Вы находитесь на странице: 1из 16

Hydrological Sciences -Journal- des Sciences Hydrologiques,39,5, October 1994

443

Hydrodynamic derivation of a variable parameter Muskingum method: 2. Verification


MUTHIAH PERUMAL Department of Continuing Education, University ofRoorkee, Roorkee 247667, India Abstract The hydrodynamic derivation of a variable parameter Muskingum method and its solution procedure for estimating a routed hydrograph were presented in Part I of this series (Perumal, 1994a). In this paper, the limitations of the method, the criterion for its applicability and its accuracy are discussed based on the assumptions used. The method is verified by routing a given hypothetical inflow hydrograph through uniform rectangular cross-section channels and comparing the results with the corresponding numerical solutions of the St. Venant equations. The stage hydrographs as computed by the method are also compared with the corresponding St. Venant solutions. It is demonstrated that the method closely reproduces the St. Venant solutions for the discharge and stage hydrographs subject to the compliance of the assumptions of the method by the routing process. Dmonstration hydrodynamique d'une mthode de Muskingum paramtres variables: 2. Validation Rsum La dmonstration hydrodynamique d'une mthode de Muskingum paramtres variables et la description de la procdure permettant d'estimer un hydrogramme ont fait l'objet de la premire partie de ce papier (Perumal, 1994a). Dans cette seconde partie on discutera les limites de la mthode ainsi que les conditions sous lesquelles elle est applicable et prcise en s'appuyant sur les hypothses formules. La mthode est valide en tudiant le routage d'un hypothtique hydrogramme d'entre travers des biefs de sections uniformment rectangulaires et en comparant les rsultats la solution numrique des quations de Saint Venant. Les limnigrammes calculs par notre mthode sont galement compars aux solutions de Saint Venant correspondantes. Il apparat que la mthode reproduit fidlement les solutions des quations de Saint Venant pour les dbits et les niveaux pourvu que les hypothses de la mthode soient satisfaites au cours du processus de routage. INTRODUCTION An approach for the derivation of a variable parameter Muskingum method from the St. Venant equations for routing floods in channels having any shape of prismatic cross-section and flow following either Manning's or Chezy's friction law was presented in Part I of this series (Perumal, 1994a) along with the solution procedure for estimating the routed hydrograph using this method.
Open for discussion until I April 1995

444

Muthiah Perumal

The definition sketch of the Muskingum routing reach of length Ax is shown in Fig. 1. According to one of the assumptions used in the development of the method, during unsteady flow the stage at the middle of the reach corresponds to the normal depth of that discharge which passes at the same instant of time at a distance / downstream from the middle of the reach. Let this discharge be denoted as Q3 and the section where it occurs be denoted as section (3). The inflow and outflow sections are represented as sections (1) and (2) respectively. This paper brings out the limitations, the applicability criterion and the accuracy of the method based on the assumptions of the method.

Fig. 1 Definition sketch of the Muskingum reach. EVALUATION OF THE APPLICABILITY OF THE METHOD It was noted while deriving this methodology (Perumal, 1994a) that the hydrograph to be routed should satisfy the criterion | l/S0(dy/dx) | < 1 at any time for a successful application, with S0 and dy/dx denoting the channel bed slope and the water surface slope respectively. Fortunately, there is no need to separately evaluate the applicability criterion, as the method has the inherent ability to bring out its inapplicability in the form of computational instability while routing a given inflow hydrograph for a small single reach length. The small single reach length is such that routing results arrived at the same location by using further sub-divisions of this small single reach would not produce significantly different results from that of the single reach solution. However, if further sub-division of the small reach length leads to a computational problem then the inapplicability of the method for routing a given flood hydrograph is evident. The method may be forced to work by identifying a minimum reach, by a trial and error approach, for which there is no computational problem while routing. However, when a longer single reach length is used to avoid this computational problem, the inaccuracy in the results will increase as the assumption of linear variations of water surface and discharge for such a long reach may not hold good. Nevertheless, one may use judgement in accepting such results depending on their utility.

Muskingum method derivation: verification

445

ACCURACY OF THE METHOD


The accuracy of the solution using the proposed method depends on the satisfaction of the assumptions involved in the development of the method. As a consequence of these assumptions, there are two major sources of approximations involved: (a) approximation with reference to the use of the assumption of a linear variation of water surface along the routing reach; and (b) approximation due to a Binomial series expansion of the energy slope in estimating the distance / between the mid-section and the normal flow section (section 3) of the reach. The error introduced due to the former approximation can be minimized by reducing the length of the routing reach. However, if the truncation error introduced by retaining only the first two terms of the Binomial series expansion, as adopted in the present method, is significant, then routing using an increased number of sub-reaches would compound the inaccuracy in the solution due to the recursive nature of the solution equation, especially when the magnitude of the term | HSJJdyldx) | is near to unity. The accuracy of the computed stage hydrograph depends on the accuracy of the computed discharge hydrograph which in turn depends on the satisfaction of the assumptions of the method. The implications of this statement will be explained later by routing a given flow hydrograph through rectangular crosssection channels using this method. APPLICATION The flow hydrograph defined by a four parameter Pearson Type-Ill distribution expressed by the following equation was used in all the test runs: t
i (7-1)

T p

exp

(i-f/g
y

. ^~

l)

where Ib is the initial steady flow (100 m3 s~'), Ip is the peak flow (1000 m3 s 1 ), t is the time to peak (10 h), and y is the skewness factor (1.15). This form of inflow hydrograph has been used for testing flood routing methodologies by various investigators (NERC, 1975; Weinmann & Laurenson, 1979; Ponce & Theurer, 1982; Garbrecht & Brunner, 1991). A rectangular channel with a uniform bed width of 50 m was used for all the test runs and the hydrograph was routed following the solution procedure described by Perumal (1994a) for a distance of 40 km from the inflow section by subdividing the reach into a single reach and various numbers of sub-reaches. The method was tested on three different channels as given in Table 1. Ten test runs with details as indicated in Table 2 were made in order to have a better understanding of the proposed methodology. A routing interval of 15 min was used on all runs.

446

Muthiah Perumal

Table 1 Channel configurations


Channel type Bed slope (S) 0.0002 0.0002 0.002 Manning's roughness (n) 0.04 0.02 0.04

Table 2 Test runs details


Test run no. Channel type Solution required at (km) Total reach length (km) No. of subreaches Length of subreach (km)

1 2 3 4 5 6 7 8 9 10

1 1 1 2 2 2 3 3 3 2

40 40 40 40 40 40 40 40 40 5

40 40 40 40 40 40 40 40 40 5

1 2 3 1 8 40 1 8 40 1

40 20 13.33 40 5 1 40 5 1 5

THE PERFORMANCE CRITERIA The performance of the method was evaluated by comparing its solution with the exact solution obtained using the St. Venant equations based on the following criteria as adopted by Weinmann (1977): (a) accuracy of the simulated peak outflow; (b) accuracy of the simulated time to peak outflow; (c) accuracy of the conservation of mass; (d) accuracy of the simulated peak stage; and (e) accuracy of the simulated time to peak stage. The accuracy of the reproduction of the hydrograph shape and size was evaluated by the Nash-Sutcliffe criterion (Nash & Sutcliffe, 1970). RESULTS AND DISCUSSION Hydrograph reproduction Table 3 shows the summary results of all the test runs with reference to reproducing some pertinent characteristics of the discharge hydrographs under various unsteady flow conditions. Table 4 shows the summary results of the corresponding test runs with reference to the reproduction of peak stage characteristics. Figure 2 shows the results of discharge hydrograph routing in channel type 1 corresponding to test runs 1-3. It shows the inflow hydrograph, the exact solution and the corresponding routed hydrographs arrived at using the present method for three cases, referring to the sub-division of the 40 km reach

Muskingum method derivation: verification

447

Table 3 Summary of results showing reproduction of pertinent characteristics of discharge hydrographs


Test run Peak flow and time to peak Proposed method St. Venant' s solution 3 3 2(m "') t (h) Q (m s"') tW e(%) 1 2 3 4 5 6 7 8 9 10 765 765 765 885 885 885 993 993 993 985 14.25 14.25 14.25 13.00 13.00 13.00 12.25 12.25 12.25 10.25 711 677 669 868 856 856 983 993 993 979 14.25 14.25 14.25 13.00 13.25 13.25 12.25 12.25 12.25 10.25 -7.06 -11.50 -12.55 -1.92 -3.28 -3.28 -1.00 0.00 0.00 -0.61

tP (h) 0.00 0.00 0.00 0.00 0.25 0.25 0.00 0.00 0.00 0.00

Variance explained 97.50 97.88 97.43 99.27 99.73 99.72 99.13 99.97 99.98 99.99

EVOL(%) -0.29 0.11 0.50 0.12 0.12 0.12 0.32 0.41 0.42 0.03

Table 4 Summary of results showing reproduction of pertinent characteristics of stage frydrographs


Test run Maximum stage and time to maximum St. Venant's solution Proposed method yP (m) 10.21 10.21 10.21 7.43 7.43 7.43 6.14 6.14 6.14 7.93 ',(h) 16.00 16.00 16.00 13.75 13.75 13.75 12.25 12.25 12.25 11.25 yP (m) 11.47 10.30 10.03 7.82 7.46 7.43 6.11 6.14 6.14 8.13

',00
17.00 16.75 17.00 14.00 14.25 14.25 12.25 12.25 12.25 11.50

y per ( m )

', (h)
1.00 0.75 1.00 0.25 0.50 0.50 0.00 0.00 0.00 0.25

i 2 3 4 5 6 7 8 9 10

1.26 0.09 -0.18 0.39 0.03 0.00 -0.03 0.00 0.00 0.20

1000

30 Time (hr)

Fig. 2 Routing results for channel type 1.

448

Muthiah Perumal

into a singe reach and two and three equal sub-reaches for routing the given inflow hydrograph. Routing using more than three equal sub-reaches, in the case of channel type 1, resulted in the abortion of solution due to the development of high negative values of the Muskingum weighting parameter 6 in the first reach. Figure 3 refers to the results of the estimated stage hydrographs corresponding to the routed discharge hydrographs of test runs 1-3 arrived at by using the stage-discharge relationship given by Perumal (1994a).

if if
i -

'. "\N\ \-\\

Single reach solution Two sub-reaches solution Three sub-reaches solution

\ \

if i i

if
\

il
i i i 1 I i

VS*. NV "^V.

- \ y
, 1 , ! i

10

20 Time

30 ( hr )

40

50

60

Fig. 3 Computed stage hydrographs for channel type 1.

3 sub-reaches - 2 s u b - reaches -single reach

. 3 sub - r e a c h e s 2 s u b - reaches single reach 30 Time ( h r )


40 50 60

Fig. 4 Variation of 1/SJdy/dx) computed at the inlet of the reach (channel type 1).

Muskingum method derivation: verification

449

Figures 5 and 7 refer to the results of discharge hydrograph routing in channel types 2 and 3 corresponding to test runs 4-6 and 7-9 respectively. They show the inflow hydrograph, the exact solutions and the corresponding routed hydrographs arrived at by using the present method for three cases, referring to the subdivision of the 40 km channel reach into a single reach, and 8 and 40 equal sub-reaches for routing the given inflow hydrograph. Figures 6 and 8 refer to the results of estimated stage hydrographs corresponding to the routed discharge hydrographs shown in Figs 5 and 7 respectively.
1000 1 800 1 i i v> ip If i M * Inflow

.
xx

Single

reach

solution

\ \
e 600

M u l t i p l e reach solutions ( S a n d 40 s u b - r e a c h e s )

; f
; jr
\ \ \

400

1
200

II

; if

\ \

- <1
i i fir i 10
~i

. j 20

i Time

i. 30 (hr)

a 40

_i

i i 50

60

Fig. 5 Routing results for channel type 2.

30 Time ( hr)

Fig. 6 Computed stage hydrographs for channel type 2.

450

Muthiah Perumal

It is seen from Fig. 2 and Table 3 that the reproduction of the exact solution characteristics including that of peak flow is achieved better by single reach routing, except at the beginning where the infamous "dip" characteristic of the Muskingum method dominates. It is also seen from Fig. 2 that routing with two and three sub-reaches reduces the size of the dip.

1000

Ti me (hr )

Fig. 7 Routing results for channel type 3.

5 -

Exact solution Single reach solution Multiple reach solutions (8 and 40 s u b - r e a c h e s )

3 -

10

20

30 Time ( h r )

40

50

60

Fig. 8 Computed stage hydrographs for channel type 3.

Muskingum method derivation: verification

451

It was shown by Perumal (1992) that the assumption of linear variation of discharge along with the condition that section (3) lies between the midsection and the outflow section of the reach is responsible for the formation of negative or reduced outflow at the beginning of a routing solution. Therefore, the higher the magnitude of | l/S0(dy/dx) |, and hence the higher the discharge variation in space, dQ/dx, and the length of the reach, the higher would be the dip or reduced outflow at the beginning of routing. The factors which influence the formation of higher \l/S0(dy/dx)\ are the steepness of the hydrograph, small bed slope and higher roughness coefficient. There is almost no difference between the routing results obtained using two and three sub-reaches. Although the size of the dip reduces considerably in these cases, the peak flow estimated is overattenuated when compared with that of single reach routing. However, it is not appropriate to discuss the results of discharge hydrograph routing in isolation from the results of the corresponding estimated stage hydrograph. It is seen from Fig. 3 that the estimated stage hydrograph corresponding to single reach routing is overpredicted for the entire exact solution in general, and very much so near the peak region. Also, it is interesting to note that the stage hydrographs estimated corresponding to two and three sub-reaches routing of the discharge hydrograph reproduce the exact solution of the stage hydrograph closely. Therefore, the inability of the method to reproduce both the stage and discharge hydrographs in a consistently better manner may be attributed to the non-compliance of the assumptions used in the development of the method by the routing process in channel type 1. Figure 4 shows the variation of the nondimensionalized slope l/S0(dy/dx) of the exact solution vs. time at the inlet of the reach corresponding to the given inflow hydrograph. Also shown are the corresponding variations of l/S0(dy/dx) estimated from a single reach, and two and three subreaches routing using the relationship:

1 ?1 = - _L ?.

(2)

So dx Be2 dt where c is the wave celerity, B is the top width of the water surface and dQ/dt denotes the rate of change of discharge with reference to time. It is seen from Fig. 4 that the estimated magnitudes of l/S0(dy/dx) for all the cases of routings based on a single reach and two and three sub-reaches is much greater than the corresponding actual l/S0(dy/dx) even though it is less than unity for the entire inflow hydrograph, thus satisfying the applicability criterion of the method. It may be noted that when the absolute maximum of l/S0(dy/dx) during the rising part of the hydrograph is less than, but nearer to unity, the truncation error introduced by retaining only the first two terms of the Binomial series expansion of Sf causes the solution of the method to be different from the exact solution. Due to the recursive nature of the solution equation, the truncation error also propagates resulting in the deviation from the actual solution. This aspect can be visualized from Fig. 5, especially, when

452

Muthiah Perumal

l/So(dy/dx)>0. The truncation error increases with the increase in the subdivisions of the reach. It may be noted that while theoretically the maximum value of \IS0(dyldx) has to be nearer to unity, the estimated + l/S0(dy/dx)> 1 due to truncation errors. It is observed that when the number of sub-reaches is increased beyond three in channel type 1, the routing leads to high negative values of 6 during the first sub-reach routing while both the inflow and outflow are receding. As seen from Fig. 4, routing based on an increasing number of sub-reaches greatly increases the truncation error in the estimated + l/S0(dy/dx), resulting in the location of section (3) (where the normal discharge corresponding to the stage at the middle of the reach occurs) far downstream of the outflow section (section (2)), thus estimating a higher magnitude of negative 6. Due to the recursive form of the solution equation, section (3) is subsequently located still farther away resulting in unbound negative values of 6 leading to the abortion of the solution algorithm. Despite the introduction of large truncation errors, the overall reproduction of exact discharge and stage hydrographs are not severely affected by the routing solutions obtained using two and three subreaches of the 40 km reach. Besides, the conservation of volume criterion is not violated with the parameter EVOL < 0.5%. One may accept or reject such results depending on their practical utility. Figures 5 and 7, corresponding to test runs 4-6 and 7-9 respectively, show the close reproduction of the respective exact solutions, but with the development of the dip at the beginning of the single reach routing solution. These cases demonstrate that the approximations involved in arriving at the present method are not too severe to affect the close reproduction of the exact solutions. A close reproduction of the stage hydrographs of the exact solutions by this method can be seen from Figs 6 and 8. In the cases of routings in channel types 2 and 3, it can be seen from Figs 5 and 7 that there is no difference between the results based on the routings using 8 and 40 equal subdivisions of the 40 km reach. This leads to the inference that there exists a limit on the use of the number of subdivisions of a reach for yielding improved results, beyond which no improvement can be obtained with increased computational effort. Wave attenuation characteristics As the same inflow hydrograph was routed in all these studies for the same distance, it may be inferred that the attenuation characteristics of the routed hydrograph can be studied effectively in terms of the characteristics of the rating curves observed during the passage of the inflow hydrograph at the inlet of the reach corresponding to each channel type. The variations in the rating curve characteristics are the result of differing roughness and slope characteristics of the channels. Figure 9 shows the rating curves obtained at the inlet of the reach for all three channel types. The narrowing of the loop for channel type 2 compared with that of channel type 1 is due only to the reduction in the

Muskingum method derivation: verification

453

Manning roughness value from 0.04 for channel type 1 to 0.02 for channel type 2. The rating curve obtained at the inlet of channel type 3 is not distinguishable from the steady flow rating curve. It is seen that the higher the bed slope and the lower the Manning roughness coefficient the narrower will be the loop of the rating curve. Under such conditions, the routing results using the described method closely reproduce the exact solutions, and when the rating curve almost corresponds to that of a steady state relationship, the flood wave is governed by kinematic wave characteristics as shown in Fig. 7. Also the peak stage values of the computed and exact solutions shown in Fig. 8 coincide with the corresponding peak discharge value of the routed hydrograph shown in Fig. 7.

Inflow

discharge ( m

s )

Fig. 9 Loop rating curves at inlet of the reaches for all channel types. When the rating curves are characterized by a wider loop width, the inflow hydrograph attenuates as shown in Figs 2 and 5, with less attenuation in channel type 2 than in type 1. It is to be noted that when the inflow hydrograph is characterized by a wide loop rating curve, as in the case of routing in channel type 1, the overall reproduction of the exact solution is not satisfactory when compared with the reproduction of the exact solution in channel type 2. The formation of a wide loop rating curve implies the development of significantly higher water surface slope dy/dx when compared with the bed slope S0.

Variation of travel time Figure 10 shows the variation of the Muskingum travel time parameter K, with reference to inflow discharge during a single reach routing in all three channel

454

Muthiah Perumal

types. The inflow discharge hydrograph remained the same for all the routing studies made herein. Since the wave celerity was always positive, the inference arrived at from Fig. 10 for a single reach routing can be generalized for multiple reach routings also. The relationship of the present method exhibits the formation of a loop with larger K for a given discharge in a rising limb of the hydrograph and a smaller K for the same discharge observed in the falling limb of the inflow hydrograph due to the accounting of water surface slope. It is also seen that the range of variation of K and its magnitude decreases with increase in the bed slope and reduction in channel roughness. It can be inferred from Figs 9 and 10 that a wide loop rating curve results in a longer travel time of flood discharge with a large variation range.

100

300

500 inflow (m3 s1 )

700

900

1000

Fig. 10 Variation of K for single reach routing. Variation of weighting parameter Figure 11 shows the variation of the other Muskingum parameter, the weighting parameter 6, with the inflow discharge corresponding to a single reach routing. Before discussing these results, it is necessary to look briefly into the aspects of the variation of 6 from a physical consideration. An insight into the variation of 6 can be gained by studying the expression for 6 given by the equation (Perumal, 1994a):

6 =1- . 1

(3)

2 Ax When section (3) lies between the midsection and the outflow section of the routing reach, 0 < 6 < 0.5. When it coincides with section (2), 6 = 0 as in the Kalinin-Milyukov method. When it is located downstream of the midsection of the reach as well as downstream of section (2), 6 < 0.

Muskingum method derivation: verification

455

01

100

300

500 inflow ( m s )

700

900

1000

Fig. 11 Variation of 6 for single reach routing.

It is seen from Fig. 11 that a wide variation in the magnitude of 6 is observed with reference to inflow for channel types 1 and 2. The relationship exhibits a wide loop with a higher magnitude of 6 for a given discharge in the rising limb of the inflow hydrograph and a smaller magnitude of 6 for the same discharge in the falling limb of the hydrograph. The wider the loop of the rating curve, as noted in Fig. 9, the wider also the loop of this relationship which characterizes the attenuation of the inflow hydrograph while routing for the same length of single reach. However, very insignificant variations of 6 with the magnitude being closer to 0.5 are observed corresponding to inflow in channel type 3. This is due to the insignificant role of the term l/S0(dy/dx) in characterizing the inflow hydrograph in channel type 3. The absence of a loop with regard to the relationship of 8 and its presence with regard to the relationship of K suggest that the flood wave propagation in channel type 3 is governed by the celerity of the flood wave with no effect of channel storage. This behaviour is characteristic of kinematic wave propagation. When / > Ax/2, then 6 < 0. Under such a condition, the outflow discharge would be greater than the normal discharge Q3 passing at section (3). This situation was realized in test run 10 in which the 6 values corresponding to each time level of routing were negative and thus the outflow discharge was greater than normal discharge Q3 at all the routing time levels. Figure 12 shows the routing results of test run 10 for a single reach solution, along with the exact solution. The estimated normal discharge corresponding to this single reach routing is also shown. Confirming the interpretation of equation (3), the outflow discharge hydrograph is observed ahead of the normal discharge hydrograph. Figure 13 shows the variation of 6 with inflow estimated for this single

456
1000

J^p

Muthiah

Perumal

V
800

\\

B 600

V fi

Inflow Exact solution Direct solution Estimated normal discharge hydrograph at section (3)

\\

. -

400

.Tl

ii

V V

VV

200h 100

jl ji
1 10

V V
^X
20

^ * - J_
30 Time (hr )

40

50

60

Fig. 12 Hydrograph at 5 km due to direct routing when outflow discharge section is upstream of normal discharge section (channel type 2).

a a

5 _

100

300

500

700
1

Inflow ( m 3 s

900

1000

Fig. 13 Variation of 6 for single reach routing at 5 km (channel type 2).

Muskingum method derivation: verification

457

reach routing. Negative values of 6 were estimated for the entire routing process with the limits -2.21 < 6 < -0.08. Although the possibility of 6 < 0 was indicated by Dooge (1973) from a mathematical consideration, the argument from the physical point of view was put forward by Strupczewski & Kundzewicz (1980). Later Wong (1984) also offered an interpretation of negative 6 based on the same lines as presented herein, although he did not use this interpretation for relating 9 with channel and flow characteristics. While 6 can assume negative values of any magnitude theoretically, there exists a limit from the computational point of view. When section (3) coincides with the midsection of the reach, i.e. ,1 = 0, then 6 = Vi. This represents the situation in which the normal discharge coincides with the normal depth at the midsection of the reach, thus depicting kinematic wave propagation. Figure 7 demonstrates the kinematic wave propagation in channel type 3 with 6 being closer to 0.5 as shown in Fig. 11. CONCLUSIONS Evaluation of the applicability of the variable parameter Muskingum method (Perumal, 1994a) reveals that the hydrograph to be routed should satisfy the criterion | l/S0(dy/3x)| < 1 at any time for the successful application of the method. However, the method has the built in ability to bring out its inapplicability in the form of computational instability when this criterion is not satisfied. Ten test runs were made by routing a given hypothetical inflow hydrograph using this method in rectangular cross section channels to test its ability in closely reproducing the St. Venant solutions of the discharge and stage hydrographs. In all the test runs, the inflow hydrograph satisfied the applicability criterion of the method defined by | l/S0(dy/dx) | < 1. However, it was found that in some test runs, although this criterion was satisfied, when certain portions of the inflow hydrograph are characterized by | \IS0(dyldx) | values which are nearer to unity, the method failed to reproduce the St. Venant solution closely. It was found that truncation errors introduced by retaining only the first two terms of the Binomial series expansion of Sf were responsible for this inaccuracy. The variation of the Muskingum parameters corresponding to a single reach routing were studied with reference to the inflow hydrograph. It was found that the wider the loop of the rating curve observed at the inlet of the reach, the wider also the loop of this relationship. For kinematic flood wave movement, the variation of 6 was insignificant, its value being close to 0.5. In all the cases studied, the conservation of volume was not violated by the method, with the maximum error being only 0.5%.

REFERENCES
Dooge, J. C. I. (1973) Linear theory of hydrologie systems. ARS Tech. Bull. no. 1468, US Agriculture Res. Serv., Washington, DC, USA.

458

Muthiah Perumal

Garbrecht, J. & Brunner, G. (1991) Hydrologie channel-flow routing for compound sections. J. Hydraul. Engng 117(5), 629-642. Nash, J. E. & Sutcliffe, J. V. (1970) River flow forecasting through conceptual models part I - A discussion of principles, /. Hydrol. 10, 282-290. Natural Environment Research Council (1975) Flood studies report, volume III - Flood routing studies. London, UK. Perumal, M. (1992) The cause of negative initial outflow with the Muskingum method. Eydrol. Sci. J. 37(4), 391-401. Perumal, M. (1994a) Hydrodynamic derivation of a variable parameter Muskingum method: 1. Theory and solution procedure. Hydrol. Sci. J. 39(5) (this issue) Ponce, V. M. & Theurer, F. D. (1982) Accuracy criteria in diffusion routing. /. Hydraul. Div. ASCE 108(6), 747-757. Strupczewski, W. G. & Kundzewicz, Z. W. (1980) Muskingum method revisited. /. Hydrol. 48, 327342. Weinmann, P. E. (1977) Comparison of flood routing methods for natural rivers. Report no. 2/1977, Dept. of Civ. Engng, Monash University, Victoria, Australia. Weinmann, P. E. & Laurenson, E. M. (1979) Approximate flood routing methods: a review. J. Hydraul. Div. ASCE 105(12), 1521-1536. Wong, T. H. F. (1984) Improved parameters and procedures for flood routing in rivers. Ph.D. dissertation, Monash Univ., Victoria, Australia. Received 8 February 1993; accepted 7 April 1994

Вам также может понравиться