on The bank of_The,Extent,of,Land,Use,Impact,on,Water,Regime,in,the,Vseminka,Catchment

　　Pavel Kovar and Darina Vassova 　　Faculty of Environmental Sciences, Czech University of Life Sciences, Prague 16521, Czech Republic
　　Received: July 1, 2011 / Accepted: August 26, 2011 / Published: February 20, 2012.
　　Abstract: The paper deals with the impact of land use changes on water regime. An assessment was carried out in order to determine the extent to which the main components of the water balance on the experimental catchment Vseminka have been influenced by land use changes (region Vsetinske Hills, the Czech Republic). For this reason, the water balance model WBCM-5 was implemented for the period of 30 years in a daily step, with particular focus on the simulation of the components of direct runoff and of subsurface water recharge. In the selected years of the period 1980-2009, major changes were made in land use and significant fluctuation of rainfall-runoff regimes were observed (e.g. dry year 1992, flood year 1997 and normal year 2009). After WBCM-5 parameter calibration it was observed that some water balance components can change in relation to substantial land use changes, even up to dozens of percent in a balance-consideration, i.e. in daily, monthly and yearly or decadal values, specifically as far as the components of interception and also of direct runoff and of subsurface water recharge are concerned. However, a different situation appeared during the investigation of significant short-term rainfall-runoff processes. There were about seven real flood events during the same period on the same catchment which were analysed using the KINFIL-2 model (time step 0.5 hr). Land use change, positive or negative scenarios, were also analysed during this period. As opposed to long-term water balance analyses, only a 10% difference in the hydrograph peak and volume was observed. In summary, the authors have shown that it is always important to distinguish a possible land use change impact on either long-term balance or short-term runoff. Otherwise, as often found in over simplified commentaries on flood events in the mass media, the actual impact of land use changes on water regime may be misunderstood.
　　Key words: Land use change, water balance, rainfall-runoff event, hydrological model.
　　 1. Introduction??
　　How is the water regime influenced by land use change? This important question is often discussed by hydrologists [1-7] as well as by the public, particularly after flood events. Therefore, it is quite important to further investigate this problem from the scientific point of view, particularly when the common EU policy includes plans for massive land use changes in the years to come.
　　The most important factors affecting direct runoff are rainfall depth and rainfall duration. The other factors are climate conditions, physiographic conditions, land use etc.. The authors wanted to determine which of these factors influence the rainfall-runoff process to a greater, or to a lesser degree. Their investigations specifically addressed land use change. Two fundamental approaches were applied:
　　? water balance analysis of long-term period of rainfall, runoff and free water evapotranspiration on the Vseminka experimental catchment in the vegetation periods (April 1 to October 31) 1980-2009 with particular emphasis on land use change on significant water balance components;
　　? rainfall-runoff analysis of isolated short-term events investigating the influence of land use changes on direct runoff hydrograph, its shape and peak.
　　Both approaches were applied through the implementation of the adequate mathematical models. The aim of this case study was to assess the impact of land use on water retention and accumulation capacities of the tested catchment. The tools which were used in the analysis were two models: the WBCM-5 model of water balance [7] and the KINFIL-2 model of short-term rainfall-runoff events[8, 9]. Both models respect spatial variability of the model parameters within the GIS environment. The partial aims were:
　　? to determine the physiographic and vegetation factors of the catchment and thus to identify the parameters of each model;
　　? to select observed rainfall-runoff time series of water balance as well as of the significant short-term events for the models parameter calibration;
　　? to simulate the time series of the existing processes and to validate the parameter values;
　　? to simulate the selected “scenario situations”characterizing land use changes;
　　? to compare hydrological regimes of actual and scenario situations.
　　 2. Material, Methods and Results
　　 the Vseminka catchment were described in a paper which has been published in previous years [8, 10]. Other characteristics, such as the main physiographic characteristic, hydrological soil groups and their physical properties are provided in Table 2. The land use mosaic pattern is given in Fig. 1.
　　The catchment geology belongs to the Tertiary Carpathian formation, with a dominant flies structure(changing sandstones and gneiss). When weathering, it forms silty soils with clay gneiss. There are mostly cambisols generating on this geological substrate. The hydraulic properties of soils are briefly described in Table 3.
　　2.2 Water Balance Modelling
　　The simulation of water balance on the Vseminka catchment for the sake of water regime dependence on land use change was implemented by the WBCM-5 model [10, 11]. The structure of the model is physically based with mathematical description of the mutually interconnected hydrological processes in the time step ?t = 1 day, as follows:
　　? potential evapotranspiration, PE;
　　? interception, AIR;
　　? direct runoff generation and its transformation,
　　SOF;
　　? soil moisture dynamics of the upper soil zone(root zone SMC);
　　? unsaturated zone dynamics, ?WP, and actual evapotranspiration, ?E;
　　? saturated zone dynamics, ?WZ, groundwater flow, GWF, total flow, STF.
　　Volumetric fit of all water balance storages is controlled through the water balance equation:)
　　SRAIN?????? (1)
　　where SRAIN is rainfall depth (mm), AE actual evapotranspiration (mm), STF total runoff (mm), ?WP change of soil moisture content of unsaturated zone(mm), ?WZ change of subsurface water, there: ?W =?WP + ?WZ (mm).
　　The core of WBCM-5 model is unsaturated soil water dynamics (its filling and exhaustion) which is described by the Richards equation [6], numerically solved in the form of finite differences:
　　(2) where θ is soil moisture (–), H is pressure height (m), z is depth of infiltration front (m), t is time (s) and K(θ) is unsaturated coefficient of hydraulic conductivity(m·s-1).
　　Three major data groups are required for the WBCM-5 model:
　　(1) Hydrometeorological data: Observed daily values of rainfall and runoff and daily free open water evaporation (or data for potential evapotranspiration). Daily rainfall depths, at least, 5 days before water balance accounting are needed, too.
　　(2) Hydrological assessment of the “range values”of porosity, field capacity of root zone and whole unsaturated zone, their depths, saturated hydraulic conductivities and sorptivity values (in transects or downscaled using geo-statistical methods). CN values (USDA NRCS) help, too.
　　(3) Land use data: Agri and forest data on land use, types of crops and forest (aging and structure), cropping pattern, catchment management.
　　There are 12 WBCM-5 parameters. Six of them have a clear physical value (areas, depths, soil hydraulic parameters). The other two parameters can be automatically optimised through the Rosenbrock procedure [12]. These are the maximum capacity of both unsaturated (SMAX) and saturated (GWM) zones in mm. The remaining four parameters are adjusted according to the real situation, or according to the scenario situation, in order to reflect the land use. These are as follows:
　　? DROT depth of root zone (mm);
　　? WIC upper limit of interception capacity (mm);
　　? CN Curve Number (-) (USDA NRSC);
　　? BK slope of the master depletion curve (log scale in days).
　　For the Vseminka catchmnet the daily data from the vegetation period of the years 1980, 1995, 1997 and 2009 were selected to present water balance affected by land use change. The seasons 1995 and 2009 were at upper (1995) and at lower (2009) levels of a normal range, 1997 and 1980 were wet and dry seasons, respectively. Daily rainfall and runoff data were processed from the Slusovice station, daily water evaporation data from the Vizovice station. Land use maps were elaborated from the 1: 10,000 (ZABAGED) scale system through GIS (ArcInfo). Table 4 provides the results of all important water balance equation components for all seasons mentioned above.
　　Table 5 provides the method of the parameters adjustment. SMAX and GWM parameters were adjusted automatically while four others were arranged to reflect their physical meaning when land use was changed. The criterion for computation of this water balance was chosen according to observed and computed decadal (in 10 days step) surface runoff, given by coefficient of determination which equals to the Nash-Sutcliffe coefficient (RE) broadly used in hydrology [13]. The goodness of fit, evaluated by coefficient of determination RE (i.e. Nash-Sutcliffe efficiency coeficient) and the coefficient of variation PE, is evaluated for the best fit RE → 1.0 and PE →0.0 (both RE and PE are dimensionless). The values of RE in this study are never less than 0.87, which shows still the acceptable fit [14]. Every season was documented with a decadal subtracting graph (rain minus evaporation, minus runoff, plus/minus change in subsurface water). Fig. 2 provides the water balance components thus presented for the season of the year 1995 (as a higher “normal range”).
　　The WBCM-5 model was subsequently used for the simulation of two scenarios, representing the changes of land use on the Vseminka catchment, with data of flood year 1997. The changes of main water balance components, i.e. total runoff STF, direct runoff SOF, actual evapotranspiration AE, total interception AIR, and changes of subsurface water storage ?W, have been published earlier [12]. There are only three basic scenarios of land use change in this paper, which are as follows:
　　? Scenario 0: The existing situation according to the measured hydrometeorologic data on daily values of rainfall, free water evaporation and runoff at the outlet.
　　? Scenario A: The situation of the Vseminka catchment after 30% deforestation when cutting forest on slopes steeper than 20% (e.g. wind disaster).
　　? Scenario E: The target situation, where 124 ha of arable land are grassed and 107 ha of meadows are afforested.
　　Table 6 indicates the value of the main components of water balance in the existing situation (calibrated values) and in both scenarios (A, E) for flood year 1997. The new parameter values of the scenarios A and E introduce the alternation in CN-values, according to the USDA NRSC Method [15, 16], WIC-values and BK-values according to our former experiences [11]. Other parameter values remain unchanged (SMAX, GWM, DROT).
　　Table 7 gives the land use areas for the scenarios A and E. Table 8 provides the values of water balance components for both scenarios (A, E) reflecting the existing situation 1997 (Scenario 0).
　　The changed values of the most important water balance components, i.e. direct runoff SOF and subsurface water storage ?W, are provided in a graph in Fig. 3.
　　The results obtained by various implementations of the water balance model WBCM-5 clearly indicate that the total flow STF and direct flow SOF are in logical relationships with land use changes; scenario A varies up to +22%, and scenario E as much as -29%, in each case when compared to the existing status, i.e. scenario 0. The practical impact of scenario E is evident in the context of non-structural flood measures. The changes of subsurface water storage ?W are also important, i.e. from the water balance viewpoint, when the variance between the preferred scenario E, and least-preferred scenario A, differs by up to 65.5 mm. On the other hand, the values of actual evapotranspiration AE do not differ significantly (to 7.6 mm only) because of the non-homogenous character of the catchment.
　　2.3 Isolated Rainfall-Runoff Events
　　Another tool in the form of the KINFIL-2 model was used for the simulation of short isolated rainfall-runoff events on small catchments. This model is based on the solution of infiltration process, using the Green-Ampt approach according to the Morel-Seytoux method [17]: where Ks is saturated hydraulic conductivity (m·s-1), zf is depth of the infiltration front (m), θs is saturated soil moisture content (-), θi is initial soil moisture content (-), Hf is saturation pressure below the infiltration front (m), i is rainfall intensity (m·s-1), Sf is storage suction factor (m), tp is ponding time (s) and t is time (s).
　　The basic task, when implementing the KINFIL-2 model, is the determination of soil parameters Ks and Sf (at field capacity FC value) which correspond to their conceptual values CN = f (Ks, Sf) [17, 18]. The second component of the model is its runoff part, which simulates the propagation and transformation of direct runoff by kinematic wave:
　　(6) where h, t, x are ordinates of depth, time, and position(m, s, m), respectively, α, m are hydraulic parameters(-) and re is effective rainfall intensity (m·s-1).
　　Eq. (6) is a core element of the direct runoff component of the KINFIL-2 model, and in its programmed version it is solved using the Lax-Wendroff explicit scheme [19], after transforming it into the finite difference form. Both models, KINFIL-2 as well as WBCM-5, were used for the reconstruction of the rainfall-runoff events of the year 1997. However, as opposed to the WBCM-5 model implemented throughout the whole year, the KINFIL-2 model was only used for two significant flood events in July 1997. Table 9 provides the most important characteristics of both flood waves, when antecedent catchment saturation was on the 3rd degree, i.e. on the field capacity of the upper soil layer.
　　The resulting hydrographs of both reconstructed flood waves, which were assessed by the KINFIL-2 model, are provided in Figs. 4a and 4b. The fit of measured and computed hydrographs is very close. This fact indicates that the model parameters were well calibrated. The high degree of fit has encouraged the next step, which was a scenario simulation. For this purpose the same scenarios, i.e. A for 30% forest cut and E for target optimal land use as in the WBCM-5 model, were implemented. Table 10 shows the changes of the KINFIL-2 parameters, reflecting the corresponding changes in land use on the Vseminka catchment.
　　Besides the reconstructed hydrographs, Figs. 4a and 4b also provide the designed scenario hydrographs deterministically, corresponding to the changed land use on the Vseminka catchment. The changed discharges, in coincidence with the analyses made on other experimental catchments, are not significant and they do not exceed the differences higher than 10% in hydrograph peaks, as well as in runoff volume.
　　 3. Discussion
　　This paper is one of the series of experiments focused on scenario simulations of land use and their rainfall-runoff reactions. Short-term flood events investigated on several experimental catchments in the Czech Republic, particularly on the Vseminka, were tested by the KINFIL-2 model with the short time step?t = 0.5 hr. None of the realistic land use changes, including very robust impacts (30% deforestation), caused the increase of peak discharges usually of less than 10% on more than 30 evaluated events. There were a few exceptions (three only) of between 10% and 12% increase on the catchments Vseminka and Rusava. All “positive scenarios”, representing the change of arable land to permanent grassland or afforestation, might decrease peak discharges to 5%-10%.
　　Long-term water balance simulations, with the time step ?t = 1 day on the same catchments, have shown that the major water balance components usually change depending on land use changes more significantly, even in the order of dozens of percent. This is also due to long-term water balancing observation (usually one season or one year, at least). In particular, total runoff and specifically direct runoff components are sensitive to land use changes in a long-term scale. Similarly, infiltration, recharging subsurface water storage is usually in a reciprocal relation with direct runoff. These two last components are very important as a criterion of positive or negative impact of land use change. Furthermore, it is very logical in water balance accounting that the interception component generally depends on the vegetation stage and on the quality of the canopy. Therefore, deforestation, specifically before a substitutive weed-herb canopy has grown, affects actual evapotranspiration.
　　By using both models, KINFIL-2 and WBCM-5, it has been found that the changes of the major balance components in the tested periods were much greater than those in short flood events in reaction to land use changes. In particular, direct runoff in a long-term balance scale can be affected by expressive land use even to several dozens of percent, while in the short-term, flood events as mentioned above, to a maximum of only about 10%. Therefore, it is necessary to clearly determine whether runoff of events, or of balance-periods as a reaction of land use change, are to be evaluated separately.
　　 4. Conclusion
　　Using the WBCM-5 model, with daily data, extreme hydrological situations were clearly identified. If implemented on-line, this can give early warning against floods or droughts.
　　When the model parameters are well calibrated in a real simulation then their physically acceptable alteration can likely quantify the impact of land use on water regime. This can be done also in a quantification of water balance components.
　　The combination of long-term models (e.g. WBCM model), simulating water balance, and short-term models (KINFIL, HEC-HMS, etc.), simulating a short event, are appropriate tools for a comparative analysis of the land use change impact on water regime.
　　Good and diversified land use with a high percentage of forest and permanent grassland can have a positive influence on direct runoff formation, infiltration increase and thus on the retention capacity of a catchment. Obviously, one should analyse impacts of land use change in a short-term (events) and in long-term (balance) time scale separately, when assessing a catchment vulnerability and sustainability.
　　Acknowledgments
　　The authors hereby wish to acknowledge the support given to the project by the Czech Ministry of Education, Youth and Sport, through the Research Grant MSMT 2B06101 “Optimization of agricultural and flood-plain landscape in Czech Republic”.
　　 References
　　[1] P. Kovar, P. Pech, Modelling floods with respect of changes in land use on small catchmnets, in: Proceedings of the International Conference “Interpraevent 96”, Germany, 1996.
　　[2] M. Falkenmark, L. Andersson, C. Yapijakis, Water, a reflection of land use, Swedish Natural Science Research Council of UNESCO, Stockholm, Sweden, 1999, p. 128.
　　[3] EU Water Framework Directive 2000/60/EC of the European Parliament and of the Council, EU WFD, Strassbourg, Oct. 23, 2000, p. 96.
　　[4] V.T. Chow, Applied Hydrology, McGraw Hill, New York, 2000, p. 582.
　　[5] R. Lal, Integrated Watershed Management in the Global Ecosystem, CRC Press, Boca Rator, 2002, p. 395.
　　[6] K.J. Beven, Rainfall-Runoff Modelling, The Primer, John Wiley & Sons, Chichester, 2006, p. 360.
　　[7] P. Kovar, D. Vassova, Impact of arable land to grassland conversion on the vegetation period water balance of a small agricultural catchment, Soil and Water Research 5(2010) 128-138.
　　[8] P. Kovar, P. Cudlin, M. Herman, F. Zemek, M. Korytar, Analysis of flood events on small river catchment using the KINFIL model, Journal Hydrology and Hydromechanics 50 (2002) 157-171.
　　[9] P. Kovar, V. Kadlec, Use of rainfall-runoff model KINFIL on the Hucava catchmnet, Zpravy Lesnickeho Vyzkumu 53 (2008) 211-222. (in Czech)
　　[10] P. Kovar, P. Cudlin, J. Safar, Simulation of hydrological balance on experimental catchment Vseminka and Drevnice in the extreme periods 1992 and 1997, Plant, Soil and Environment 50 (2004) 478-483.
　　[11] P. Kovar, P. Cudlin, M. Korytar, F. Zemek, M. Herman, Comparative study of water balance on experimental catchments Vseminka and Drevnice, Rostlinná Vyroba 47(2001) 260-266.
　　[12] H. H. Rosenbrock, An automatic method for finding the greatest or least value of a function, Computer Journal 3(1960) 175-184.
　　[13] J. E. Nash, J.V. Sutcliffe, River flow forecasting through conceptual models, Part I, A discussion of principles, Journal of Hydrology 10 (1970) 282-290.
　　[14] WMO, Simulated Real-Time Intercomparison of Hydrological Models, WMO Publ. No. 779, Geneva, 1992, p. 241.
　　[15] Urban Hydrology for Small Watersheds, US Department of Agriculture, Soil Conservation Service, US SCS, Washington D.C., 1985, p. 162.
　　[16] National Engineering Handbook-Section 4: Hydrology, Soil Conservation Service, US SCS, Washington D.C., 1986.
　　[17] H.J. Morel-Seytoux, J.P. Verdin, Extension of the Soil Conservation Service, Rainfall-Runoff Methodology for Ungauged Watersheds, Report No. FHWA/RD-81/060, University of Colorado, Fort Collins, 1981, p. 79.
　　[18] P. Kovar, Possibilities of determination of design discharges on small catchmenst using the KINFIL model, Vodohospodarsky Casopis 2 (1992) 197-220.
　　[19] P.D. Lax, B. Wendroff, Systems of conservation laws, Communications on Pure and Applied Mathematics 13(1960) 217-237.