Analysing effects of climate change and forestry management on water yield using conceptual models

Transcription

1 Man's Influene on Freshwater Eosystems and Water Use (Proeedings of a Boulder Symposium, July 1995). IAHS Publ. no. 230, Analysing effets of limate hange and forestry management on water yield using oneptual models HUAXIA YAO, MICHIO HASHING & HEROMU YOSHEDA Department of Civil Engineering, The University of Tokushima, 2-1 Minami-josanjima, Tokushima 770, Japan Abstrat A physially-based hydrologial model is set up to analyse the effets of land use alteration and limate hange on water resoures. The integration of water and energy transfers is emphasized by a feedbak solution algorithm, and nonuniform-nonlinear features are inluded in onsidering soure area variability, infiltration and soil drainage. The model is first tested on a small experimental site in China using a day-byday iteration, with 30-days runoff deviations falling within % of the observed for ases of unultivated and forested land. The results suggest that afforestation might redue runoff by % and raise evaporation by %. The model was improved and tested on the Shirakawadani basin in Japan using an hourly iteration. This yielded a tiny error of 1.65% for the annual water yield and a good determination oeffiient (0.976) for daily series. Hydrologial proesses are shown to be most sensitive to temperature, rainfall and radiation. Climate hange ould inrease either annual evaporation or runoff by 15.6% or 6.3% respetively. Monthly runoff hanges are shown to range from 1% to 15.6%. INTRODUCTION In water resoures and land use management, physially-based hydrologial models have proved to be of great value for larifying the mehanisms of hydrologial hange aused by human ativities and for providing information for management. The problems of global limate hange and regional forest redution are being studied at multiple sales but loal or regional phenomena appear to be most sensitive (Bruijnzeel, 1990). In this study, a physially-based oneptual model is proposed and applied to one site in entral China and another site in western Japan. MODEL DEVELOPMENT A basin is usually assumed to be a system of reservoirs or layers regulating the water yle, whih is oneptually homogeneous in the horizontal dimension but differs in the vertial dimension (Yao et al., 1994). The struture of the model, with seven layers, is shown in Fig. 1(a). Net radiation and preipitation provide the inputs to the model. These are then transported or transformed in and between these layers. At the same time, heat-vapour fluxes

2 AIR CANOPY H2 Huaxia Yao et al. evapotranspi ration J> uptake < SURFACE RETENTION > H2 TOP SOIL ROOT SOIL IS > ( BOTTOM SOIL ) overland R, outflow R, outflow R. C AQUIFER y baseflow R runoff R (a) Fig. 1 Model struture: (a) flux hart; (b) variable soure area; () drainage width and length. () and runoff are the outputs. Four layers (boundary atmosphere, tree anopy, topsoil and root soil) are signifiant both for heat and water transfers, while the other three layers (surfae retention, bottom soil, and ground aquifer) are effetive merely for the water yle. Energy budget equations Aording to most models andfielddata already established, the energy budget in a soilvegetation-atmosphere system is mainly ontrolled by net radiation, sensible and latent heat fluxes (Grob, 1993). Net radiation R n is separated into two parts: R n (l e"** - ) ontributes to the anopy layer, and R n e' kl to the topsoil layer, where L is the leaf area index and k is the anopy's weakening fator. Then the energy budget in the anopy layer is written as: R n (l-- kl ) = \(E C + T r )+Hl (1) where E is evaporation from interepted water storage during a time interval, T r is the transpiration of the trees, and Hi is the sensible heat flux from anopy to air layer. It is aeptable that whenever intereption water exists, transpiration is far less than leaf surfae evaporation, so only E is inluded in that ase. Otherwise T r is onsidered alone. Heat fluxes may be expressed following the priniple of aerodynamis as: E = P L^ar q a )lr a HI = p p L(T r T a )lr a

3 Effets of limate hange and forestry management on water yield 239 in whih q al and q a are speifi air humidities on a leaf surfae and above the anopy, T t and T a the temperatures of the anopy and the air above, r a the turbulent flux resistane and r sl the stomatal resistane. The energy budget at the soil surfae and in the topsoil is expressed as: R n e' kl = \(E SS+ EJ+H2 (2) where E ss is the evaporation from surfae water, E sl is the evaporation from the topsoil, and H2 is the sensible heat flux from soil to atmosphere. Heat fluxes depend on the gradient of humidity or temperature, wind speed and the topsoil water ontent (W r ), i.e. K = P(<7 ~ <la)'r a -W T ) and H2 = p p (T s - T a )lr a. Water balane equations Aording to the law of fluid ontinuity, eah layer in the model gives a orresponding water balane equation. Taking any layer into onsideration, let F be the inflow amount, R the outflow, W 0 the initial water storage in the layer, and W e the storage at the end of the interval. Then the water budget is: W e -W 0 = F-R (3) Inflow F and outflow R are determined by a representative storage status W, whih is neither the initial W 0 nor the end W e. Water storage and movement may show nonlinear or nonuniform harateristis beause of atastrophi rainfall, flutuating radiation or the vegetation-soil heterogeneity. Of ourse, different layers have different relationships between W and F or R. Canopy intereption storage W Ce = Wo + vp-e-p* (4) where P is the gross rainfall, a is the proportion of forest over, P^ is the sum of drip and stemflow. Evaporation E (or P ds ) relates to representative W. Surfae retention storage ( W D ) Overland flow is estimated by a Hortonian formula: R d = k d W D e < k d = k M + k P x (5) in whih the routing oeffiient k d is related to rainfall, emphasizing the nonlinear rainfall pattern. Here a variable soure area is used to reflet nonuniform water storage on the surfae (see Fig. 1(b)). Then W D, R d, E ss and I n are related to the soure area A d whih is determined by W D, total basin area A, and the reession-beginning storage depth W DA, as A d = A Wp/W^. Infiltration I n at aieaa d is usually different from infiltration I a at area A - A d, and we have: la = V,WW r )«/ = tf+rj ( 6 ) where V t is the topsoil water ondutivity, W T and W Ts the representative and saturated ontents of the top layer, I r the infiltration into the root layer, and R ts the saturated

4 240 Huaxia Yao et al. throughflow i.e. R ls = V t sm<j> (D t + H,)B/A. Here 4>, B, H, and D, are the slope angle, draining width, thikness and longitudinal length respetively, as shown in Fig. 1(). Finally, the water budget is: where P x is the net rainfall. W De = Wm+Px-Ess-hx-R* (7) Topsoil storage (W 7 ): W Te = W m+ 1 [ -E S[ (A-A d )/A-I r -R l (8) where the area-averaged infiltration I, and atual throughflow R, are obtained from equation (9) (W T j-being soil water field apaity). j - j d +j a _ R, = R«1 -exp w T-w Tf -. ~ n ~w -w VY Ts YYJJ Root soil storage (W R ) and bottom soil storage (W B ) W Re = W m + I r -T r -I b -R r (10) W Be = W B0 + l b -P b (11) in whih I r and I b (infiltration to bottom layer) are alulated in the same way as I a, and lateral drainage R r and vertial drainage P b are similar to R r There is no lateral drainage in the bottom layer. Beause the stomatal resistane onnets transpiration T r with soil water W R and evaporation E st also depends on W T, the energy budget group of equations (1) and (2) and water budget group of equations (3)-(l 1) are therefore ombined. All of these equations have to be simultaneously solved. Furthermore, the representative W T, W R and W B themselves are ontrolled by the desired W Te, W Re and W Be whih are unknown. So, the above mentioned proesses suh as heat fluxes, soil inflow and outflow are derived with a speial feedbak-iteration method (Hashino et al., 1994). (9) Temperature and runoff Canopy and topsoil temperatures for eah step are written as: T, = r a /(p p L)[R n (l-t-' L )-\(E + T r )] + T a (12) T, = r a /(p p )(R n e- kl -\E s ) + T a (13) in whih the mean evaporation rate E s = E SS -AJA + E SI (A - A^/A. Groundwater flow is linearly expressed as:

5 Effets of limate hange and forestry management on water yield 241 R g ' being the flow of the former interval Then total runoff is available now: R =R d +R [ + R r +R g (14) (15) APPLICATION: PREDICTION OF AFFORESTATION EFFECTS The Shiqiaopu experimental station on water and soil onservation is loated in a hilly area of the Notian County, Hubei Provine of China, at latitude 30 42'N and longitude 'E. The observation plot seleted for use is of 3 km 2 area, on average m wide, m long, and on a 31 slope. The sandy loam soil is underlain by granite bed rok. Annual temperature, preipitation and runoff are 16.7 C, mm and mm, respetively. Regarding land use, the small basin was unultivated land with sparse grasses overing only 11.0% of the ground, during period After that time, plantation forestry was introdued and a mixed forest stand of pine, oriental oak (Querus variabilis) and Chinese fir had been formed by the year Tree over rose to 52.8%. In the model, apart from physial onstants or oeffiients, there are 38 parameters. Twenty-three of these are diretly determined from on-site investigation and some published experiment results, and the important fifteen ones are determined by an optimization tehnique, SIMPLX. As for Shiqiaopu, three points need to be explained: the topsoil is too thin to be onsidered; groundwater flow in total streamflow is negligible; and the time interval of alulation was one day. Calibration was implemented separately for before plantation and for after plantation. Parameter values relevant to the above two' land use types are listed in Table 1. Runoff over a 30 day period was used to assess model validation for water yield simulation (see Fig. 2(a)). Six tests for the grass ase and eight tests for the forest ase demonstrated deviations of % from the observed runoff. By using parameter values for the pre-plantation period ( ) and running the model again for , it was found that afforestation would derease water yield by % and inrease évapotranspiration by % (Fig. 2(b)). EFFECTS OF CLIMATE CHANGE: SENSITIVITY The Shirakawadani experimental basin is loated about 100 km west of Tokushima City, Western Japan (latitude 33 52'N, longitude 'E). The 23 ha area is distributed with natural broad-leaf trees (oak years old) and artifiial needle-leaf trees (Japanese edar years old). Elevation ranges from m, with a mean slope angle 41, and brown forest soils. Automati observation inludes air temperature, relative humidity, net radiation, wind speed, rainfall, runoff disharge and soil water. Data for January 1991 through November 1992 were used to alibrate parameters. In this ase, the model was run on an hourly basis. Model parameters area listed in Table 1 and outputs of daily and monthly runoff are shown in Fig. 3. Estimates of annual water yield are also satisfatory. Taking the 1991 year, the mm rainfall

6 242 Huaxia Yao et al, Table 1 Calibrated parameters of the model for two sites. Parameter Shiqiaopu: Shirakawadani ff (m) H z (m) /i(km 2 ) s a A s L 5 m (mm) Zo(rn) d(m) k W Ts (mm) W T y (mm) W Tw (mm) W& (mm) W Rf (mm) W Rw (mm) W & (mm) W B/ (mm) 'A/m ( m ) Grass Forest Forest See text for key to symbols Parameter Shiqiaopu: Shirakawadani tsmin ( ffl ) '/w ( m ) K optimized H, (mm) H r (mm) H b (mm) V, (mm h 1 ) V r (mm h" 1 ) VjflMh- 1 ) C d C k n B(m) D,(m) D r (m) K Grass Forest Forest o 200 -= 400 E 600 IT C ouu 600 (a) Observed Simulated = 200 n rl nn, B1 B2 B3 B4 B5 B6 (b) Evaporation A1 A2 A3 A4 A5 A6 A7 A8 u Runoff _CH f-» T A1 A2 A3 A4 A5 A6 A7 A8 Bl 6 June - S July 1966 B J uly 1966 B3 17 April - 2 May 1969 B4 BS 27 June - 26 July May 25 June 1970 B6 26 June - 25 July 1970 Al 11 Mav 9 Jnue 1979 A2 10 June - 9 July 1979 A3 12 May - 12 June 1980 A4 8 Julv - 7 August 1980 AS A6 17 June 18 July July 20 August 1981 A7 14 June - 13 July 1982 A8 28 June 30 July 1982 Fig. 2 (a) Parameter tests at Shiqiaopu before (B) and after (A) plantation; (b) afforestation influene.

7 Effets of limate hange and forestry management on water yield 243 ^ o 100 E r e I M I Observed Simulated ,2-9.S* Differene (%) M J F M A M J J A S O N D Fig. 3 Monthly runoff at Shirakawadani. is balaned by mm évapotranspiration (43.5% intereption loss, 36.3% transpiration and 2% soil evaporation) and mm runoff (8.9% overland, 15.0% rapid throughflow, 21.8% delayed throughflow and 54.3% baseflow), with a tiny annual runoff error of 1.65%. There is also a good agreement between the simulated and observed runoff series; the hourly series has a orrelation oeffiient of to the observed, and daily runoff has a orrelation oeffiient of Then an attempt was made to estimate the hydrologial influene of limate hange, Base llmate(1991): rainfall , évapotranspiration 515.1, runoff mm # 0) a eu J= O ^3 CO o CO 5» 111 o O) a -C o À H ; y I I a b d e f g h l l - ABCDEFGH Senario Single Change Multi-Change a b d e f g h l I A B C D E F G H T0 ( C) P Rn (%) H, (%) temperature preipitation net radiation air humidity wind speed leaf-area-index stomatal resistane L ra <*> Fig. 4 Hydrologial sensitivity to limate hanges.

8 244 Huaxia Yao et al. under a bakground of global greenhouse warming. Long-term future limate regimes an not be universally predited by any numerial system available and limate hange at the loal sale is also unertain. Nevertheless, the sensitivity of hydrology to a range of limate hange senarios should be assessed for the sake of future resoures strategies. Aording to literature reviews (Waggoner, 1990; Japan Meteorologial Ageny, 1990), possible variations of meteorology and plant growth in a 2 x C0 2 atmosphere may be desribed by single or ombined fators, suh as temperature, rainfall and radiation. Runoff response to eah limate senario was alulated by applying the model to the Shirakawadani (see Fig. 4). It was found that temperature, rainfall and net radiation are the most sensitive foring fators for hydrologial proesses. A single-type senario ould inrease annual évapotranspiration by 8% and runoff by 3%, and a ombined-type senario ould raise these up to 15.6% and 6.3 % respetively. This basin has high rainfall and a humid sub-tropial limate, so that the other fators of air humidity, wind speed, leaf area and stomatal resistane show little ontribution. A more reasonable senario (Seino, 1992) was also applied. The seasonal distribution of three main fators in limate hange were gained from an improved GISS-GCM model. Corresponding influenes are a 13% inrease in annual evaporation and a very small derease in annual runoff. But the response of monthly runoff flutuates between 1 and.6% (Fig. 5) beause the predited rainfall differs season by season, and inreased annual rainfall just ompensates for the enhaned evaporation demand. In onlusion, global warming in the next entury will probably bring about sustainable effets on water yield, plant water requirements and the seasonal omposition of the water balane. Climate Change Senario DJF MAM JJA SON Year T a ( C).0 P(%) -4.0 R (%) JFMAMJJASOND Fig. 5 Possible hange under 2 X C0 2 limate senario. Y REFERENCES Bruijnzeel.L. A. (1990) Hydrology ofmoist Tropial Forests and Effets of Conversion: A State of Knowledge Review. A Publiation of the Humid Tropis Programme, IHP, UNESCO. Grob, G. (1993) Numerial Simulation of Canopy Flows. Springer-Verlag, Germany. Hashino, H., Yoshida, H. & Yao, H. (1994) An hourly model for energy and water yle in forested athment. In: Pro. International Symposium on Forest Hydrology (Otober 1994, Tokyo) (ed. by T. Ohta), Japan Meteorologial Ageny (1990) Climate Changes as Greenhouse Gases Inrease (in Japanese). Press of Ministry of Finane,Japan. Seino, H. (1992) Probable effets of CO, limate warming on water resoures of rop prodution. In: Pro. Workshop on the Effets of Global Climate Change on Hydrology and Water Resoures at the Cathment Sale (February 1992, Tsukuba), Press of Ministry of Constrution, Japan. Waggoner, P. E. (1990) Climate Change and U.S. Water Resoures. John Wiley & Sons In. Yao, H., Hashino, M. & Yoshida, H. ( 1994) Analyzing effets of deforestation and afforestation on streamflow by using a physially-based oneptual model. J. Japan So. Hydrol. and Wat. Resour. 7(3), 1963.