Author(s) First Name Middle Name Surname Role Ana Lúcia Denardin Da Rosa

Transcription

1 This page is for indexing purposes and will not be printed in the conference proceedings book. The text area is picas (~12.3 cm 19 cm) to fit in the printed proceedings book without reduction. When printed on letter paper there will be wide margins. Please see instructions at Ana Lúcia Denardin Da Rosa Federal University of Santa Maria - UFSM (optional) Eng.analucia@yahoo.com.br Eloiza Maria Cauduro Dias De Paiva Federal University of Santa Maria - UFSM eloizadepaiva@gmail.com Repeat Author and boxes as needed. (optional) Repeat Author and boxes as needed. (optional) Publication Information Pub ID Pub Name Pub Date ASABE will complete Put short name of your conference here ASABE will complete

2 Natural flows obtained from a simplified model and the influence of rice irrigation in these flows Rosa, Ana Lúcia Denardin 1; Paiva, ELoiza Maria Cauduro Dias 1. Federal University of Santa Maria - UFSM Abstract. This study aimed to apply and assess the simplified methodology used by Paiva et al (2006 b) in a basin of the Vacacaí River to obtain information about natural flows. The analyses to determine the water demands of rice farming and to sub-divide the basin were based on hydrological information and use of geoprocessing techniques. The study of use and occupation of the soil showed that the flood plains of the Vacacaí River are covered by rice fields that occupy 8.06% of the basin area. Factors such as scarcity of rainfall data as well as uncertainty in the demand value adversity affected the goodness of model fit. However,it was possible to conclude that the methodology was really efficient in determining the natural flow of sub-basins with approximately 3456 km² of area, and that due to factors that were not considered in the model (e.g. spreading into rivers and storage purposes) it was not possible to obtain better results for sub-basins of larger dimensions. Among the ranges of the discretized model, the ones which presented the best results were those with monthly generation and propagation of flow as well as the generation of daily flow and propagation of 7 day average flow, highlighting the monthly simulation due to the smoothing of monthly maximum flows which were not well represented. It was concluded that in areas where the number of rice fields is higher the flows in the fluviometric stations are underestimated, the knowledge about natural flows by managers becomes then relevant. Key words: natural flow, rice irrigation, basin of Vacacaí River Introduction Water at adequate quantity and quality is necessary for the development of a region. Thus, efficient actions are necessary for planning and managing water resources to assure their availability at adequate standards. According to Paiva et al (2006 b), planning is possible only by knowing the quantitative values considering its temporal and spatial distribution scales. To do so, it is important to implement a long term hydrological monitoring system that allows this measurement. Brazil has a large continental territory and consequently it presents high costs for a dense network of hydrological stations. Therefore, it is very common to find a reduced number of stations in a certain region of the country with a series of short term data especially at small hydrologic basins and at regions with poor economic interest, according to Genovez & Pio (1999). Besides the scarce information on river water availability, it is important to notice that water withdrawal actions and the construction of dams interfere in the environment and may impact in the data from the series of registered flow that are used by studies and researches in general. It is very important to know about natural flows because the flows observed at the stations reflect the changes made by man in the environment. In this way, this work aims at determining the natural flows of Vacacaí River Basin applying the methodology developed by Paiva et al (2006 b). 1 Material and Methods The study area is the Vacacaí River Basin (Figure 1) that is located in Rio Grande do Sul State at the coordinates 29º 45 and 30º 45 South and 54º 30 and 53º 15 West, covering a total area of about 11,616 Km². It is inside the agroecological regions of Campaign and Central Depression. Data about the basin soil use are in Table 1. Table 1. Area and percentage of basin soil use classes. CLASS AREA Km² % Water % Developing Crops % Developed Rice % Vegetation % Field % Urban Area % Exposed Soil % Cultivation Soil %

3 Figure 1 Basin Location 2 SMAP Hydrologic Model ( Soil Moisture Accounting Procedure ) The SMAP model, proposed by Lopes et al. (1981) is a deterministic, conceptual, concentrated, rainflow type hydrologic model that has been largely used in Brazil due to its relative simplicity as demonstrated in Figure 2. To apply this model it is not necessary to have large data series and only a few parameters are needed to calibrate it. Figure 2 SMAP Model Structure. Source: Modified by Paiva et al 2006 b In this model, the equations of movement amount are replaced by a series of reservoirs and the passage of water from one to another is calculated from the amount of water in each reservoir (state variable) and from the parameters that represent certain physical characteristics of the basin, Crawford and Linsley (1966 apud OLIVEIRA et al., 2006). To the daily version there are six parameters to be calibrated : (i) Sat (maximum volume stored at soil reservoir in mm), (ii) Ai (initial abstraction in mm), (iii) Capc (soil field capacity, dimensionless), (iv) Crec (groundwater recharge), (v) Kkt (number of days in which the base flow drops to half of its value) and (vi) K2t (number of days in which the direct flow drops to half of its value), two state variables that describe the initial basin conditions: (i) Tuin (initial moisture in mm.mm -1 ) and (ii) Ebin (initial base flow m 3.s -1 ) as well as the Pcoef and Epcoef coefficients used to adjust precipitation and potential evapotranspiration, respectively. To the SMAP monthly discretization we have only two reservoirs, being that the surface reservoir is eliminated due to the fact that the dumping occurs in intervals that are shorter than a month and the field capacity concept used at the soil reservoir is also eliminated (LOPES, 1999). In this version there are four parameters to be calibrated: (i) Sat (maximum volume stored at soil reservoir in mm), (ii) Pes (surface flow parameter, dimensionless), (iii) Crec (recharge coefficient, dimensionless) and (iv) Kkt (number of days in which the base flow drops to half of its value), two state variables that describe the basin initial conditions:

4 (i) Tuin (initial moisture in mm.mm -1 ) and (ii) Ebin (initial base flow m 3.s -1 ) as well as the Pcoef and Epcoef coefficients used to adjust precipitation and potential evapotranspiration, respectively. 3 Natural flows calculation The natural flows calculation was made by applying the simplified methodology suggested by Paiva et. al (2006 b). In this model the basin is simulated as a hydrologic system composed of several sub-basins and characteristic points (CPs) which represent, in a segmented shape, the drainage network. The CP location is at discretion of the user but the choice of the place should take into account the points which represent significant demands, reservoir exit and sub-basin estuary (UFRGS, 1998). Figure 3 schematically demonstrates the general concept in which the sub-basin processes the transformation from rain to flow through the SMAP model and at the exit of each sub-basin the fuzzy demands and returns are discounted. The flow propagation and the discount of concentrated demands and returns occur in the CPs. Generate flow in the sub-basin (SMAP) (SMAP) Propagation flow by the CPs Flow calculated in outfall Flow observed in outfall Computing fuzzy demands and returns Computing demands and returns concentrated CALIBRATION Figure 3- General conception of the proposed model. Source: Paiva et al. (2006 b) In this current study the simulations are made in three different time intervals: flow generation at month level through SMAP monthly version and propagation at month level; generation of mean flows at 7 days through a modification to SMAP monthly version to adequate it and 7day mean flow propagation; and the generation of daily flows using SMAP daily version and 7 day mean flow propagation. 4 Used data Hydrologic flow and precipitation data obtained from the site of the National Water Agency (Agência Nacional de Águas - ANA) have been used. Historical precipitation series were from five rainfall stations, but none of these showed a period greater than two consecutive days of failures. These failures were completed using the mean values from the other stations. The mean basing precipitation was estimated through the Thiessen polygon method. The flow series used were from the fluviometric stations represented in Table 3. Evapotranspiration data were obtained from the paper by Paiva et al. (2006 b). To apply the developed methodology it is necessary to determine the amount of water used by users of hydrologic resource. For demand it was considered only the amount of water used at rice crop because the rice crop demand is largely superior to other demand types according to Pruski et al. (2007) and Paiva et al. (2006 a). The demand was indirectly calculated through satellite images where the caption geographic location and the crop irrigation area were determined. Besides, the average consumption per hectare was also estimated. The determination of this consumption was achieved by adjustment attempts changing the demand and a final water consumption of m³/ ha was reached. Due to our unawareness of the daily water withdrawal, the characteristic demand values were considered to each time interval and then repeated each year; however, to this study the returns for fuzzy and concentrated demands were considered null. Table 3 Stations Name Code Type Using Period Operator Granja Umbu Rainfall jan/96 dec/04 ANA Ponte São Gabriel Rainfall jan/96 dec/04 ANA Caçapava do Sul Rainfall jan/96 dec/04 ANA São Sepé Montante Rainfall jan/96 dec/04 ANA Passo dos Freires Rainfall jan/96 dec/04 ANA Ponte São Gabriel Fluviometric jan/96 dec/04 ANA Passo do Rocha Fluviometric jan/96 dec/04 ANA São Sepé Montante Fluviometric jan/98 dec/04 ANA

5 5 SMAP model parameters calibration The parameter calibration for sub-basins in SMAP model was made through the SCE-UA global algorithm from a series of flows observed at the sub-basin outfall. Although there were flow stations in the intermediary sub-basins their data were not computed in the objective function. The Nash and Sutcliffe (E NS) efficiency index was the objective function used in calibrating. The determination coefficients (R²) and the error in the total drained volume ( V) were calculated to evaluate the model adjustment quality. The idea was to simulate the model using the strategy of putting together the sub-basins in groups with similar characteristics (soil type, soil use and occupation, and altimetry) so that all sub-basins in a group have the same parameter values during the calibration. Group 1 was predominantly formed by Argisols, field followed by vegetation and with developed rice crop covering 5% of the sub-basin area. In Group 2 the predominance was of Neosol, field followed by vegetation and with developed rice crop covering from 1% to 2.9% of the sub-basin area. In Group 3 the predominance was of Argisols, field followed by vegetation and with developed rice crop covering from 7% to 8.5% of the sub-basin area. In Figure 4 we present the sub-basins, the Characteristic Point location and the hydrologic stations used in the simulation. Figure 4 Sub-basins, CP location and hydrologic stations Firstly, the model was applied to determine the parameters of the sub-basins located between the Vacacaí River source and sub-basin 12. The calibration was done using the data from the fluviometric station code for the period from Jan/96 to Dec/01. When the indexes that evaluate the model were analyzed for the intermediary stations code and , we observed that it was not having an adherence of the observed and calculated hydrograph to those stations. In this way, the parameters of the Vacacaí River upstream sub-basin were calibrated and the values from the calibrated parameters were adopted in the simulations of the downstream sub-basins. In this context, the first sub-basin to be calibrated was the one that presented the station in its outfall with data from Jan/96 to Dec/04. The calibrated parameters of this sub-basin were applied in the model simulation that calibrated the parameters of the sub-basins up to the fluviometric station covering the sub-basins 1,2,3,4, and 9. To verify if the adopted technique was giving good results, a validation using the data from the station for the period from Jan/02 to Apr/04 as well as the data from the station corresponding to the period from Jan/98 to Apr/04 was performed. Finally, with the parameter already calibrated data from the station were again applied to the model simulation to ensure a good estimative to the intermediary sub-basins flow. The adopted technique allowed model validation to occur together with calibration because the subbasins calibrated parameters and the upstream were used to calibrate the sub-basins and the downstream. 6 Results and Discussion The first attempts to calibrate the model were made using the Passo das Tunas fluviometric station code , covering 10 CPs. After many adjustment attempts it was possible to see that the model presented satisfactory results only for the Passo das Tunas station. However, to the intermediary sub-basins, especially

6 the one that has the Ponte São Gabriel station ( ) in its outfall, the model response was unacceptable as demonstrated in Table 4. Table 4 Evaluation of the adherence of the calculated and observed hydrograph Using the Passo das Tunas station for calibration Calibration Jan 96 Dec 2001 Ponte São Gabriel Station Daily/ 7days % 7 days % Monthly % Passo do Rocha Station Daily/ 7days % 7 days % Monthly % Passo das Tunas Station Daily/ 7days % 7 days % Monthly % In this way, the intermediary sub-basins which have a fluviometric station in its estuary were calibrated. In this context, the sub-basin 1 was calibrated using data from the period between Jan/96 and Dec/04 from the Ponte São Gabriel station ( ). Table 5 shows the evaluation of the adjustment quality. Although the simulation in daily level represent with more details the hydrologic cycle, the simulation with monthly generation and propagation of the mean flows presented better results in the calibration of the SMAP model parameters. Table 5 Evaluation of the adherence of the calculated and observed hydrogram Ponte São Gabriel Station Calibration Jan 96 Dec 2001 Daily/ 7days % 7 days % Monthly % It was possible to see that in the monthly simulation the maximum flows were damped, what may have given better values to the indexes that evaluate the model. The model was not capable of representing well the maximum flows, especially a flow peak that occurred in December Once the monthly simulation presented the best results for the determination coefficient and for the Nash and Sutcliffe efficiency index, the permanence curve showed in Figure 5 refers to this simulation. It is possible to notice that the permanence curve of the natural flows during irrigation period is below the permanence curve of the natural flows considering the whole year, that is, the natural flows in summer are lower than in the rest of the year. Values of the calibrated parameters for sub-basin 1 were imputed to simulate the model up to sub-basin 9. In Table 6 we show the values of the indexes that evaluate the model performance when using data from Passo do Rocha station ( ) in calibrations and validation stages. As observed in basin 1 simulation, the model presented better results to daily/7 days and monthly simulations, in the calibration phase and in the validation phase as well. Again, it was verified that high flows were underestimated for daily/7 days and 7 days simulations and the simulations with generation and propagation of monthly flows and propagation of 7 days mean flows show the best results. We believe that the reason why the simulation that generates and propagates 7 days flows did not show better results is related with the suppression of the surface reservoir in the SMAP model, that is, the damping of this reservoir in intervals shorter than 7 days. Figure 6 refers to the permanence curves of observed and natural flows in the monthly simulation interval.

7 M o n th ly flo w (m ³.s -1 ) Permanence of observed flows Permanence of natural flows Permanence of observed flows - with irrigation Permanence of natural flows - with irrigation Demand Permanence Figure 51 Permanence Curve of the observed and natural flows at the Ponte São Gabriel fluviometric station ( ) monthly simulation interval period 1/1996 to 4/2004. Table 6 Evaluation of the adherence of the calculated and observed hydrograph Passo do Rocha Station Calibration Jan 96 Dec 2001 Validation Jan 2002 Apr 2004 Daily/ 7days % % 7 days % % M o n t h l y f l o w ( m ³. s - 1 ) Monthly % % Permanence of observed flow Permanence of natural flow Permanence of observed flow - with irrigation Permanence of natural flow - with irrigation Demand Permanence Figure 6 Permanence Curve of the observed and natural flows at the Passo do Rocha fluviometric station ( ) monthly simulation interval period 1/1996 to 12/2001 Results from the flow permanence curve and the estimative of withdrawal for irrigation demonstrated a large interference of the irrigation consumption in the registered flow at the fluviometric stations. This is confirmed when we observe the intersection of the demand curve with the observed and natural flow permanence curve in irrigation periods. By observing Figure 6, the intersection of the demand and the observed flow permanence curve in the period of irrigation occurred approximately in 44%, differently of what happens in the intersection of demand and natural flow permanence curve that happens in 54%, confirming the large impact of rice crop irrigation in the registered flows. The parameters values found during model calibration in the three versions, daily/7 days, 7 days and monthly were used to validate the model with data from São Sepé Montante station ( ) that is located at the outfall of sub-basin 7. Table 7 represents the model evaluation.

8 Table 7 Adherence evaluation of the calculated and observed hydrograph São Sepé Montante Station Validation Jan 98 Apr 2004 Daily/ 7days % 7 days % Monthly % The model was simulated using the Passo das Tunas station ( ), but at this time the flows generated at the intermediary sub-basins were close to reality once they have been calibrated beforehand. The results obtained for Passo das Tunas station however were not considered satisfactory for any of the simulations in accordance with Table 8, a fact that may be related to the basin dimensions. Table 8 Evaluation of the adherence of the calculated and observed hydrograph Passo das Tunas Validation Jan 96 Dec/2001 Daily/ 7days % 7 days % Monthly % We believe that the reason why the model did not present good results to this simulation is related with the basin concentration time. In the model, the inflows at the CP the most upstream of the hydrographic basin upstream reach the most downstream stretch at the same simulation time interval, that is, the water balance is accomplished disregarding the effects of storage and wave damping at river floods. Through Table 8 it was possible to notice the importance of using more than one model evaluation index because in this case the determination indexes (R²) and correlation index (R) are acceptable, differently of the Nash and Sutcliffe (ENS) coefficient and the total volume drained error ( V). 7 Final Considerations At the first attempts to calibrate the parameters of the SMAP model, the sub-basin parameters were calibrated using data from Passo das Tunas fluviometric station ( ). It was then possible to observe that the model gave a satisfactory response when observing only the data from the station itself, but the model was not capable of estimating the intermediary sub-basins flows. Therefore, the intermediary fluviometric stations demonstrated to be extremely important in evaluating the adjustment quality although they had been computed in the objective function during the model calibration. In this way, they were used to calibrate the sub-basins. When we use the parameters calibrated through the intermediary sub-basins to simulate the model up to sub-basin 12, we did not obtain good results. The reason is probably related with the basin concentration time. In the model, the inflows at the CP plus the hydrographic basin upstream reach the most downstream stretch at the same simulation time interval, that is, the water balance is accomplished disregarding the effects of storage and wave damping at rivers flooding. In a general way, it is possible to say that the methodology developed by Paiva et al (2006 b) demonstrated to be very efficient in determining the natural flows of the sub-basins with approximately 3456 km² of area and that due to factors that were not considered in the model, like propagation in rivers and storage effects, it was not possible to achieve better results for larger basins. Among the model discretization intervals that were applied up to the sub-basin 9, the ones that showed the best results were those with generation and propagation of monthly flows and, with generation of daily flows and propagation of 7 days mean flows, highlighting the monthly simulation due to the damping of the maximum flows which were not well represented by the model. We believe that the simulation that generates and propagates 7 day flows did not show better results because of the suppression of the surface reservoir in the SMAP model, that is, the damping of this reservoir does not occur in intervals shorter than 7 days. We have also to consider that the results of the obtained adjustments (simulation in sub-basins with approximately 3456 km² of area) were similar to those found in the work developed by Paiva et al (2006 b) although, in this last one, the information used were more accurate because they were obtained from the users register made by Paiva et al (2006 a). Therefore, developing and testing methodologies to estimate natural flows becomes more important each day because the alteration made by man in the environment may reflect in the river flow records. Nevertheless, it is necessary to notice that in basins with scarce hydrologic data, like in this study,

9 simplified methodologies as the one developed by Paiva et al. (2006 b) should be used, since they have been showing good results in the determination of natural flows. 8 Acknowledgments Financial assistance provided to National Counsel of Technological and Scientific Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico CNPq). 9 Bibliography AGÊNCIA NACIONAL DAS ÁGUAS (ANA). Sistemas de informações hidrológicas-hidroweb. Disponível em: < Acessado em: 20 de out.de GENOVEZ, A. M.; PIO, E. D.; Avaliação da Regionalização do Hidrograma Unitário e dos Coeficientes de Snyder para bacias rurais. In: SIMPÓSIO BRASILEIRO DE RECURSOS HÍDRICOS, 13., 1999, Belo Horizonte. Anais... Belo Horizonte, LOPES, J. E. G.; BRAGA JÚNIOR, B. P. F.; CONEJO, J. G. L. Simulação hidrológica: Aplicações de um modelo simplificado. In: SIMPÓSIO BRASILEIRO DE RECURSOS HÍDRICOS, 3., 1981, Fortaleza. Anais... Fortaleza, OLIVEIRA, C. de. P. M. et al. Sistema de Suporte a Decisão para Modelagem e Previsão de Vazões de Cheias para a Bacia do Alto Tietê Utilizando o Modelo Smap Diário. In: SIMPÓSIO DE RECURSOS HÍDRICOS DO SUL-SUDESTE, 1., 2006, Curitiba. Anais... Curitiba, PAIVA, J. B. de. D. et al. Demandas de água na bacia do rio Vacacaí Mirim. In: SIMPÓSIO DE RECURSOS HÍDRICOS DO SUL-SUDESTE, 1., 2006 a, Curitiba. Anais... Curitiba, 2006 a. PAIVA, R. C. D. ; PAIVA, E. M. C. D. ; PAIVA, J. B. D.. Estimativa das vazões naturais nas subbacias do Vacacaí Mirim através de modelo simplificado.. In: I SIMPÓSIO DE RECURSOS HÍDRICOS DO SUL-SUDESTE, b, Curitiba. Anais... Curitiba, 2006 b. PRUSKI, F.F. et al. Impacto das vazões demandadas pela irrigação e pelos abastecimentos animal e humano, na Bacia do Paracatu. Revista Brasileira de Engenharia Agrícola e Ambiental, v.11, n. 2, p , UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL (UFRGS). Programa de Pós-Graduação em Recursos Hídricos e Saneamento Ambiental. HIDP-04 Análise Sistêmica de Recursos Hídricos. Porto Alegre,1998, p