Distributed rainfall runoff analysis in a flow regulated basin having multiple multi-purpose dams

Transcription

1 Preitions in Ungauge Basins: Promise an Progress (Proeeings of symposium S7 el uring te Sevent IAHS Sientifi Assembly at Foz o Iguaçu, Brazil, April 2005). IAHS Publ. 303, Distribute rainfall runoff analysis in a flow regulate basin aving multiple multi-purpose ams TAKAHIRO SAYAMA 1, YASUTO TACHIKAWA 1, KAORU TAKARA 1 & YUTAKA ICHIKAWA 2 1 Disaster Prevention Resear Institute, Kyoto University, Gokaso Uji, Kyoto, , Japan sayama@floo.pri.kyoto-u.a.jp 2 Department of Urban an Environmental Engineering, Kyoto University, Yosia Honmai, Kyoto, , Japan Abstrat Tis paper presents yrologial preitions an floo safety assessment in a flow regulate basin wit multi-purpose ams. A istribute rainfall runoff preition system is evelope by inorporating a am reservoir operation moel into a pysially base rainfall runoff moel for simulating a igly regulate Japanese river basin, te Yoo River basin (7281 km 2 ), in wi eigt multi-purpose ams are urrently operate. Te test simulation for typoon events verifie tat te system an simulate omplex floo ontrol operations su as preliminary release, peak attenuation, an water level maintaining. Following te test simulation, te evelope system was applie to assess te floo safety level of te basin. Te numerial experiments iniate tat te basin is safe for a rainfall event of a 100-year return perio ue to te newly onstrute multi-purpose ams, wereas it was safe only for a rainfall event of a 30-year return perio in Key wors am moel; istribute rainfall runoff moel; floo safety level; flow regulation; uman impat; peak attenuation; Yoo River basin INTRODUCTION Te alterations of te yrologial yle ause by environmental ange iniate tat observe information wit past or urrent onitions may not be appliable for future preitions. It is essential for te perspetive of Preition in Ungauge Basin (PUB) to unerstan te sientifi bakgroun of yrologial alterations inue by uman ativity an to preit te future yrologial yle base on tis unerstaning. One of te ruial flow regime anges is ause by am reservoirs. Tere are mainly two approaes for te assessment of flow regime ange inue by am reservoirs. One is te ata riven approa (ex. Batalla et al., 2004), wi uses isarge ata observe before an after am onstrutions; te oter one is te moel approa, wi uses yrologial moels inorporate wit am reservoir moels. Te avantage of te moel approa is te possibility of onuting various ypotetial experiments su as simulating unexperiene extreme floo events. Montalo et al. (2004) ombine am moels wit a istribute rainfall runoff moel to simulate te effets of te ams on floo attenuation. Sine te purpose of te ams tey treate in teir resear was limite to yroeletriity proution, te am oul be moelle wit a simple storage funtion.

2 372 Takairo Sayama et al. Tis paper fouses on floo ontrol wit multi-purpose ams, wi are operate in a ompliate manner base on am operation rules. Te am moel evelope in tis stuy preits outflow an water level wit te input information of inflow, upstream rainfall an ooperative am operations by moelling am operation rules. A yrologial preition system is onstrute by inorporating te am moels wit a istribute rainfall runoff moel. We evelope tis system for a igly regulate Japanese river basin, te Yoo River basin (7281 km 2 ). By using te system evelope, we assess am effets on floo ontrol to investigate ow te floo safety level as been inrease by newly onstrute ams an to unerstan wi range of floo magnitues te ams an regulate effetively. Tis paper is ompose of tree main parts. Te first part isusses te evelopment of te rainfall runoff preition system inorporating am moels. Te seon part verifies te system wit a typoon rainfall event. Te tir part presents analysis of am effets on floo ontrol. Finally, we onlue wit some insigts into ontributions of rainfall runoff analysis onsiering uman impats to PUB. A DISTRIBUTED RAINFALL RUNOFF PREDICTION SYSTEM INCORPORATING A DAM OPERATION MODEL Wole struture of te rainfall runoff preition system Te rainfall runoff preition system is onstrute base on te Objet-oriente Hyrologial Moelling System (OHyMoS). Te following four kins of element moels ompose te wole system (Fig. 1). (a) River element moel: Te kinemati wave moel is applie to a river segment, wi is prepare from te igital river-network ata an te loation information of lakesores. Ea river segment is ut to be about 3 km lengt. (b) Sub-atment element moel: A saturate-unsaturate subsurfae an surfae runoff moel (Taikawa et al., 2004) is applie to all gri-ells omposing a subatment. We use a igital elevation moel wit 250 m resolution to alulate te flow iretion an to efine te sub-atment of ea river segment. () Lake element moel: Te lake element moel is a simple mass-balane moel to simulate water level from inflow, outflow, an rainfall information. We apply tis moel to Lake Biwa, of wi outflow is simulate wit te am element moel. () Dam element moel: A am operation moel is onstrute wit an if-ten style base on ea am operation rule. We apply it to eigt multi purpose ams in te basin. We refer to tis site-speifi am operation moel as a am element moel. Sub-atment element moel Figure 2 (a) sows a semati iagram of te soil layer of te moel, wi takes into aount tree types of flow: unsaturate flow in apillary pore, saturate flow in nonapillary pores, an surfae flow on te soil surfae. In tis figure, te soil ept is D (m), te water ept orresponing to te water ontent is s (m), an te water ept orresponing to maximum water ontent in te apillary pores is (m).

3 Distribute rainfall runoff analysis in a flow regulate basin 373 Fig. 1 Data proesses an te wole struture of te system. Fig. 2 Semati iagram of te soil layer: (a) an stage isarge relationsip; (b) of saturate unsaturate subsurfae an surfae runoff moel. Figure 2(b) sows te relationsip between water ept an isarge of te moel. Let k an k a be saturate yrauli onutivities in apillary pores an in non-apillary pores, respetively, an v = k i, v a = k a i (i: slope), ten a relationsip between te isarge per unit wit q (m 2 s -1 ) an te water ept (m) are esribe as follows: ( ) ( ) ( ) ( ) ( ) ( ) < + α + < + = β v v v v v q s m s a s a,, 0, (1) (a) (b)

4 374 Takairo Sayama et al. were α equals to i 1/2 /n; n is Manning rougness oeffiient (m -1/3 s -1 ]. β [ ] is te parameter to esribe te reution of yrauli onutivity in apillary pore as te water ontent ereases. β equals k a /k so as to keep te ontinuity of te ept isarge relationsip between te apillary pore an te non-apillary pore layers. Combining tis ept isarge relationsip (1) an te ontinuity equation, we simulate rainfall runoff from ea gri-ell. Te simulate isarge forms inflow to te ownstream gri-ell. Te water flow is route until it reaes a river segment. Dam element moel By formulating te am operation rules an eision-making proesses of am operators, we evelop te am operation moel. It preits te outflow an water level of a am wit te input information of inflow, average rainfall in te am atment, an operation status of oter relate ams. All te ams loate in te Yoo River basin are multi-purpose ams. Altoug ea am as ifferent operating rules, we an ategorize all te floo ontrol operations into te following six ommon operation proesses (Iikawa, 2001): Orinary operation; Operation uner floo warning; Preliminary release operation; Peak attenuation operation; Floo release operation; an Post floo operation. Ea am is always uner one of te six operations, an we formulate te onitions to sift from one operation to anoter wit if-ten equations. Figure 3 (left) sows ow to sift te proess from one to anoter, an Fig. 3 (rigt) sows speifi water levels tat appear in te am operation rules. Fig. 3 Operation status an speifi water levels of am operation moel. TEST SIMULATION Stuy area Te Yoo River basin is analyse in tis ase stuy beause it is a typial Japanese river basin igly regulate by multi-purpose ams. Te total area of te Yoo River

5 Distribute rainfall runoff analysis in a flow regulate basin 375 basin is 8240 km 2. In tis ase stuy, sine Hirakata is te main esign target loation for esigning ams an oter river works, we fous on te upper Hirakata basin (7281 km 2 ) (Fig. 4), an refer to tis basin as te Yoo River basin in tis paper. Tere are eigt multi-purpose ams insie te basin an five of tese ams are loate in one of te sub-basins, alle te Kizu River basin (1596 km 2 ). Te Setagawa weir ontrols outflow from Lake Biwa (670 km 2 ), wi is te largest lake in Japan (Table 1). Te mean annual preipitation of te Yoo River basin is about 1600 mm. Fig. 4 Yoo River basin (7281 km 2 ). Table 1 Properties of te ams in te Yoo River basin. Dams Operation start year Catment area (km 2 ) Total volume (10 6 m 6 ) Seto Amagase Takayama Sorenji Murou Nunome Hiyosi Hinai Floo ontrol apaity (10 6 m 6 )

6 376 Takairo Sayama et al. Simulation onitions We onute a test simulation using observe rainfall an observe isarge ata uring a typoon event in Te perio of te simulation is from 25 to 29 July. We use te nearest neigboroo meto to istribute rainfall ata tat is observe by 58 raingauge stations insie te basin. Te estimate total rainfall over te basin is 149 mm. We ivie te wole basin into tree zones epening on te lan use, an assign ifferent parameters in te ifferent lan use zones. Tere are tree lan use ategories: forest, pay fiel, an urban area, among wi forest is te ominant lan use (63%). A rainfall runoff moel onsiering surfae flow is use for te pay fiel zone an te urban area zone by substituting zero in an s in equation (1), terefore, only te Manning oeffiient n is a parameter of te surfae rainfall runoff moel (pay fiel zone: 1.0 (m -1/3 s -1 ) an urban area zone: 0.3 (m -1/3 s -1 )). For te forest area zone, we use te saturate-unsaturate subsurfae an surfae rainfall runoff moel, an te parameters are as follows: n = 0.6 (m -1/3 s -1 ), D = 1.0 (m), s = 0.2 (m), = 0.1 (m), k s = 0.01 (m s -1 ). Six ams an Biwa Lake are also simulate in te rainfall runoff simulation system. Two ams onstrute after 1997 are not inlue in tis simulation. Simulation results Figure 5 sows te simulate an observe inflow an outflow at te Sorenji am. Te goo agreement between te simulate inflow wit te observe inflow verifies te rainfall runoff moel at te upstream of te am. Te outflow from te am operation moel sows tat te preliminary release an peak attenuation operations were properly simulate. Figure 6 sows te simulate an observe water level at te Sorenji am. It sows tat te water level is rawn own by te preliminary release an it is inrease by te peak attenuation operation. Fig. 5 Simulate an observe inflow an outflow at Sorenji am.

7 Distribute rainfall runoff analysis in a flow regulate basin 377 Fig. 6 Simulate an observe water level at Sorenji am. Fig. 7 Simulate an observe outflow at Seta weir. Figure 7 sows te simulate an observe outflow from te Setagawa weir an Fig. 8 sows te simulate an observe water level of Lake Biwa. Note tat te outflow is reue from 300 m 3 s -1 to 200 m 3 s -1 temporarily aroun 35 to 50 after te simulation began. Tis reution reflets te speial operation rule: Te Setagawa weir as to keep its outflow less tan 200 m 3 s -1 wen te Amagase am is uner preliminary release or peak attenuation operations inluing teir preparations. Figure 9 sows te simulate an observe inflow an outflow at te Amagase am, an we realize tat it was uner preliminary release operation from 35 to 50. Tis kin of omplex operation by multiple ams oul be simulate well beause te am operation moels an te rainfall runoff moels interat wit ea oter. Figure 10 sows te simulate an observe isarge at Kamo. Te simulation result exluing te am element moels is also isplaye in te figure. By omparing simulate yrograps wit an witout ams, it was onlue tat te ams loate upstream of Kamo oul attenuate te peak floo by aroun 1300 m 3 s -1.

8 378 Takairo Sayama et al. Fig. 8 Simulate an observe water level of Lake Biwa. Fig. 9 Simulate an observe inflow an outflow at Amagase am. Fig. 10 Simulate isarge (wit an witout ams) an observe isarge at Kamo.

9 Distribute rainfall runoff analysis in a flow regulate basin 379 ANALYSIS OF DAM EFFECTS ON FLOOD CONTROL Constrution of large-sale ams is believe to ave improve te safety level of a atment against floo isasters. However, it is not lear to wat extent tese ams ave improve te safety level or wi range of floo magnitues ams an regulate effiiently. In terms of te Yoo River basin, tere are urrently eigt large-sale ams in te basin an seven ams were onstrute after 1960 (Table 1). Conuting rainfall runoff simulation using te evelope rainfall runoff preition system, we evaluate ow te floo safety level of te Yoo River basin as been progressively inrease over te perio 1960 to 2000 wit respet to rainfall events of ifferent magnitue. METHOD Rainfall runoff simulations were implemente onsiering te ams tat were operate at te beginning of te following years: 1960, 1970, 1980, 1990, an Te simulate peak isarges at Hirakata are examine to isuss te effet of te ams. Te rainfall event use in tis stuy is te typoon event observe in 1982 (1 3 August). Tis event was osen beause it was te largest event sine 1980 wen enoug rainfall isarge sequenes were available to onut te simulation. Furtermore, in orer to examine wi floo magnitue te ams an attenuate te peak, some fators were multiplie to te 1982 rainfall pattern. Te fators are selete so tat two-ay total rainfall amounts in te Yoo River basin orrespon to te following return perios: 30, 50, 100, 150, 200 an 300 years. Te same parameters an te initial onitions use in te test simulation are use for tis assessment. RESULTS AND DISCUSSION Figure 11 sows te simulate peak isarge at Hirakata. Te ifferent lines represent ifferent years. Firstly, looking at te results of 1960 wen only te Setagawa weir existe, it an be observe tat peak isarge ause by te 30-year return perio rainfall (Q 30 ) exees m 3 s -1, wi is te allowable maximum floo isarge at Hirakata. By 1970, two oter ams, te Amagase am an te Takayama am, a been onstrute. Te omparison between te lines for 1960 an 1970 iniates tese two newly onstrute ams suessfully erease floo peaks from relatively smaller rainfall events up to aroun te 50-year return perio. By 1980, two more ams, te Sorenji am an te Murou am, a been onstrute. Tese ams enable te peak isarge to reue by about 2000 m 3 s -1 ompare wit te isarge from te 100- year an 150-year return perio rainfall. Tis is beause te Takayama am loate at te ownstream of te Sorenji am an te Murou am a a floo ontrol apaity large enoug to regulate floos of tis magnitue. By 2000, a furter tree ams, te Nunome am, te Hinai am, an te Hiyosi am, a been onstrute. It is noteworty tat, in 2000, te allowable maximum floo isarge was exeee by te isarge from te 100-year return perio rainfall (Q 100 ), wereas in 1960 tis value was exeee by te Q 30 peak isarge.

10 380 Takairo Sayama et al. Fig. 11 Simulate peak isarges at Hirakata wit ifferent magnitue of input rainfall. Years (1960, 1970, et.) enote tat te ams existe in te year are inlue in te simulation. CONCLUSIONS A rainfall runoff preition system was evelope for te Yoo River basin. Te ombination of te am operation moels an te rainfall runoff moels enable to simulate igly regulate rainfall runoff proesses. Te onlusions of te assessment of am effets on floo attenuation in te basin are summarize as follows: (a) Te ams onstrute in te 1960s were effetive in attenuating relatively small floo peaks ause by te rainfall event wit te magnitue of te 50-year return perio or smaller. On te oter an, te ams onstrute after 1970 were effetive in attenuating relatively larger floo peaks ause by te rainfall events wit te magnitue of te 100-year return perio or larger. (b) Q 30 orrespons to te allowable maximum floo isarge at Hirakata in 1960, wile Q 100 orrespons to it in However, we foun tat it as not aieve te initial esign target: te isarge ause by smaller tan te 200-year return perio rainfall annot exee te allowable maximum floo isarge. As it is isusse in tis paper, te yrologial yle in a atment an floo potential are altere by uman ativity, wi makes observe istorial reors not iretly appliable for future preitions. Moelling te water yle an te effet of uman ativity su as am reservoir operations is one of te solutions for tis type of ungauge basins. Aknowlegements A part of tis stuy is supporte by te System Moelling Approaes for Assessment of Interation Between Soial Canges an Water Cyle (PI: Kaoru Takara, DPRI, Kyoto University), wi is onute uner te CREST Program by te Japan Siene an Tenology Ageny, an Grant-in-Ai for Sientifi Resear C (PI: Yasuto Taikawa, DPRI, Kyoto University) provie by MEXT, Japan.

11 Distribute rainfall runoff analysis in a flow regulate basin 381 REFERENCES Batalla, R. J., Gomez, C. M. & Konolf, G. M. (2004) Reservoir-inue yrologial anges in te Ebro River basin (NE Spain). J. Hyrol. 290, Iikawa, Y. (2001) Framework an lumping of istribute rainfall runoff moels. Dr Eng. Tesis, Kyoto University, Kyoto, Japan (in Japanese). Montalo, N., Manini M. & Rosso, R. (2004) Floo yrograp attenuation inue by a reservoir system: analysis wit a istribute rainfall runoff moel. Hyrol. Proesses 18, Taikawa, Y., Nagatani, G. & Takara, K. (2004) Development of stage-isarge relationsip equation inorporating saturate-unsaturate flow meanism (in Japanese). Ann. Hyraul. Engng, Japan So. Civil Engrs 48, 7 12.