Water circulation in karst and determination of catchment areas: example of the River Zrmanja

Hydrological Sciences Journal ISSN: 0262-6667 (Print) 2150-3435 (Online) Journal homepage: http://www.tandfonline.com/loi/thsj20 Water circulation in karst and determination of catchment areas: example of the River Zrmanja OGNJEN BONACCI To cite this article: OGNJEN BONACCI (1999) Water circulation in karst and determination of catchment areas: example of the River Zrmanja, Hydrological Sciences Journal, 44:3, 373-386, DOI: 10.1080/02626669909492233 To link to this article: https://doi.org/10.1080/02626669909492233 Published online: 25 Dec 2009. Submit your article to this journal Article views: 322 View related articles Citing articles: 30 View citing articles Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalinformation?journalcode=thsj20

Hydrological Sciences Journal des Sciences Hydrologiques, 44(3) June 1999 373 Water circulation in karst and determination of catchment areas: example of the River Zrmanja OGNJEN BONACCI Civil Engineering Faculty, University of Split, Matice hrvatske str. 15, 21000 Split, Croatia Abstract Karst hydrological investigation of the sinking stream problem of the River Zrmanja is presented. The aim of this analysis is to assess the feasibility of constructing three hydroelectric power plants (HEPP) along the River Zrmanja course. This paper presents a suitable and simple hydrological methodology that can be applied to scarce available data obtained on complex karst terranes. The paper presents a complex but common case of water circulation in a karst system. The primary objectives of the investigations were: (a) to analyse the underground karst connections, (b) to analyse discharge conditions along the River Zrmanja, and (c) to define variations in the catchment area along the River Zrmanja. The fact that the hydrological regime of the River Zrmanja is highly variable, due to the water losses along the open streamflow, strongly influenced the selection of the locations and heights of the HEPP dams. In spite of many hydrological, meteorological and hydrogeological measurements, the River Zrmanja catchment is insufficiently gauged. This dictates the use of a simple, empirically-based hydrological methodology. The Turc (1954) and Coutagne (1954) formulas were used in determination of annual total runoff. Using these simple hydrological methods, some important engineering answers were obtained. This is a first step towards application of sophisticated hydrological models, needing large amounts of reliable data. La circulation de l'eau dans le karst: l'exemple de la Rivière Zrmanja Résumé On expose les recherches menées sur l'engouffrement de l'eau dans le karst sur l'exemple de la Rivière Zrmanja. Le but final de ces recherches est d'évaluer la possibilité de construction de trois usines hydroélectriques le long du lit de la Rivière Zrmanja. On utilise la méthodologie hydrologique simple, quelle il est possible d'appliquer dans le cas où l'on dispose de peu de données dans un environnement karstique complexe. On expose le cas complexe mais fréquent de la circulation de l'eau dans le karst, où il a été analysé: (a) les liaisons karstiques souterraines, (b) les modifications de l'écoulement le long du cours de la Zrmanja, et (c) la définition des modifications des bassins versants le long du cours de la Zrmanja. Le fait que le régime hydrologique de la Rivière Zrmanja soit fort variable à cause d'un engouffrement des eaux le long du lit, est prépondérant dans le choix de l'emplacement et de la hauteur des barrages. Malgré les nombreuses mesures hydrologiques, météorologiques et hydrogéologiques effectuées, le bassin versant de la Rivière Zrmanja est insuffisamment connu, ce qui impose l'utilisation d'une méthodologie hydrologique simple, basée sur l'expérience. Les formules de Turc (1954) et de Coutagne (1954) ont été utilisées pour l'estimation de l'écoulement total annuel. Le recours à ces méthodes hydrologiques simples donne néanmoins des réponses significatives du point de vue de l'ingénieur. Il s'agit d'un premier pas avant l'utilisation de modèles hydrologiques numériques complexes, nécessitant un grand nombre de données hydrométéorologiques fiables. INTRODUCTION One of the characteristic features of open streamflows in karst is that they either completely sink into the underground or have partial water losses along their course Open for discussion until 1 December 1999

374 Ognjen Bonacci (Bonacci, 1987). These losses can be important in particular river sections, whereas in other places they are small and difficult or even impossible to observe without performing special simultaneous discharge measurements (Bonacci, 1987). The losses depend on hydrogeological-hydrological conditions and especially on the groundwater level (GWL) which exhibits rapid spatio-temporal variation. The water losses through the river bed are significantly affected by silting up of the wetted area of the stream channel. In addition to the water losses, groundwater recharge occurs along particular sections. This recharge often comes from distant karst regions outside the topographic catchment area. Determination of exact catchment boundaries and the contributing areas of springs and rivers in karst is a difficult problem due to strong, direct, complex and often unpredictable interactions between groundwater and surface water flowing from different karst regions (Bonacci, 1988). In this paper hydrological methods based on annual water balance will be used to estimate the relationships between recharges and losses in the River Zrmanja catchment (Bonacci, 1985b; Grgic, 1987). Investigations were carried out to study the feasibility of constructing three hydro-electric power plants (HEPP) on the River Zrmanja (Bjedov, 1995). Respective dams are located near the Prevjes and Zegar gauging stations and about two kilometres downstream of the River Krupa mouth (Fig. 1). The primary objectives of investigations were: (a) to analyse the underground karst connections; (b) to analyse discharge conditions along the River Zrmanja; and (c) to define variations in the catchment area along the River Zrmanja caused by karst. Combined with the results of geological, hydrogeological, geophysical, speleological and other investigations, results obtained from hydrological investigations will help to determine the definite position and height of HEPP dams along the River Zrmanja. BASIC CHARACTERISTICS OF THE STUDY AREA The River Zrmanja catchment is located in the central part of the Dinaric karst region of Croatia (Fig. 1(a)) between 44 00' and 44 25'N and 15 35' and 16 20'E. The exact hydrological catchment area and boundaries are not known, although numerous investigations and groundwater tracings have been carried out. Figure 1 presents the topographic boundary of the River Zrmanja catchment and surrounding rivers. Figure 1 suggests that water from the Gracac plateau might have direct connection with the River Zrmanja. Numerous tracer tests showed that the upper Gracac plateau is indeed a part of the River Zrmanja hydrological catchment. Figure 2 presents the cross-section A-B-C, shown on Fig. 1, and suggests that the River Zrmanja is connected with the River Krka. Groundwater levels drawn in Fig. 2 had to be estimated in spite of the fact that as many as 20 piezometers were drilled in the Zrmanja catchment. Even this relatively large number of piezometers was not sufficient to obtain a precise delineation of the karst aquifer in the deep and well developed Dinaric karst. The piezometers are grouped in the river valley. They were drilled with the intention of investigating water losses from reservoirs. Due to the great

Water circulation in karst and determination of catchment areas 375 Legend Permanent karst spring Q Swallow hole (ponor) Connections between ponor and spring o-topographic basin limit - Cross-section (A-B-C) «River flow direction 1) Discharge gauging station Lu Precipitation gauging station Fig. 1 Location maps indicating (a) the study area, and (b) rivers, karst springs, swallow holes, gauging stations and topographic basin limits. variability of measured GWL values, these data were futile as far as aquifer geometry determination was concerned. The principles underlying this behaviour are explained by Drogue (1980). The total length of the River Zrmanja from the Zrmanja spring (Fig. 1) to its mouth in the Adriatic Sea is 69 km. Figure 3 presents a longitudinal cross-section of the Zrmanja with the location of the six gauging stations. The "suspended" section, with a length of about 25 km, is also designated. It can be seen that three subsections with swallow holes and water losses and four subsections without water losses exist in this "suspended" section. This is confirmed by simultaneous discharge measurements. During the driest years the River Zrmanja dries up for a maximum length of 18 km.

376 Ognjen Bonacci,6RACAC PLATEAU SUPPOSED GROUND WATER LEVEL Fig. 2 Cross-section A-B-C (from Fig 1 ). KRKA RIVER CATCHMENT 4 r-i 1000-900 - 800-700 600 500 «00-300 - _; [ma. s ALTIT UDE nrt^rrrrr^. ' vr»rrm 200 100 Y J75ma,s.(, H [ma.s.l. ] -300-200 100 -J- \ \ RIVER SUBSECTION WITH SPRINGS D> RIVER SUBSECTIONS WITH SWALLOW HOLES r. SUSPENDED" SECTION Legend Bottom of the Zrmanja River bed with permanent surface flow Bottom of the Zrmanja River bed with temporary Q surface flow Gauging station O Groups of piezometers Fig. 3 Longitudinal cross-section of the River Zrmanja with indication of gauging stations and groundwater level. There are eight precipitation gauging stations and two meteorological stations located in and near the River Zrmanja catchment. Table 1 presents the main characteristics of the precipitation gauging stations located in and near the River Zrmanja catchment in the 1953-1990 period. The relationship between mean annual air temperature, t, in C and altitude, H, in m a.s.l. defined for this part of the Dinaric region (Bonacci, 1985b) is given by following equation: / = 15.2-0.012ff+ 4.28 x 10" H 2 (1) Average annual temperature in the study area varies from 9 C (the Gracac plateau) to 14 C (mouth of the River Zrmanja at the Adriatic Sea coast). The mean annual rainfall varies from 1100 to 2100 mm with an average of about 1600 mm. The climate

Water circulation in karst and determination of catchment areas 377 Table 1 Main characteristics of the precipitation gauging stations located in and near the River Zrmanja catchment in 1953-1990 period (Fig. 1). Serial number of gauging station Stationnante Altitude Mean annual precipitation (m a.s.l.) (mm) 2 3 4 5 6 7 8 9 Obrovac Raduc Zegar Brnicevo Graiac* Mokro Polje Velika Popina Zrmanja Vrelo Plavno 57 600 60 610 560 200 650 300 443 1200 1950 1350 2100 2035 1080 1850 1550 1360 JO ^ Knin* 234 1100 ^ * meteorological station. is continental but influenced by the Adriatic Sea. The summers are very hot with air temperature as high as 35 C, whilst the winters are cold with temperatures as low as -10 C (especially on the Gracac plateau). The Zrmanja catchment is underlain by Lower Triassic dolomites and limestone, Cretaceous carbonates, dolomite breccias and Quaternary alluvial deposits. A multitude of fractures, sinkholes and caves in a well developed subcutaneous zone allow rainfall falling on the catchment to move rapidly into subterranean karst passages (Fritz, 1972). Because of this the average annual total runoff coefficient is greater than 0.5 and ranges from 0.4 in dry to 0.8 in wet years. The annual total runoff coefficient is defined as the relationship between total annual runoff at an investigated point and annual precipitation falling on its catchment. There are four large poljes in the karst and many small and medium depressions (from 0.1 to 2 km 2 ), covered with arable and non-permeable soil, mostly terra rossa. Polje in the karst represents a depression with relatively gently sloping bottom from the spring zone to the swallow-hole zone. The polje is a part of the karst where overland flow exists (Bôgli, 1980; Bonacci, 1987; Ford & Williams, 1989). These four poljes represent the only areas in the catchment where living conditions are favourable for human habitation and where intensive agricultural production is possible. Table 2 presents data on the mean monthly and annual water temperature measured at two gauging stations, Palanka and Jankovica Buk (see Fig. 3 and Table 3), and air temperature of the Zrmanja catchment in the period 1953-1990. From these data it can be noted that, while air temperature affects water temperature very strongly at the Jankovica Buk, at Palanka station its influence is almost negligible, indicating that Palanka station receives mainly groundwater runoff. The Palanka station and the Jankovica Buk station are respectively 10 km and 56 km distant from Zrmanja spring. Since 1955 numerous underground connections among swallow holes and springs have been determined using fluorescein tracing. Hydrogeological connections between the Gracac plateau and the spring zones along the north bank of the River Zrmanja are direct and complete. As a result, almost 100% of the fluorescein placed in the swallow holes on the Gracac plateau appeared in the springs of the Zrmanja's tributaries, the Krupa and Dobarnica rivers (Fig. 1), or in the spring zones near the River Zrmanja bed.

378 Ognjen Bonacci Table 2 Mean monthly and annual water and air temperatures (1953-1990). Period Water temperature, t ( C) Mean catchment air temperature, t ( C) Palanka Jankovica Buk January February March April May June July August September October November December Year 6.1 6.2 7.4 8.7 9.5 9.9 11.6 11.5 9.2 8.4 7.9 7.0 8.6 7.9 9.0 9.5 11.6 13.6 15.2 19.0 18.0 14.8 13.0 11.0 8.7 12.6 4.9 8.0 11.6 16.4 20.2 22.3 21.9 18.2 13.4 8.8 5.4 12.9 Flow velocity varies from 5 to 50 cm s" 1, which is typical of the Dinaric karst (Bonacci, 1987; Bonacci & Magdalenic, 1993). The difference in altitude between the swallow holes in the Gracac plateau and the springs in the River Zrmanja topographic catchment is on average 500 m. Special interest exists for any underground connection between the Zrmanja and Krka rivers due to the possibility of construction of a HEPP in the "suspended" section of the River Zrmanja (see Fig. 3). Figure 4 represents fluorescein concentration hydrograph measured on the Miljacka spring during November 1985. The dye was injected on 7 November 1985 in a swallow hole near the Mokro Polje gauging station and emerged after 252 h in the Miljacka spring. The flow velocity for the first arrival was 1.2 cm s" 1. Detailed analyses using discharges and water volumes were not 19 ' 20 ' 21 ' 22 ' 23 ' 24 ' 25 ' Nov. 1985 Fig. 4 Fluorescein concentration hydrograph measured at the Miljacka spring in the River Krka catchment.

Water circulation in karst and determination of catchment areas 379 possible due to the complex hydrological conditions and the fact that only a small part (about 30%) of the fluorescein was recovered. The existence of two peaks indicates the complexity of the underground connection between the swallow hole and the spring. Tracings were carried out during a long-lasting dry period in December 1986 and January 1987 when the River Zrmanja was partially dried up. The fluorescein dye was injected into small ponors in the river bed near the Ervenik station. All of the dye emerged 5 km downstream in the River Zrmanja subsection with springs (see Figs 1 and 3). The flow velocity was relatively slow, about 0.4 m s 4, mainly due to the extremely dry conditions. Investigations and measurements should be continued. The GWL measurements were carried out on the "suspended" section with the three groups of piezometers shown in Fig. 3. The first group was located near the Prevjes station and had nine piezometers, the second group (with three piezometers) was located near the Ervenik station and the third group (with eight piezometers) was located near the Zegar station. It was not possible to construct the complete flow system for the karst aquifer underlying the River Zrmanja using the data obtained from these piezometers. The maximum decrease rates in GWL are not great and amount to about 0.5 m day" 1. The maximum amplitude of the GWL ranges from 7 to 36 m and the maximum rate of increase after heavy rains varies from 1 to 5 m day" 1, which is low in comparison with other regions of the Dinaric karst (Bonacci, 1987, 1995; Bonacci & Zivaljevic, 1993). Using GWL measurements (Borelli, 1966; Bonacci, 1995), the effective porosity, n e, is estimated to be about 0.5%. All of the above mentioned parameters and their values proved that underground capacity for water retention on analysed section is not remarkable. THE DISCHARGE CONDITIONS ALONG THE RIVER ZRMANJA The hydrological regime of open streamflows in karst depends mainly upon the interaction between groundwater and surface water. Some basic hydrological data for the six gauging stations along the River Zrmanja are given in Table 3. The topographic areas are defined using surface morphology. The data exhibit how the karstification of the catchment affects the average annual discharges. It can be seen that the hydrological behaviour on the "suspended" section is anomalous. Total mean annual discharge at the downstream Ervenik station is about 10% less than at the upstream Palanka station. Table 3 Main characteristics of the gauging stations along the River Zrmanja for the 1953-1990 period (Fig- 3). Station number 1 2 3 4 5 6 Station name Palanka Prevjes Mokro Polje Ervenik Zegar Jankovica Buk Datum plane, H (ma.s.l.) 254.90 225.61 195.61 121.55 50.38 2.93 Distance from spring, L (km) 10 14 20 30 41 56 Topographic area. A T (km 5 ) 156 183 228 296 382 666 Total mean annual discharge, Q (m 3 s' 1 ) 5.2 5.4 5.0 4.8 9.8 39.0

380 Ognjen Bonacci 0 -j 1 1 < 1 H H I H I! 1 1 1 1 t J 1 p> 0 1 2 3 4 5 6 7 S 9 10 11 12 13 U 15 16 17 18 DISTANCE ALONG THE ZRMANJA RIVER (km) Fig. 5 Graphical presentation of simultaneous discharge measurements taken out along the River Zrmanja from Prevjes to Ervenik (Grgic, 1987). Figure 5 shows the results of one of the many sets of simultaneous discharge measurements made on the river section between the Prevjes and Ervenik stations. The example presented confirms the variations in the behaviour of the losses along the natural streamflow. The largest losses occur in the subsection Prevjes-Mokro Polje and amount to 0.4, 0.7 and 2.5 m 3 s' 1 in the 1953-1990 period for minimum, mean and maximum discharges respectively (Grgic, 1987). Losses on the subsection Mokro Polje-Ervenik exist only during low water and vary from 0.3 to 0.8 m 3 s" 1. The average daily discharges jq i+ \ at the downstream station of section / for day j are plotted against the difference in discharges between the downstream and upstream sections jaqi (Fig. 6), where: AQ^JQM-JQ, (2) When the value of jaqi is negative, water is lost into the karst underground in section /. A positive sign defines "normal" hydrological circulation. The analyses indicate that, during such "normal" circulation along the "suspended" section, the river is in hydraulic connection with the aquifer. Analysing the results from Fig. 6, it can be seen that there are constant intersections of the curves of the relationship (Qi+\ - Q t ) with the ordinate axis, which do not vary significantly from one year to another. The intersection points have been named "limit discharges" (Bonacci, 1985a, 1987) since they represent important hydrological data. The area bounded by the curve and the ordinate axis is the negative domain (checked in Fig. 6) and is an indication of the quantity of losses. The losses are larger in the Prevjes-Mokro Polje subsection than in the Mokro Polje-Ervenik subsection. Special attention from a hydrological point of view should be paid to the analysis of the "limit discharges". These can be explained

Water circulation in karst and determination of catchment areas 381 3 4 5 6-1 0 1 2 3 AQ=Q 3 - Q 2 ( m 3/s) AQ = 0^ -Q 3 (m3/s) Fig. 6 Analysis of losses and recharges along the River Zrmanja from Prevjes to MokroPolje (1953-1990). by the fact that losses due to karstification either decrease, completely disappear or become smaller than the lateral intercatchment recharges when the discharge is greater than the limit discharge. The River Zrmanja dries up at only two of the six gauging stations analysed, i.e. at the Mokro Polje and Ervenik. On the Mokro Polje profile the River Zrmanja dried up in 42% of the years in the period 1953-1990. On the average it is dry on 24 days per year. On the Ervenik profile, the River Zrmanja went dry every year except for the wet 1978 year, on average 76 days per year. The year 1983 was especially dry, and on the Mokro Polje and Ervenik profiles there was no surface flow for 134 and 187 days respectively. It is uncommon for the river to go dry in the winter period; it occurred only twice during the 1953-1990 period. DETERMINATION OF THE CATCHMENT AREAS FOR SOME PROFILES ON THE RIVER ZRMANJA Everything previously discussed confirms the extremely complex and spatiotemporally variable hydrologic-hydrogeologic situation in the catchment of the River Zrmanja as a result of karstification. In this section the actual hydrological contributing areas, for the six gauging stations given in Table 3 and shown on Fig. 3, are defined. Using the precipitation data obtained at ten precipitation gauging stations the mean annual areal precipitation is determined by isohyetal method for the 1953-1990 period. The next step was to estimate the annual runoff deficit, D, expressed in mm, for each catchment area covered by the six chosen gauging stations (Table 3). The annual

382 Ognjen Bonacci total runoff deficit, D, is calculated using the Turc (1954) and Coutagne (1954) formulas. The Turc (1954) formula is: D = P/j0.9+ (P 2 /L 2 ) (3) where P is annual precipitation in the catchment area expressed in mm, and L is defined as: L = 300 + 25/ + 0.05/ 3 (4) where / is the average annual air temperature of the catchment area, expressed in C. The Coutagne (1954) formulas are: D = P~-P 2 X (5) A. = 1/(0.8 + 0.14?) for (0.125/X) < P < (0.5/A,) (6) D = P for P < (0.125/A.) (7) 25 = 0.2 + 0.035/ forp>(0.5/?0 (8) The parameters D, P and / are the same as in the Turc (1954) formula. The process consisted of determining the catchment area which satisfies the water budget equation: R = P-D (9) where R denotes total annual runoff. This runoff should flow out from the hydrological catchment area. As only the topographic areas, A T, for the six gauging stations along the River Zrmanja (Table 3) are known, the comparison with the measured runoff at the six stations reveals the relationship between the topographic, AT, and real hydrological, A, catchment areas. This serves as an indicator of the influence of karst on particular locations on the River Zrmanja. Table 4 presents the relationships between precipitation, P, and annual runoff, R, defined by the Turc (1954) and Coutagne (1954) formulas for six gauging stations for the 1953-1990 period. Those relationship are expressed by the linear equation: R = ap-b (10) where a and b are linear regression coefficients determined by the method of least squares. The coefficients of linear correlation, r, were defined for each case and were Table 4 Relationship between annual runoff R and annual precipitation P for six gauging stations along the River Zrmanja (1953-1990). Station number 1 2 3 4 5 6 Station name Palanka Prevjes Mokro Polje Ervenik Zegar Jankoviéa Buk Equation (10) [R-aP-- b] defined using Turc formula Coutagne formula # r =0.96P-426 R c =l.0\p-5)5 R T =0.96P~470 R c =\.0lP-542 #T-=0.97P~489 R C =\.01P~564 R T =0.9SP~4S5 /?c= 1.01P-574 J? r =0.96P-505 Re-0.9SP-537 tf r =0.96/>-512 R c = 1.01/»-578

Water circulation in karst and determination of catchment areas 383 always greater than 0.99. The values of R calculated by equation (10) present the expected mean annual runoff from the topographic catchment area without the influence of the karst. The influence of the karst on the runoff variability has been further investigated by establishing the difference between the runoff, Q, as defined by field measurement (the last column in Table 3) and the total runoff, R, defined by equation (10). This runoff, R, can be expressed as either effective annual precipitation in mm or as mean annual discharge in m 3 s" 1. Table 5 gives the average annual discharges (1953-1990) expressed in m 3 s" 1 defined by the Turc equation, R T, compared with measured discharges Q (Table 3). It should be stressed that in this case Rj is practically equal to Re and so the calculations with only one of Turc or Coutagne formulas is needed. Before discussing of the results in Table 5 it should be recognized that the accuracy of the methods used is estimated to ±10%, so more accurate results could not be expected. Previous results indicate that the topographic catchment areas, AT, for the gauging stations Palanka, Prevjes and Zegar can be accepted as the actual hydrological catchment areas. A negative discharge difference AQ = -2.2 m 3 s" 1 has been established on the Mokro Polje profile. It can be explained either by significant losses caused by water sinking through the swallow holes located on the wetted areas in the River Zrmanja "suspended" sector or by the fact that the topographically defined catchment is greater than the actual hydrological catchment. The first assumption, however, seems more probable. The difference established at the Ervenik gauging station was as much as AQ = -33 m 3 s" 1 which is an average 69% of the measured discharge. The situation found on the Mokro Polje-Ervenik subsection is even more evident than on the previous Palanka-Mokro Polje section. The groundwater there is essentially lower than the level of the River Zrmanja bed and the infiltration of surface water is high. Most of this water returns to the River Zrmanja just downstream from the Ervenik station and partly flows to the Miljacka spring in the River Krka catchment. At the Jankovica Buk station the established difference is Ag = +20.0 m 3 s" 1, which is 51% of the measured discharge. This quantity definitely indicates that the Jankovica Buk station receives recharge from an area greater than its topographic catchment area. This can be explained physically by the connection established with the Gracac plateau with a catchment area of 715 km 2. Using the relationship: Table 5 Comparison of measured Q and calculated,r r mean annual discharges (1953-1990) on six River Zrmanja gauging stations. No. Station name Measured Runoff defined by Discharge difference Ag Q discharge, g Turc equation, R T AQ = Q-R T ---100 (mv) (mv) ( rv) (0 } R r 1 Palanka ' 52 lo ' +02 +3^8 L04 2 Prevjes 5.4 5.3 +0.1 +1.9 1.02 3 Mokro Polje 5.0 7.2-2.2-44.0 0.69 4 Ervenik 4.8 8.1-3.3-68.8 0.59 5 Zegar 9.8 10.0-0.2-2.0 0.98 6 Jankovica Buk 39.0 19.0 +20.0 +51.3 2.05

384 Ognjen Bonacci (A:A T ) = (Q:R T ) (11) where A is actual hydrological area, A T is the topographic catchment area (both in km 2 ), Q is the measured runoff and R T is runoff defined by the Turc (1954) formula (both in m 3 s* 1 or in mm), it is possible to estimate A for the Jankovica Buk station as 1367 km 2. This value is greater than the topographic area of 666 km 2 (Table 3) by 701 km 2, that is by an area comparable with the hydrological catchment area of the Gracac plateau (715 km 2 ). The influence of snow cover on hydrological processes of the River Zrmanja catchment is important, but measured data do not exist. In this situation the Turc (1954) formula based on the annual hydrometeorological data is more convenient than other complex models. CONCLUSIONS This paper presents a hydrological analysis adapted to karst conditions. The basic time unit for analysis was one year. Referring to the computation of the runoff deficit, D, using Turc and Coutagne formulas, Réméniéras (1965) states that they are often quite precise. Although these expressions contain only the average annual rainfall in the catchment, the formulas (especially Turc's) have been applied worldwide with great success, particularly to ungauged catchments (de Montmollin et al., 1979; D'Oliviera & Mimoso, 1973; Rodier, 1983; Bonacci & Magdalenic, 1993). Turc (1961) developed equations to determine monthly and 10-day water deficit. This approach needs more data and is less reliable than the annually-based (equations (3) and (4)) model. Using the proposed procedure no reliable conclusions on seasonal and monthly hydrological fluctuations can be drawn. It seems that the catchment areas in welldeveloped karst regions, as in the River Zrmanja catchment, are not constant in time. They change depending upon the general water quantity situation, best represented by the GWL. In the River Zrmanja catchment case, the existing GWL data are not sufficient for drawing reliable conclusions. Consequently the obtained values for the hydrological catchments covering the area of 156 km 2 upstream of the Palanka station and the area of 1387 km 2 upstream of the Jankovica Buk station represent only average values for the mean annual hydrological state. Figure 7 gives a simple schematic conceptual presentation of how the River Zrmanja catchment hydrology and hydrogeology function. The Turc (1954) and Coutagne (1954) formulas do not guarantee a high degree of accuracy, but are quite applicable in engineering practice as reliable indicators. Engineers are often in situation where they must quickly solve similar complex problems. According to our experience, the method presented here could work effectively. Numerous hydrological, hydrometric, hydrogeological and speleological investigations were carried out during the period 1953-1990 to satisfy the requirements of the design and construction of three HEPPs in the course of the River Zrmanja. Most of these works have been described here from the hydrological standpoint with the aim of solving engineering problems. The analyses described in this paper stress the need for inter- and multi-disciplinary approaches incorporating several methods

Water circulation in karst and determination of catchment areas 385 GRACAC PLATEAU TOPOGRAFIC BASIN LIMITS. ZRMANJA CATCHMENT < LU to ZRMANJA..--7 R IVER ZRMANJA 5 Legend Q < MILJACKA SPRING r 1 KRKA SPRING Swallow hole # Permanent karst spring II1 River in the karst KRKA CATCHMENT Fig. 7 Schematic conceptual presentation of River Zrmanja hydrology and hydrogeology functioning. I Underground connection and techniques in the study of water circulation in karst. It can be stated that many dilemmas have been solved but that the investigation has pointed to the necessity of performing further measurements and analyses. REFERENCES Bjedov, T. (1995) Visenamjensko iskoristavanje akumulacijskih jezera rijeke Zrmanje (Multipurpose exploitation of reservoirs on the River Zrmanja, in Croatian). Gradevinar (Zagreb) 47(11), 679-684. Bôgli, A. ( 1980) Karst Hydrology and Physical Speleology. Springer Verlag, Berlin, Germany. Bonacci, O. (1985a) Hydrological investigation of Dinaric karst at the Krcic catchment and river Krka springs. J. Hydrol. 82,317-326. Bonacci, O. (1985b) The influence of karst on the water discharge in open streamflows-example of the Zrmanja catchment. Nas Krs XI( 18-19), 7-19. Bonacci, O. (1987) Karst Hydrology. Springer Verlag, Berlin, Germany. Bonacci, O. (1988) Determination of the catchment areas in karst. In: Karst Hydrogeology and Karst Environment Protection (ed. by Y. Daoxian) (Proc. IAH/IAHS Guilin Symn., China, October 1988), 606-611. 1AHS Publ. no. 176. Bonacci, O. ( 1995) Ground water behaviour in karst: example of the Ombla spring (Croatia). J. Hydrol. 165, 113-134. Bonacci, O. & Magdalenic, A. (1993) The catchment area of the Sv. Ivan spring in Istria (Croatia). Ground Water 31(5), 767-773. Bonacci, O. & Zivaljevic, R. (1993) Hydrological explanation of the flow in karst: example of the Crnoievica spring. J. Hydrol. 146,405^119. Borelli, M. (1966) O gubicirna vode iz kraskih akumulacija. Akumulacija Busko Blato (Water losses from karst reservoirs. Busko Blato reservoir, in Croatian). Saopstenja Inst. J. Cerni 36, 53-58. Coutagne, A. (1954) Quelques considérations sur le pouvoir évaporant de l'atmosphère, le déficit d'écoulement effectif et le déficit d'écoulement maximum. La Houille Blanche 6, 360-369.

386 Ognjen Bonacci de Montmollin, F., Olivier, R. & Zwahlen, F. (1979) Utilisation d'une grille d'altitudes digitalisées pour la cartographie d'éléments du bilan hydrique. (Use of a digitized grid in the mapping of the water balance in a river basin.) J. Hydro!. 44, 191-209. D'Oliviera, E. E. & Mimoso, J. J. (1973) Application of Coutagne's and Turc's formulas to southern Mozambique rivers. In: Design of Water Resources with Inadequate Data (Proc. Madrid Symp., June 1973), 141-154. UNESCO/WMO/IAHS Studies and Reports in Hydrology vol. 16, no. 2. Drogue, C. (1980). Essai d'identification d'un type de structure de magasine carbonats, fissurés. Mem. Hydrogeol. Ser. Soc. Geol. France II, 101-108. Ford, D. C. & Williams, P. W. ( 1989) Karst Geomorphology and Hydrology. Unwin Hyman, Winchester, MA, USA. Fritz, F. (1972) Razvitak gornjeg toka rijeke Zrmanje (Morphological evolution of the upper Zrmanja course). Carsus lugoslaviae 8/1, JAZU, Zagreb, 1-19. Grgic, B. ( 1987) Komponente povrsinskog i podzemnog bilansa voda rijeke Zrmanje od Prevjesa do Zegara. (Components of surface and underground water balance of the River Zrmanja from Prevjes to Zegar, in Croatian). Unpublished MSc Thesis, Zagreb Univ., Croatia. Réméniéras, G. (1965) L'Hydrologie de l'ingénieur. Eyrolles, Paris, France. Rodier, J. (1983) Hydrological computations for water resources development with inadequate data. In: Hydrology of Humid Tropical Regions (ed. by R. Keller) (Proc. Hamburg Symp., August 1983), 447^*58. IAHS Publ. no. 140. Turc, L. ( 1954) Le bilan d'eau des sols. Troisième journée de l'hydraulique, Alger, 36-43. Turc, L. (1961) Evaluation des besoins en eau d'irrigation; évapotranspiration potentielle. Ann. Agr. 12, 13-49. Received 15 April 1998; accepted 22 October 1998