Você está na página 1de 9

ISSN 1068-3739, Russian Meteorology and Hydrology, 2010, Vol. 35, No. 12, pp. 842850.

Allerton Press, Inc., 2010. Original Russian Text L.S. Kuchment, A.N. Gelfan, V. N. Demidov, 2010, published in Meteorologiya i Gidrologiya, 2010, No. 12, pp. 7687.

A Spatial Model of Snowmelt-Rainfall Runoff Formation of the Mountain River (by the Example of the Upper Kuban River)
L. S. Kuchment, A. N. Gelfan, and V. N. Demidov
Water Problems Institute, Russian Academy of Sciences, ul. Gubkina 3, Moscow, 119333 Russia Received April 14, 2009 AbstractA spatial model is proposed of snowmelt-rainfall runoff formation of the mountain river enabling to take account of spatial inhomogeneity of the river catchment and vertical zoning of physiographic and meteorological conditions. The model describes the processes of snow cover formation at various altitudes and snow melting, infiltration into the soils, evaporation, and overland, subsurface and riverbed flows. The verification of the model was carried out from the observational data in the Kuban River basin up to the town of Armavir.

DOI: 10.3103/S1068373910120083

INTRODUCTION The main peculiarities of the formation of mountain river runoff are the vertical zoning of physiographic conditions at the catchment and the spatial inhomogeneity of its structure. Therefore, when describing the runoff formation processes, it is desirable to strive for the most possible detailing of spatial runoff formation picture. However, for the majority of mountain catchments a sparse network of hydrometeorological measurements is typical; that often results in the need to confine oneself to use the models with lumped parameters for the description of runoff formation processes where the meteorological impact on the catchment and its characteristics are presented in the form of the values averaged for the catchment area [1, 2, 6]. In many cases, such approach provides the acceptable results; however, it may lead to considerable errors if the hydrometeorological conditions differ considerably from the usual ones. In this paper, the model is proposed of runoff formation with distributed parameters enabling to take account of spatial variations of catchment characteristics and meteorological conditions; as a result, one can most fully use the available hydrometeorological and geographic information, in particular, the radar precipitation measurements and satellite data. The model is based on the finite-element schematization of catchment area and river basin network which takes account of the spatial variability of relief, main soils, vegetation cover, and the river network structure. The realization of the model and its verification were carried out by the example of the Upper Kuban River basin up to the town of Armavir (catchment area is 16900 km2). The computation of formation of the maximum snowmelt-rainfall flood for the period of observations which took place in the Kuban River basin in June 2002 was used as one of the model verification possibilities. CATCHMENT DESCRIPTION AND INITIAL DATA The Kuban River springs from Ullukam glacier on the western slope of Mount Elbrus at the altitude of 3080 m. The river feeding occurs mainly as a result of snow melting and liquid precipitation falling. The glacial runoff volume in the outlet of Armavir does not exceed 23%. In the area of the towns of Ust-Dzhegut and Nevinnomyssk, the Kuban River riverbed is diked and the Bolshoi Stavropol and Nevinnomyssk canals start here which get almost 50% of the runoff. About 10% of the Kuban River runoff up to Armavir is formed as a result of the inflow of return waters from irrigated fields situated downstream of Nevinnomyssk. The highest meteorological station in the Kuban River basin, Klukhorskii Pereval, is situated at the altitude of 2039 m and the significant part of the territory of the basin remains uncovered with meteorological observations.
842

A SPATIAL MODEL OF SNOWMELT-RAINFALL RUNOFF FORMATION

843

Fig. 1. The finite-element schematization of the Kuban River catchment.

The vertical zoning in the variability of soils and vegetation cover is clearly observed in the river basin. As the locality altitude increases, the chestnut chernozems change into mountain-forest brown soils and then into mountain-meadow soils. Up to the altitudes of 600800 m, the catchment area is completely ploughed up and is involved into the farming industry. The forest vegetation (mainly, beech, birch, and pine) is situated at the altitude up to 2000 m. In the highland areas, the vegetation cover is represented by the vegetation of subalpine, alpine, and nival belts. Owing to the vertical zoning of meteorological conditions, differences in the relief type, and vegetation cover changes, the formation processes of the snow cover and snow melting as well as of the water inflow into the river network differ considerably by intensity and duration in different parts of the Kuban River basin. In the upper catchment part, a considerable portion of rainfall precipitation is kept in the snow cover and the snow melting takes place mainly during daylight hours in the absence of cloudiness. The surface overland runoff is formed as a result of the snowmelt and rainfall water income to the watertight surface of rocks when the water income intensity exceeds the infiltration intensity. The water soaked into the soils is kept in small pores by capillary forces or flows down the large pores usually reaching shallow weakly permeable temporary aquiclude along which it moves in the horizontal direction in the form of subsurface overland runoff up to the riverbed network. A small portion of water soaks through the temporary aquiclude into deeper layers forming the ground runoff.

MODELING THE SNOWMELT-RAINFALL RUNOFF FORMATION OF THE KUBAN RIVER The finite-element schematization of the Kuban River catchment up to the outlet of Armavir (16900 km2) is demonstrated in Fig. 1. The main riverbed of the Kuban River and those of its seven main tributaries (Malyi and Bolshoi Zelenchuk, Urup, Teberda, Ullu-Khurzuk, Ullu-Kam, and Uchkulan) are divided into 64 linear riverbed parcels along which the riverbed runoff takes place.
RUSSIAN METEOROLOGY AND HYDROLOGY Vol. 35 No. 12 2010

844

KUCHMENT et al.

On the river slopes, 512 typical trapezoidal area elements are marked out with homogeneous type of soils, vegetation cover, and topography along which the one-dimensional runoff of surface and subsurface waters occurs. The slopes of area elements as well as their geometric characteristics were specified using the topographic map with the scale of 1 : 250000. The snow cover formation and its melting were computed using the model describing the variations of height and density of snow cover, ice and nonfrozen water content in it using the following equations:
dH s 1 -1 = r w [ X sr 0 - ( M + E s )(r i I s ) ] - V , dt d (r i I s H s ) = r w ( X s - M - E s ) + F, dt d (r w Ws H s ) = r w ( X l + M - E l - R w ) - F, dt

(1) (2) (3)

where Hs is the snow cover height; Is and Ws are the volume contents of ice and nonfrozen water in the snow cover, respectively; Xs and Xl are the intensities of the inflow of solid and liquid precipitation to the snow cover surface, respectively; r w, r i, and r 0 are the densities of water, ice, and newly fallen snow, respectively; Es is the sublimation intensity; El is the evaporation intensity of liquid water from snow; M is the snow melting intensity; F is the intensity of freezing of snowmelt water or liquid precipitation in the snow; Rw is the intensity of water loss in the snow cover; V is the intensity of snow cover compaction under influence of its own weight; t is the time. The snow compaction intensity (cm s1) was computed using the formula
V = v1r s H s2 , exp( v 2Ts + v 3r s ) 2

(4)

where r s is the snow cover density averaged by depth (g cm3), equal to r s = r i I s + r wW s ; v1, v2, and v3 are the empirical coefficients equal to [8, 12] 2.8 106 cm2 s1 g1, 0.08 C1, 21 cm3 g1, respectively. It was assumed that the sublimation occurs only in case of dry snow and in the presence of liquid water, only the evaporation takes place. To compute the evaporation El, the dependence was used
LE E l = k E d a r w Ws , rs riIs , rs

(5)

and to compute the sublimation intensity, the dependence was used


Ls E s = k E d a

(6)

where kE is the empirical coefficient. The snow melting intensity M was considered to be proportional to the positive air temperature Ta and was computed using the empirical melting coefficient b:
b(T - Tcr ), Ta > Tcr ; M = a Ta Tcr , 0,

(7)

where Tcr is the air temperature below which the snow melting does not occur. The intensity F of freezing of liquid precipitation or snowmelt water in the snow was computed using the formula
b |Ta |, Ta < 0C Ws > 0; F= Ta 0C Ws = 0. 0,

(8)

The water loss in the snow cover takes place only if the water content in the snow cover exceeds its water-retaining capacity and is determined as
Rw = Xl + M El wmax dH s , dt

(9)

where the volume water-retaining snow capacity was computed from the empirical formula
wmax = 0.11 0.11 rs . rw
Vol. 35 No. 12

(10)
2010

RUSSIAN METEOROLOGY AND HYDROLOGY

A SPATIAL MODEL OF SNOWMELT-RAINFALL RUNOFF FORMATION

845

The variations of air temperature with the altitude on the territory drawn to jth meteorological station were determined according to the linear dependence
Tj(z, t) = Tj(z0j, t) a(z z0j),

(11)

where a is the lapse rate; T(z0j, t) is the air temperature at the altitude of the jth meteorological station. The increase in the precipitation Rj(z, t) with altitude was taken into account using the linear dependence
Rj(z, t) = Rj(z0j, t)(1 + h(z z0)),

(12)

where h is the coefficient of precipitation increase with altitude; Rj(z0j, t) is the precipitation amount at the jth meteorological station. The dynamics of the content and vertical transport of moisture in the soils was computed using the diffusion equation of soil moisture
q q = - K (q ) , D( q ) t z z

(13)

where q is the volume soil humidity; D(q) is the diffusion coefficient; K(q) is the vertical hydraulic conductivity; z is the vertical coordinate. It was assumed that the subsurface runoff forms in the upper one-meter layer at the humidity being higher than the field water capacity, and below, the aquiclude is situated; the vertical moisture flow through it is equal to the hydraulic conductivity at the full saturation. The hydrophysical characteristics of soils K(q), y(q ), and D(q) were presented as the empirical dependencies proposed by Campbell [9] and including four constants: Ks is the hydraulic conductivity of soils at full saturation; q s is the soil porosity; y s is the barbotage potential; l is the index of pore size. At the upper soil boundary, the moisture flow was considered to be equal to
J = Jm = D(q) q + K for Jm Xl E; z z = 0

(14)

J = Xl E for Jm > Xl E,

where Jm is the infiltration intensity at maximum humidity of the upper soil layer. The intensity of soil moisture evaporation was computed using the formula
q0 - qr E = KE da (t) q - q r s ,

(15)

where da is the air humidity deficit; KE is the empirical coefficient; q0 is the humidity of upper soil layer; q r is the residual humidity. The finite-element method described in [3] was used for the numerical integration of Eq. (9). To describe the surface runoff, the equations of kinematic wave were used in the following form:
bs hs q s + = R e bs ; t x 1 1/ 2 5/ 3 qs = is hs bs , ns

(16)

where hs is the water flow depth; is is the slope angle; qs is the water discharge; bs is the width; ns is the roughness coefficient of Manning; Re is the exceeding of the intensity of the inflow of snowmelt waters and (or) liquid precipitation to the slope surface over the absorbing capacity of soils minus the evaporation loss. The subsurface runoff computations were carried out using the system of equations
( q s - q fc )bs hg t + q g x = R g bs ;

(17)

qg = K g i s h g b s ,

where q s and q fc are the maximum humidity and field water capacity of soils, respectively; qg and hg are the discharge and depth of subsurface runoff; Kg is the horizontal hydraulic soil conductivity. The intensity of the subsurface runoff feeding Rg was computed from the formula
RUSSIAN METEOROLOGY AND HYDROLOGY Vol. 35 No. 12 2010

846

KUCHMENT et al.

Fig. 2. The comparison of (1) computed snow cover thickness with (2) the actual one measured at the Klukhorskii Pereval meteorological station in winter (a) 1970/71, (b) 1971/72, (c) 1972/73, (d) 1973/74, (e) 1974/65, (f) 1975/76, (g) 1976/77, (h) 1977/78, (i) 1978/79, and (j) 1979/80.

Rg =
0

zs

( q - q jc )dz , t

(18)

where zs is the soil layer where q > q fc. It was assumed that the subsurface water runoff occurs at the same slopes and geometry of area elements as the surface runoff. The equations of kinematic wave describing the surface and subsurface runoffs were numerically integrated by means of the method of finite elements in combination with the method of Galerkin [4]. Unsteady water movement in the main riverbed of the Kuban River was computed using the advection-diffusion equation [5]
Q Dr 2Q 1 Q 2 + q, + = C t C x x

(19)

where Q is the water discharge; q is the lateral inflow per the unit of the riverbed length; x is the distance; C is the rate of propagation of flood waves; Dr is the diffusion coefficient. The parameter C was derived from the formula C = 1.5u, and the coefficient Dr, from the relationship
RUSSIAN METEOROLOGY AND HYDROLOGY Vol. 35 No. 12 2010

A SPATIAL MODEL OF SNOWMELT-RAINFALL RUNOFF FORMATION

847

Fig. 3. The dynamics of snow cover formation at the altitude of (1) 2000, (2) 2400, (3) 2800, and (4) 3200 m, respectively, computed from the data of Klukhorskii Pereval meteorological station (19721973); (5) the snow cover height measured at this station.

Dr =

Q , 2Bi 0

(20)

where B is the average river width; i0 is the average bottom slope; u is the average current speed. The numerical realization of advection-diffusion equation was carried out by means of the finite-element method described in [4]. For the tributaries of the Kuban River, the computations of unsteady water movement in the riverbed were carried out using the equations of kinematic wave. CONSTRUCTION AND VERIFICATION OF THE MODEL The construction and verification of the model were carried out from the diurnal data of meteorological observations of temperature, precipitation, and snow cover height at 11 meteorological stations (Armavir, Otradnaya, Nevinnomyssk, Kian, Cherkessk, Zelenchukskaya, Karachaevsk, Arkhyz, Uchkulan, Teberda, Klukhorskii Pereval), from the data of daily water discharges in the main riverbed of the Kuban River near the rural settlement named after Kosta Khetagurov and near Armavir and at three outlets on the tributaries: the Uchkulan RiverVerkhnii Uchkulan aul; the Ullu-Kam RiverKhurzuk aul; the Teberda River Teberda settlement, as well as from the data on the water intake to Bolshoi Stavropol and Nevinnomyssk canals and on the dumping of return waters to the Kuban River through the Barsuchki River for ten-day periods from 1971 to 1980. The calibration of model parameters of runoff formation was carried out from the data on the water discharge of the Kuban (at the outlets near the rural settlement named after Kosta Khetagurov and near Armavir), Uchkulan (Verkhnii Uchkulan), Ullu-Kam (Khurzuk), and Teberda (Teberda) rivers for the period from 1971 to 1975. The runoff formation at the Kuban River catchment was considered in spring-summer period from March 1 to September 30. The coefficients a and h in (11) and (12) were determined from the observational data at meteorological stations and turned out to be equal to 0.005 C/m and 0.001 m1, respectively. The melting coefficient b was selected by means of the comparison of the snow cover height values computed on the basis of the model of formation of snow cover and snow melting (1) with those measured at Klukhorskii Pereval and Teberda meteorological stations. In Fig. 2, the temporal course is demonstrated of measured and computed values of snow cover height for ten seasons at Klukhorskii Pereval meteorological station at b = 5.5 mm/( C day). As is clear from the figure, on the whole, a close correspondence of values is obtained. The model reproduces in a verisimilar way the dynamics of snow accumulation in different altitude zones where the snow cover measurements are absent (in Fig. 3, the computation is presented carried out from the data of Klukhorskii Pereval meteorological station for the period from July 1972 to May 1973). Hydrophysical characteristics of main soil types prevailing at the catchment were specified on the basis of the data cited in [7]. The field water capacity q fc of the chestnut chernozems, mountain-forest brown and mountain-meadow soils was assumed to be equal to 0.31, 0.22, and 0.26, respectively; the porosity q s, to 0.50, 0.43, and 0.44; and the residual humidity q r, to 0.15, 0.10, and 0.12. The barbotage potential y s (kPa) was determined using the empirical dependence [11]:
ys = 100(0.108 + 0.341qs),
RUSSIAN METEOROLOGY AND HYDROLOGY Vol. 35 No. 12 2010

(21)

848

KUCHMENT et al.

The ratio of mean square computation error s D of the daily water discharge at different outlets of the Kuban River and its tributaries to the mean square deviation sf of actual water discharges at these outlets the Kuban River the rural settlement named after Kosta Khetagurov 0.48 0.49 0.45 0.45 0.56 0.42 0.66 0.60 0.54 0.63 0.53 the Uchkulan RiverVerkhnii Uchkulan aul 0.47 0.47 0.40 0.48 0.48 0.44 0.70 0.53 0.47 0.54 0.50 the Ullu-Kam RiverKhurzuk aul 0.54 0.49 0.44 0.60 0.58 0.48 0.67 0.62 0.68 0.75 0.58 the Teberda RiverTeberda 0.54 0.45 0.33 0.57 0.47 0.67 0.49 0.40 0.52 0.49 the Kuban RiverArmavir 0.51 0.63 0.59 0.66 0.68 0.71 0.57 0.50 0.69 0.44 0.60

Year

1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 Average

and the index of pore size l, using the formula from [9]:
qs yr l = ln q ln y , r s

(22)

where the sucking pressure at the residual soil humidity yr was assumed to be equal to 1500 kPa. The value of hydraulic conductivity at full saturation Ks (cm/s) was derived from the empirical dependence from [10]: qs - qr . Ks = 108 (23) 2
ys

The hydraulic conductivity coefficients turned out to be equal to 0.0020, 0.0045, and 0.0035 cm/s, respectively. It was assumed that the ground runoff is represented in the lower part of the abatement branch of hydrograph and the value of hydraulic conductivity of weakly permeable layer at the lower soil boundary was determined by means of the coordination of observational data with the computation results of water discharge at this part of hydrograph. According to the calibration results, the value of horizontal conductivity Kg of different soil types turned out to be equal to 0.001 m/s. The portion of watertight areas in the highland region amounted to 0.2; the empirical coefficient KE at computations of evaporation from the soil surface turned out to be equal to 6.94 109 m s Pa; the roughness coefficients of Manning at computations of water movement along the slopes ns and along the riverbed network n, 0.2 and 0.05 s m1/3, respectively. The model verification was carried out on the basis of the data of water discharge measurements for the period from 1976 to 1980. In the table, the ratios are given of mean square computation error to mean square deviation of actual value of average diurnal water discharge for every year of calibration and testing periods. As is clear from the table data, on the whole, the runoff hydrographs for the Kuban River and its tributaries were computed with satisfactory accuracy. As an example, in Fig. 4, the comparison is demonstrated of measured and computed runoff hydrographs for the Kuban River near Armavir in spring-summer period of 19711980. MODELING THE SNOWMELT-RAINFALL FLOOD IN JUNE 2002 An attempt was made to reproduce the formation conditions and to compute the runoff hydrograph in the period of catastrophic flood in June 2002 as the estimation of the potential of developed model of the Upper Kuban River runoff formation. The interest in such computation is caused by the fact that the runoff formation conditions in that period differ considerably from those for which the model calibration and verification were carried out.
RUSSIAN METEOROLOGY AND HYDROLOGY Vol. 35 No. 12 2010

A SPATIAL MODEL OF SNOWMELT-RAINFALL RUNOFF FORMATION

849

Fig. 4. Hydrographs for the Kuban River near Armavir plotted from the (1) observational and (2) model data for (a) 1971, (b) 1972, (c) 1973, (d) 1974, (e) 1975, (f) 1976, (g) 1977, (h) 1978, (i) 1979, and (j) 1980.

The flood was caused by long-lasting shower-type precipitation which started on June 18 and occupied a considerable part of the territory of the Northern Caucasus. The maximum water levels of the Kuban River registered at different outlets in the period from June 20 to June 22 exceeded the maximum levels for the whole observational period. In particular, this exceeding amounted to 172 cm near Armavir. Unfortunately, for the part of the Kuban River catchment situated upstream of the outlet near the rural settlement named after Kosta Khetagurov, only the daily data on precipitation, air temperature, and humidity were available from observations at Klukhorskii Pereval meteorological station (the data for other meteorological stations (Teberda, Uchkulan, and Karachaevsk) were absent). In Fig. 5, the comparison is demonstrated of observed and computed hydrographs of the Kuban River near the rural settlement named after Kosta Khetagurov. As is clear from the figure, on the whole, the computed hydrograph is coordinated well enough with the one plotted from the observational data both by discharge value and by the dates of the peak values (the data on the water levels and discharges in the period of flood wave passage in the upper reaches of the Kuban River from June 20 to July 4, 2002 are absent). CONCLUSION The model is proposed of snowmelt-rainfall runoff of the mountain river taking account of spatial inhomogeneity and vertical zoning of runoff factors. The verification of the model was carried out from the data of hydrometeorological observations in the Upper Kuban River basin, in particular, from the observations in the period of passage of the highest snowmelt-rainfall flood for the period of observations in June
RUSSIAN METEOROLOGY AND HYDROLOGY Vol. 35 No. 12 2010

850

KUCHMENT et al.

Fig. 5. The hydrograph of the Kuban River near the rural settlement named after Kosta Khetagurov in 2002 from the (1) observational and (2) model data.

2002. The satisfactory coincidence is obtained of computed hydrographs with those plotted from the observational data. The proposed model accuracy can be increased by means of the assimilation of the radar spatial precipitation measurement data and satellite snow cover measurements. ACKNOWLEDGMENTS The work was supported by the Russian Foundation for Basic Research (project No. 08-05-00375). REFERENCES
1. Yu. M. Denisov, Computation Scheme of Mountain River Hydrograph (Gidrometeoizdat, Leningrad, 1965) [in Russian]. 2. V. D. Komarov, V. M. Mukhin, and A. Ya. Polunin, Model of Snow Cover Formation and Water Inflow to the Mountain Basin Surface, Trudy Gidromettsentra SSSR, No. 163 (1976) [Trans. Hydrometcenter of the USSR, No. 163 (1976)]. 3. L. S. Kuchment, Hydrological Forecasting for Water Resource Management, in Results of Science and Technology, ser. Land Hydrology (VINITI, Moscow, 1981) [in Russian]. 4. L. S. Kuchment, V. N. Demidov, and D. A. Bryazgin, Application of an Advection-Diffusion Model for Calculations of Unsteady Flow in Riverbeds, Meteorol. Gidrol., No. 5 (1999) [Russ. Meteorol. Hydrol., No. 5 (1999)]. 5. L. S. Kuchment, V. N. Demidov, and Yu. G. Motovilov, River Runoff Formation (Nauka, Moscow, 1983) [in Russian]. 6. V. M. Mukhin, Model of Mountain River Runoff Formation, Trudy Gidromettsentra SSSR, No. 186 (1977) [Trans. Hydrometcenter of the USSR, No. 186 (1977)]. 7. Resources of Surface Waters of the USSR, Vol. 8: Northern Caucasus (Gidrometeoizdat, Leningrad, 1973) [in Russian]. 8. E. A. Anderson, Point Energy and Mass Balance Model of a Snowcover. NOAA Technical Report National Weather Service No. 19 (U.S. Dept. Commerce, Silver Spring, MD, 1976). 9. G. S. Campbell, A Simple Method for Determining Unsaturated Hydraulic Conductivity from Moisture Retention Data, Soil Science, 117 (1974). 10. S. Mishra and J. C. Parker, On the Relation between Saturated Conductivity and Capillary Retention Characteristics, Ground Water, 28 (1990). 11. W. J. Rawls, D. L. Brakenseik, and K. E. Saxton, Estimation of Soil Water Properties, Trans. Amer. Soc. Agric. Engin., No. 5, 25 (1982). 12. Z. Yosida et al., Physical Studies on Deposited Snow, Contrib. Inst. Low Temp. Sci., No. 7 (1955).

RUSSIAN METEOROLOGY AND HYDROLOGY

Vol. 35

No. 12

2010

Você também pode gostar