浦白川流出試験地の降雨一流出概念モデノレ - nied...
TRANSCRIPT
防災科学技術研究所研究報告 第53号 1994年3月
浦白川流出試験地の降雨一流出概念モデノレ
S.K.JAIN*・岸井徳雄**
Rainfa11-Runoff lModel1img of CatchmeIlts im Umjiro River
Basin Using A Comeptua1Mode1
By
S.K.JAIN*and T.KISHII榊
*∫㎜〃Z0〃,M肋〃α〃〃∫肋加ぴ助〃010馴,〃肋
**M肋〃α1地∫θ肌〃〃∫ガエ肋〃肋肋∫肋舳α〃D云∫α∫伽P閉θ〃0〃
Abstmct
The conceptual mathematica1mode1s are popu1ar1y used for rainfa11runoff
model1ing of a catchment.A conceptua1mode1has been app1ied for rainfa1l-runoff
modelling of the experimental catchments of NIED located in Chiba Prefecture-the
catchment up to Kakinokidai(are0.15sq.km.)and the catchment up to Tsukizaki
(area9.04sq.km.).The mode1results are reasonab1y good、
Key words:rainfa11,runoff,conceptua1mode1,experimenta1basin
キーワード:降雨,流出,概念モデル,流出試験地
1. Introduction
The rainfal1runoff mode1ing of a catchment using a conceptua1mode11ies intermedi-
ate between the detai1ed physica11y based mode1s and the empirical models.The term
“conceptua1”,is used to denote those mode1s which re1y on simp1e arrangement of
interlinked conceptua1e1ements,each representing a segment of the land phase of the
hydrologic cyc1e.The most common1y used element in a conceptua1model is the storage.
*防災科学技術研究所 気圏・水圏地球科学技術研究部
榊同 気圏・水圏地球科学技術研究部
一19一
Report of NIED,No.53,March1994
Each of these unequa1sized storage usua11y has one input and one or more outputs and
is used to represent a catchment storage,e.g.surface detention,soi1moisture etc,The
1inear reservoirs and chamels are used for routing purposes,The mode1ling basical1y
consists of a set of ru1es which govern moisture f1ow from one e1ement to another.Since
this is a non-iterative accounting procedure,these mode1s are computationa11y very
efficient.
The conceptua1mode1s were initia11y deve1oped for sma11homogeneous areas.
However,they have been successfu11y app1ied to basins having wide variations in
topography and vegetation and1arge area.The input data requirements for these mode1s
are quite modest and can be easi1y met with.Ciriani et a1(1977)and B1ackie&Eeles
(1985)give a1ucid discussion on phi1osophy and app1ications of these mode1s.
A conceptua1mode1has been used for rainfall-runoff mode11ing of two experimenta1
catchments of NIED,JAPAN.These catchments are1ocated in the Urajiro basin in
Chiba Prefecture.
2.Des㎝iption of the mode1皿sed
In the mode1used in the present study,the catchment is represented with the he1p of
three storages.The first storage termed as surface storage,represents the water stored
on the catchment surface.It has a maximum storage capacity given by Sm、、(mm).The
second storage represents the catchment soi1moisture storage andhas a maximum water
ho1ding capacity given by Cm、、(mm).The third storage represents the ground water
zone.It may be mentioned that most of the existing mode1s make use of simi1ar
arrangement of1inear reservoirs with varying degrees of comp1exity.
The rainfa1l is input to the surface storage.The water may1eave this storage
through evaporation,infi1tration or over1and f1ow.The moisture content of this storage
at any time is denoted by SURF.If SURF>Ep(potential evaporation in mm/hr),the
actua1evapotranspiration is at the potentia1rate e1se evapotranspiration(ET)takes
p1ace from the1ower storage at a1esser rate.The infi1tration of water from this storage
to the soi1storage takes p1ace at the rate INF:
INF=(1-C、。iI/Cm、、)*Fi,f if SURF>0
=O otherwise (1)
where Finf is a factor(mm/hr)control1ing the infi1tration rate.It may be noted that
when C、。i1=Cm、、,INF wi11be zero.One may visua1ize that in this event the surface and
一20一
Rainfall-Runoff Model1ing-S.K.Jain&T.Kishii
the soi1moisture storages have merged and the downward movement of moisture is
comp1ユted as described be1ow.
If at any instant SURF>Sm、、,the excess water over Sm、。,f1ows as over1and f1ow
(OF).The OF is routed through a1inear reservoir LR l with time constant K。.
The water infi1trated from the surface storage enters the soi1storage.The outf1ow
from this storage can take p1ace through ET1osses,interf1ow or recharge to the
groundwater zone.If the surface storage is empty,ET takes p1ace from the soil storage
at a rate Ea given by
E、=C、。ll/Cm、、*Ep (2)
and Cm、、is dep1eted by Ep*dt where dt is the1ength of computation interva1in hour.If
SURF>Ep*dt,the actual ET is SURF+E、*dt.The maximum va1ue of E.is Ep.
If the contents of soi1storage exceed a thresho1d denoted by FC,moisture f1ows out
of it as interf1ow and recharge to groun〔1water.The excess moisture availabIe for these
tWO iS:
Exw=(C、。i1/Cm、。一FC)*Ewf if C、。i1/Cm。。>FC (3)
where Ewf is a factor(mm/hr)control1ing the volume of excess water1The vo1ume of
interf1ow is given as:
IntF=Exw*Ci.t (4)
and the recharge to groundwater is
RECH=Exw*(1-Cint) (5)
where Cj,t is a dimension1ess coefficient which governs the division of the excess
moisture between recharge and interf1ow.The interf1ow is routed through a linear
reservoir LR2with time constant Ki.
The gromdwater zone behaves as a1inear reservoir whose time constant is KG.The
moisture comes out of it as the baseflo“(BF).The f1ow coming out of the reservoirs
LR1,LR2and LR3is combined and then routed through a1inear reservoir,LR4,to
yie1d the discharge from the catchment,denoted by TF.The box diagram of the mode1
structure is given in Fig.1.
一21一
Report of NIED,No.53,March1994
Rain ↑E。↓ ↓
Sm。。 Surface Storage SURF LR l OF →
↑E日
Cm。、 Soi1Storage C.oil LR2IntF → LR4 TF
BF →
Groundwater Zone(LR3)
Fig.1 Structure of the Mode1
The input to the mode1consists of the va1ues of various mode1parameters,the
period of simu1ation,and the time step size.The initia1contents of various storages are
a1so specified.The rainfa11and potential evaporation data for the period of simu1ation
are given as input.This mode1has been successfu11y used to simu1ate the response of a
few basins.
3. Th6stuay area and data皿sed
The data of the Urajiro river experimenta1catchment of National Research Institute
for Earth Science and Disaster Prevention(NIED).Japan,were used in this study.The
Urajiro river is a tributary of the Yoro river.The catchment1ies near the1atitude35.
18’and1ongitude140.08’in the Chiba prefecture.The catchment e1evation varies from
284m to55m and the areas of0,15sq.km.up to the Kakinokidai and9,04sp.km.up to
the Tsukizaki gauge&discharge measurement sites were mode1ed.The Urajiro basin
is in humid c1imate,the normal amua1rainfa11is1820mm.In some years.the basin a1so
experiences very sma11amount of snow which usual1y me1ts within a few hours.The soi1
is coarse,most1y vo1canic ash type having1arge hydrau1ic conductivity.The index map
of the basin is given in Fig,2,
The daiIy rainfa11data of the Kakinokidai station were used.The discharge(at
Kakinokidai and Tsukizaki)are recorded using an ana1og chart-recorder and the values
read at10-minute interva1were used to compute mean-hour1y discharge.These va1ues
were used to obtain the dai1y and month1y discharges.The data for the period1980-84
were used,For estimation of potentia1evaporation from the basins,the meteoro1ogica1
data at the Ushiko station were used.The hour1y data for wind ve1ocity sunshine-hours
and air temperature were extracted from the Automatic Meteoro1ogical Data Acquisi一
22一
Rainfall-Runoff Modelling-S.K.Jain&T.Kishii
工
C
+ ・、。 ■’
φ、
お キ
・。…一 1 ⊂ o ■];hik
o o
“ …
τ5ukiZOk’
手
A1TS]klZOk■
8=Koklnokidoi
C Doiyom0D Yo[ogow0EI…h’z]kq
■1WO-e-eWl gOuging 5-O-lOn
■1Ro lngOuqe 昌t ot1o〔
o 1 2k m
山\ KO…;uuro
O … lO-m
ア。。。Rl”
Chlbo ClW
{
Fig.2 Index map of the Urajiro experimentfal basin showing locations of
rain and discharge measurement sites.
tion System(AMeDAS)database and the dai1y humidity data were taken fromthe Japan
Meteoro1ogica1Agency(JMA)Pub1ications.The procedure for Penman method,recom-
mended by Dorrenbos and Pruitt(1977)was used to estimate PE.Since the Ushiko
station is1itt1e away from the basin(13km),the estimation of PE may be somewhat
unrepresentative.
4.Pammeter estimation
The ca1ibration process requires a procedure to evaluate a given ca1ibration and then
to further adjust the parameters if required The various aspects of ca1ibration of a
conceptual mode1have been discussed by Ibbit and O’Dome11(1973),Johnston and
Pi1grim(1976),and Gupta and Sorooshian(1988).Several criteria are avai1able in the
1iterature to test the efficiency of a rainfa1l-runoffmode1,see Nash&Sutc1iffe(1970)and
Garrick et al(1978).
A two-stage process appears to work we11for ca1ibration of the model described
above. The first stage invo1ves matching the vo1umes of observed and simu1ated
一23一
Report of NIED,No.53,March1994
hydrographs on month1y basis.In the second stage,the shape of the simu1ated hydro-
graph may be matched with the shape of the observed hydrograph.This approach gives
a f1exibility to the modeler to adjust the mode1parameters in1ight of the objectives of
the study,e g.,whether peak f1ow mode11ing is more important or low f1ow mode11ing.
Out of the ten parameters,four are time constants of various reservoirs which do not
affect the vo1ume of the simulated hydrograph(except the time constant of groundwater
reservoir)and hence need not be ca1ibrated in the first stage, Therefore,on1y six
parameters,name1y Sm、。,Cm、、,FC,Fi,f,Cm、。,and Ewf were determined.The Rosen-
brock method,which is a search technique,was used for optimization.The objective
function adopted was
Min Z=Σ(VO。一VS。)2 (6)
where,
VOt=Vo1ume of observed hydrograph in mm for month t,
VSt=Vo1ume of simu1ated hydrograph in mm for month t.
5. 正己esults
The vo1umes of observed and simu1ated discharge hydrographs at Kakinokidai on
month1y basis for the period Jan.1980to Nov.1984have been p1otted in Fig.3.The
graph shows a reasonab1y good match between the observed and the computed hydro-
graphs There appear to be some inconsistencies in the input data for a few months-the
discharge coefficient(vo1ume of discharge/vo1ume of rainfa1l)was far greater than one.
The rain gauging station whose data were used(Kakinokidai)is1ocated in the1ower
altitude region and therefore the rainfa11estimations may be on1ower side.Moreover,
for this catchment,the possibi1ity of exchange of f1ow with the neighbouring areas in the
subsurface zone cannot be ignored.
The pIot of observed and simulated hydrographs for the Tsukizaki station for the
period Jan.1980to Nov.1984is given in Fig.4.A reasonably good match was a1so
obtained for this station too.The catchment up to Kakinokidai is a hi1Iy area whi1e the
major part of catchment up to Tsukizaki is re1atively flat.It is also seen from the graph
that the computed discharge peaks are general1y on the higher side of the observed
discharge up to1982whi1e they are on the1ower side subsequent1y.This might have been
caused by some catchment change.
24
Rainfan-Runoff Mode11ing-S.K.Jain&T.Kishii
mm/hr10
8
6
4
2
0
mm/mon400.
Rainfa-1工 a1二 Kakinokidai
Month1 Dischar e at KAKINOK工DAI
320.00
240.O O
160.00
80.00
0.00
一 Sim- Obs
1980 1981 ユ982 1983 1984
Fig.3 PIot of observed and simu1ated hydrographs at the Kakinokidai
station.The corresponding rainfan has been p1otted in the upper
graph.
Table1 Optimum Mode1mentS
Parameters for the Catch一
Parameter Up to Kakinokidai Up to Tsukizaki(Area0.15sp km) (Area9.04sp km)
Smax 47.94 60.50
Cmax 42,00 38.75
FC O.19 O.23Finf 0.72 O.87Cint O.48 0.55
Fwf O.81 0.88
CKG 400.OO 400.00
Obj Fun. 35,978. 49.O00.
Model Eff. 0.85 O.69
一25一
Report of NIED,No.53,March1994
mm/hr10 OO Rainfa11 at Kakinokidai
00
00
00
00
o o
㎜/mon400.O O Month1 Dischar e at TSUKI ZAKI
320.O O
240.00
160.O O
80.00
0.00
SimObs
Fig.4
1980 1981 1982 1983 1984
P1ot of observed and simu1ated hydrographs at the Tsukizaki
station.The corresponding rainfa11has been plotted in the upper
graph.
Overall the resu1ts are acceptable.It may be mentioned that no attempt was made
to match the shape of the hydrographs because short term data were not avai1ab1e.The
values of the various mode1parameters for the two basins are given in Tab1e.1.
6. CoIlclusions
A simpIe conceptua1mode1has been used to simu1ate the rainfa11response of two
sma11research catchments,The resu1ts are reasonab1y good for both the catchments.
一26一
Rainfall-Runoff Modelling-S.K.Jain&T.Kishii
7.AckIlow1edgemets
The authors are thankful to Dr M.Maki,of NIED,for providing the AMeDAS data
which were used to estimate the potentia1evaporation for the Urajiro catchment and to
Ms.Y.Aoki and Ms.K,Endo for assistance in data processing.
References
1)B1ackie,J.R,and C.W.O.Eeles.(1985)Lumped catchment models,in Hydrological Forecasting,M.
G.Anderson and T.P.Burt(ed),John Wi1ey and Sons,New york.
2)Ciriani,T.A.,Maione,U.,and J.R,Wallis.(1977)Mathematical Models for Surface Water
Hydrology,John Wiley&Sons,London.
3)Dorrenbos,J.,and W.O.Pruitt(1977),Guidelines for Determining Grop Water Requirements,FAO
Irrigation and Drainage Paper No.24,Rome.
4)Garrick.,M.,C.Cmnane,and J.E,Nash.(1978)A criterion of efficiency for rainfall-runoff model,
Journa1of Hydro1ogy Vol.36,375-381.
5)Gupta,V-K一,and S,Sorooshian.(1985),The automatic calibration of conceptual catchment models
using derivative-based optimization algorithm.Water Resources Research,2!(4),473-485.
6)Ibbitt,R.P一,and T,O’Dome1l.(1971)Fitting methods for conceptua1catchment mode1s,Joumal of
Hydraulic Division,ASCE,97(HY9),133ユー1342.
7)Johnston,P,R.,and D.H.Pi1grim.(ユ976)Parameter optimization for watershed models.Water
Resources Research,12(3),477-486.
8)Nash,J-E.,and J.V.,Sutcliffe.(1970)River f1ow forecasting through conceptua1mode1s,Part1:A
discussion of principles,Joumal of Hydrology,VoL10,382-290.
一27一