runoff. Instantaneous tracer concentrations for any time during the storm runoff was estimated by direct interpolation from these trend lines (see Rodhe, 1987). A, and B~ were determined as incrementally adjusted weighted mean tracer concentrations (see McDonnell et al., 1990). This allows continuous adjustment of the tracer content of precipitation on the catchment based on the concentration and timing of each sample (see eqn. (8)): I1

A, = Z,=, ~A, Y.=,P, II

(8)

where A, is the weighted mean precipitation tracer concentration, P is the gross increment of precipitation with each sample, and A~is the corresponding tracer concentration. AI of the last precipitation sample is used for hydrograph separation for the post-precipitation period of storm runoff. This method takes into consideration the fact that incident precipitation at any time within a storm event will mix with some prior precipitation on the latter's path to the channel, leading to a possible change in chemistry of the

78

O.O. OGUNKOYA AND A. JENKINS

incident precipitation. However, the time synchronization of mixing on the catchment and entry into the channel is not known. The main advantage in this approach is that later rain in a storm event does not influence the results of analysis for an antecedent period. Given the number of points from where data describing catchment hydrochemical characteristics have been sampled, there is a need to determine which combination (rain, soil and ground water) of the sets of data best accounts for storm-runoff hydrochemistry. The end-member mixing analysis model ((EMMA), Christophersen et al., 1990), consisting of linear x - y plots or mixing diagrams of tracer concentrations of the three end-members considered, was used to select the set of end-members and validate their inclusiveness (see Fig. 2). If the end-members used in the plot constitute an exhaustive list of the sources of the storm runoff, stream flow tracer concentrations will plot totally within a triangle bordered by the end-member concentrations. Although Cl and 6D were chosen as tracers while rainfall at R1 and RS, soil water at PH2, and groundwater at BH26 were selected as end-members, Fig. 2 shows that this set of end-members are not all inclusive. Sampling missed a key contribution to stream chemistry, water having a low chloride content and depleted in deuterium. Non-inclusiveness of endmembers significantly detracts from accuracy of hydrograph separation, but this flaw is overlooked here as the main aim is to present the model and the problems associated with its use. The runoff hydrograph of the 13 June 1989 storm is analyzed, and results are presented in terms of instantaneous and total end-member contributions to storm runoff. RESULTS Figure 3 shows the pluviograph, hydrograph and the associated chemistries of the storm runoff, while Fig. 4 shows the 'separated' hydrograph ('fixed' end-member chemistry model). The storm runoff was in response to a 26 mm catchment average rainfall with a peak intensity of 6 m m h - ' , involved an increase in streamflow from 100Is-I to peak values of 6301s -~, and a total runoff oi" 3.4 mm with a runoff coefficient of 13.1%. The 'fixed' end-member chemistry model indicates that total ~ (proportion of incident precipitation) was 19%, corresponding to a runoff coefficient of 2.5%, while total fl (proportion of soil water) and 7' (proportion of ground water) were 28% and 53%, respectively. Instantaneous ~ range from 8% to 30% while those of fl and 7 range from 3% to 46%, and 30% to 87%, respectively. Contributions at peak discharge were 30%, 40%, and 30%, for 7, fl and ),, respectively. The highest ~ occurred during peak discharge periods, and were also associated with periods of peak rainfall intensities. The lowest

ANALYSIS OF STORM H Y D R O G R A P H A N D FLOW PATHWAYS

79

n,

o ,m(9

,

o. o_ ~

6

6q,

6(y

(I-O'~'oll~,) OplJOlq~

0

i

(I "11~srt'*~N~) ,(~loodo:~ 5UlOllO.LI.n®N plo'~

¢0 N ID

3 Io

oo

0

:~

a

.~

"~

-m

~

g

~

~"1-1'13¢n'+ONV) X4.1oodo:) OUlS!lOJ,~n®N p!~oV

r..

2

0

~

\\~o

13" 0 6 . 1 1 9

o

3

_

/

-

/

.

~

.

,

"~-.~

~

0 0

~

O0

0 30000

0 0

HOURS

_oo-'--.~o o oo o-,o

~

0

-.

! O

a.

0

.200Z

z

z

N

.>

z

z O ,.< >,

~J M

o o 0

Storm runoff S t r e a m drD, CI SOil w o t e r ~rD~CI G r o u n d w o t e r d ' D ~ CI R o l n dro t C I B u l k r o i n d ' D t CI

<

•4 0 0 3 hl (9

o • ÷DJ, ..... .600

~k-~

14.06.119

_.~

L6.00

i

~..oo=

14-06" 89

'PH;~,,4~6 ,pi.oz=,o z, 4 ~ e ,o,,2 ,,4

~.-_

/

0 0 W ~

9 9

Fig. 3. Pluviograph, h y d r o g r a p h and hydrochemistry.

O

g: Id I:3 ~-110

~-40,

~. - 2 0

0 J 1"

0

3

L I00 d @

""

g

J J

13,06-89

ANALYSIS OF STORM HYDROGRAPH AND FLOW PATHWAYS 13.06- 09

E

~,.... ,4

~

Q

ip

1.2 14

I~

81 14.08.89

I.o 29. 2is

Inoident p r e o i p i t Q t l o n

[

Soil wotor

aoo~

/A,

aE IEo

~ o : ) oc

Toc

eO0

D

_a4oo v

Id ¢f 2 0 0 Z U O

o

~, 4 ~ 13,,0G,09

e 1"O'12 ,~ le I~ 202~' .OUeS

0

~,

4

e

a

IO 12 I~.

14.0e. e 9

Fig. 4. Separated hydrograph (fixed end-member chemistry model).

occurred during lulls in rainfall and at the end of storm runoff. Instantaneous [~ increased from 3% at 8 h to 46% at 11 h, involving an absolute increase of approximately 1201s -~ . Changes in [1 generally reflected rainfall intensities. The highest "; (87%) occurred on the rising limb of the storm hydrograph (9 h), at a time when the water table showed an initial response to the storm event (see Fig. 6). Low ;, coincided with periods of high rainfall intensities. The 'temporally varying end-member chemistry' model (Fig. 5) indicates that total ~ was 15%, corresponding to a runoff coefficient of 2%, while/3 was 19%, and ),, 66%. Instantaneous ~ ranged from 12% to 26%, while those of I/and 7 were 15% to 40%, and 44% to 78%, respectively. This model, by adopting apparently more realistic values of end-member chemistries, appears to emphasize the relative importance of ground water contribution to storm runoff more than the 'fixed" model. Soil water exhibited some degree of temporal and spatial variation with the waters at the valley floor (PH2 and PH4), having lower pH, but richer in Na,

82

O.O. OGUNKOYA AND A. JENKINS

-- Soil woter 600'

•

ion

It

~. 4 o c O 0 4=

~= aoo 5

0 13. o e , e s

HOURS

14" Oe' 89

Fig. 5. Temporally varying end-member chemistry model.

~il

13.06,S9 ~ '4

I~ lIP,. ~

41'1 11~ ~11 lIP li~ 114"a~0%'8~

..oo

L. 1

~ 2,10

~" ~" ~" N . G ~ u n d woter at

St,20

f

600

.o|

I,~ Fig. 6. Ground water table response to rainfall, tFixed model.)

14.06.1ag

ANALYSIS OF STORM HYDROGRAPH AND FLOW PATHWAYS

83

K, Mg, C1 and TOC than the waters at the valley side (PP2 and PP4). Within the profile, soil water at the upper horizon had higher TOC and K, but lower AI. While there did not appear to be any difference in Cl with depth, chemistry at PH2 and PP4 reflected the diluting effect of rainfall. Thus, A1 ~'total monomeric) at PH2 declined from 87 to 46/tg l-1 between l0 h and 11 h, and from 114 to 23/tgl -I between 18h and 21 h (see Table 1), with periods of dilution following closely those of peak rainfall intensities, and thus suggesting entry of more dilute rainfall into soil water reservoir. Ground water chemistry also varied both in space and time over the catchment. For instance, mean alkalinity (ANC) was higher on the valley side (BH23 = 6 2 4 p g l - t ; BH24 = 512/tgl -~) than on the valley floor (BH25 = 433#g1-1; BH26 = 212/~gl -t) during storm runoff. Also, A1 increased from 30#gl -t at 8h to 108pgl-' at 11 h at BH2. During this time, alkalinity declined from 412/zg 1-, to 13/~g 1-1 . AI subsequently declined to 16 l~g 1- ~while alkalinity increased to 480/lg 1- ~at the end of the storm runoff. Similar variability occurred at BH23 and BH24, o n l y less pronounced. Chemistry at BH26 was less variable, and exhibited lower concentrations of ANC (63 _ 7 ttg 1-t ), CI (96 -t- 4/~g 1- t ) and AI (9 + 6/~g 1-t ) than at the other sites (see Table 2). Stream pH was inversely related to discharge, and involved a unit decline in pH between 8 h and 15 h (Fig. 3), during which period, instantaneous fl increased from 3% to 40% with a corresponding increase in absolute soil water contributions greater than an order of magnitude (~- 1601 s- ~). TOC and AI were positively correlated to stream discharge but with the relationship reflecting a slight 'exhaustion' effect over the various hydrograph peaks (Fig. 4). DISCUSSION

A three-component hydrograph separation model incorporating the 'fixed' or "time-invariant', and the 'temporally varying' end-member chemistry approaches, ' have been used to estimate incident precipitation, soil water and ground water contributions to a June 1989 storm runoff in the Allt a Mharcaidh catchment. The results presented can, however, only be considered suggestive, given that the data are not exhaustive of the sources of storm runoff in the catchment. The fixed end-member chemistry model indicates that total incident precipitation, soil water, and ground water contributions are 19%, 28%, and 53°/O, respectively. The corresponding contributions obtained from the temporally varying end-member chemistry model are 15%, 19% and 66%. The results suggest ground water is the dominant contributor to runoff,

0115 0615 0715 0815 0915 1015 1115 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 0000 0100 0200 0300 0400 0600

13.06.89

*ILEq. I l ~%o. bmg I I ~ltg i- i d p S c m i at 25°C.

14.06.89

Time

Date

3.98 4.05 4.02 3.94 4.09 4.00 4.01 4.08 4.02 4.00 4.03 3.92 3.94 4.06 4.02 4.04 4.06 4.03 4.01 3.91 4.05 4.02 3.95

pH

201 218 170 208 184 199 193 195 114 192 191 200 203 177 180 184 188 19i 188 202 211 265 261

Na*

Soil water chemistry at PH2

TABLE !

32 41 36 34 30 32 31 29 15 30 27 30 32 26 24 27 27 29 26 28 29 27 27

K* 3! 42 28 32 40 28 26 30 15 26 34 38 35 25 24 28 26 28 27 32 38 34 37

Ca* 55 62 45 58 50 52 50 49 24 48 50 54 55 41 42 44 45 43 44 50 52 57 53

Mg* ll0 141 99 117 101 106 96 116 102 102 !10 113 117 96 105 102 104 99 102 123 121 159 154

CI* 49 86 45 49 45 45 44 50 42 42 43 46 44 46 45 43 45 41 42 51 50 48 46

SO~ 0 39 14 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

NO~ 160 97 ll8 166 158 160 160 137 24 152 149 163 164 127 120 138 137 151 141 138 159 176 178

ANC*

-47 -45 -43 -45 -44 -45 -42

--49 -48 --47 -45 -46

-40 -47

-44

6D ~ 22 21 18 22 23 23 19 22 22 21 22 25 23 18 22 24 22 18 20 21 21 19 21

TOC b ll7 Ill 97 100 90 88 46 74 40 98 94 114 107 44 23 57 98 80 45 71 98 12 93

AI(TM) ~ 56 54 50 55 46 55 49 48 47 50 48 56 56 49 50 40 45 50 52 54 58 45 49

Conductivityd

z z

Z

z o >

ANALYSIS OF STORM H Y D R O G R A P H A N D FLOW PATHWAYS

85

TABLE 2

Bulk rainfall arid ground water chemistry (a) Rainfall 13.06.89 at R5 Time

Amount

pH

Na

K

Ca

Mg

CI

SO4

NO3

NH4

ANC

07:00 09:00 10:00 12:00 20:00

6.5 8.5 2.5 3.5 4.5

4.70 5.03 4.25 4.00 3.99

20 41 62 65 37

5 2 5 6 5

12 6 15 12 12

7 4 16 19 10

22 47 89 89 48

44 16 65 97 84

16

7 5 6 6 7

- 9 - 3 - 19 -30 - 19

26 51 60

(b) BH26 Date

Time pH

Na

01.6.89 17:00 7.11 131 12.6.89 12:35 6.75 122 13.6.89 08:00 6.78 131 09:00 6.86 124 10:00 6.69 126 I1:00 6.76 128 12:00 6.51 127 13:00 6.84 126 14:00 6.82 124 15:00 6.82 121 19:00 7.01 130 21:00 7.00 129 14.6.89 I1:00 6.97 134

K

Ca Mg CI

SO4 NO3 A N C AI(TM) 6D

Conductivity

10 9 15 8 10 12 10 9 8 8 9 9 12

49 48 59 43 45 48 52 48 47 44 59 45 52

51 51 50 50 51 48 50 49 51 49 52 50 48

39 37 43 31 31 35 48 56 36 34 39 39 47

26 24 25 22 23 25 26 24 25 23 26 25 25

93 96 103 94 96 98 96 93 92 91 95 94 100

0 0 2 1 1 3 3 2 I I I 2 I

72 "J 75 52 56 64 66 68 60 55 76 62 70

20 12 3 !1 1 13 23 7 12 0 8 9 3

-60 -59 -59 -57 -57 -61 -58

although the relative importance of groundwater declines during peak discharges. There was a relationship between the timing of peak rainfall intensities and maximum instantaneous incident precipitation contributions, The lags in the relationship appear to reflect the moisture status of the catchment and the surface wetting that occurred before significant contributions from incident precipitation would be attained. Minimum groundwater contributions were associated with periods of peak rainfall intensities, while maximum contributions were associated with the periods of peak water table elevations. The rapidity of ground water response to the storm event may suggest occurrence of ground water ridging or capillary zone effect (Sklash and Farvolden, 1979; Novakowski and Gillham, 1988), with the build-up of the ground water reservoir and hydraulic head effecting enhanced ground water flow into the channel. Entry of rain water into the channel during periods of peak intensity

86

O.O. OGUNKOYA AND A. JENKINS

would depress the relative contribution of ground water during such periods. There was also a close relationship between the periods of maximum dilution of soil water and those of maximum contributions of incident precipitation. This and the direct relationship between the contributions of incident precipitation and soil water may suggest that infiltrating rainfall caused a dilution of the soil water reservoir, and increased soil saturation levels and the soil water hydraulic head. These latter could promote saturation overland flow and increased soil water contributions. The incident precipitation contribution indicates that 2.5% of the catchment area contributed direct runoff to the channels. This value would be much higher than the total channel area in the catchment, and thus apparently includes riparian zones. The uncertainties in the analysis due to the spatial variations in endmember chemistry were assessed using alternative end-members (e.g. ground water at BH23 instead of at BH26). It was observed that contributions of incident precipitation, soil water and ground water at the inception of storm runoff could be overestimated by up to 15%, 50% and 20%, respectively. At peak discharge, incident precipitation and soil water could be overestimated by 39% and 40%, respectively, while ground water contribution would be underestimated by 35%. During streamflow recession, precipitation contribution could be overestimated by 18% while soil water and ground water contributions would be underestimated by 50% and l 1%, respectively. As pointed out earlier, a main issue in the use of the model is that of a rational and objective selection of representative end-members given the spatial and temporal variation of end-member chemistries. The choice of end-members was guided by both EMMA and the literature (e.g. ground water at BH26 was used partially because of the knowledge that groundwater oil the hill slope may not be representative of the conditions at the channel side: see Hill, 1990). The issue of the order of entry of these waters into the channel and the site of entry within the channel system were, however, unresolved. The temporally varying end-member chemistry model attempts to alleviate this problem, but the degree of success could not be ascertained. The difficulty of determining input times of various waters into the channel and in particular, the time synchronization between sampling at the various points on the catchment, and the time of entry of these waters into the channel, puts constraint on the validity of inferences that can be d~,awn, especially, on flow pathways. CONCLUSION

Information on contributions of assumed end-nlembcrs to storm run off has been determined using a three-component hydrograph separation model.

ANALYSIS OF STORM HYDROGRAPH AND FLOW PATHWAYS

87

The model could not, however, readily provide information on specific pathways such as whether ground water entered the channel as return flow or whether overland flow became influent just before reaching the channel. In general, it can be inferred from the results that at the early stages of the June 1989 storm runoff, streamflow consisted mainly of ground water and incident precipitation. With increasing catchment wetness, incident precipitation and soil water gained dominance contributing 30% and 40% of peak discharge, respectively. Groundwater regained dominance at the end of the storm runoff. These tend to suggest that the generating process and hydrological pathways of storm runoff in Allt a Mharcaidh could be explained using the Variable Source Area Model (Troendle, 1985). The model envisages storm-runoff contribution from a channel system expanding and contracting in response to rainfall, and throughflow and ground water contribution promoted through the development of hydraulic heads. ACKNOWLEDGM ENTS

The authors are indebted to the field sampling team; Ann Whitcombe, Steve Tuck, Angela Jenkins, Roger Wyatt, Dave Waters, and Alice Robson. Bob Ferrier and Bruce Walters from Macaulay Land Use Research Institute, Aberdeen, provided soil water samples. The work was funded under the Royal Society Surface Water Acidification Programme and the data analysis was completed whilst one of the authors (O.O.O.) was a visiting scholar at the Institute of Hydrology, funded by the Royal Society. London. REFERENCES Bishop, K. and Richards, K., 1988. A study of water flow pathways and resident times in the catchment of Loch Fleet using Oxygen-18 as a natural tracer. Loch Fleet Project Monograph--Report, April 1988, University of Cambridge, Cambridge, UK. 50 pp. Bonell, M.. Pearce, A.J. and Stewart, M.K., 1990. The identification of runoff production mechanisms using environmental isotopes in a tussock grassland ca~.chment, eastern Otago, New Zealand. Hydrol. Proe., 4: 15-34. Bottomley, D.J., Craig, D. and Johnston, L.M., 1984. Neutralization of acid runoff by groundwater discharge to streams in Canadian Precambrian Shield watersheds. J. Hydrol., 75: 1-26.

Christophersen. N., Neal, C. and Hooper, R.P, 1990. Modelling stream water chemistry as a mixture of soil water end-members---a step towards a second generation acidification model. J. Hydrol., 116: 201-207. Dewalle, D.R., Swistock, B.R. and Sharpe, W.E., 1988. Three component tracer model for storm flow on a small Appalachian forested catchment. J. Hydrol., 104: 301-310. Hill, A.R., 1900. Groundwater cation concentration in the riparian zone of a forested headwater stream. Hydrol. Proc., 4: ! 2 I-I 30. Hooper, R.P. and Shoemaker, C.A., 1986. A comparison of chemical and isotopic hydrograph separation. Water Resour. Res., 22(10): 1444-1454.

88

O.O. OGUNKOYA AND A. JENKINS

Ingraham, N.L. and Taylor, B.E., 1986. Hydrogen isotope study of large scale meteoric water transport in northern California and Nevada. J. Hydrol., 85: 183-197. McDonnell, J.J., Bonell, M., Stewart, M.K. and Pearce, A.J., 1990. Deuterium variations in storm rainfall, implications for stream hydrograph separation. Water Resour. Res., 21{3): 455-458. Neal, C., Christophersen, N., Neale, R., Smith, C.J., Whitehead, P.G. and Reynolds, B., 1988. Chloride in precipitation and streamwater for the upland catchment of R. Severn, midWales: some consequences for hydrochemical models. Hydrol. Proc., 2: 155-165. Neal, C., Robson, A. and Smith, C.J., 1990. Acid neutralization capacity variations for Hafren Forest Streams: inferences for hydrological process. J. Hydroi., 121: 85-101. Novakowski, K.S. and Gillham, R.W., 1988. Field investigations of the nature of water table response to precipitation in shallow water-table environments. J. Hydrol., 97: 23-32. Obradovic, M.M. and Sklash, M.G., 1986. An isotopic and geochemical study of snowmelt runoff in a small arctic watershed. Hydrol. Proc., I: 15-30. Ogunkoya, O.O. and Jenkins, A., 1991. Analysis of runoff pathways and flow contributions using deuterium and stream chemistry. Hydrol. Proc., 5: 271-283. Pearce, A.J., Stewart, M.K. and Sklash, M.G., 1986. Storm runoff generation in humid headwater catchments. I. Where does the water come from? Water Resour. Res., 22(8~: 1263-1272. Pilgrim, D.H., Huff, D.D. and Steels, T.D., 1979. Use of specific conductance and contact time relations for separating flow components in storm runoff. Water Resour. Res., 15 ~2~: 329-339. Rodhe, A., 1987. The origin of streamwater traced by ~80. Uppsala University, Department of Physical Geography, Division of Hydrology. Report Series A 41,260 pp. Appendix 73 pp. Sklash, M.G. and Farvolden, R.N., 1979. The role ofground water in storm runoff. J. Hydroi., 43: 45-65. Swistock, B.R., Dewalle, D.R. and Sharpe, W.E., 1989. Sources of acidic storm flow in an Appalachian headwater stream. Water Resourc. Res., 25 (10): 2139-2147. Troendlc, C.A., 1985. Variable source area models. In: M.G. Anderson and T.P. Burr {Editors), Hydrological Forecasting. Wiley and Sons, Chichester, pp. 347-403.