Beruflich Dokumente
Kultur Dokumente
ISBN 978604821338-1
ABSTRACT
In this study, the Canoabo reservoir storage volume variation is analyzed during the period 1993-2009. The analysis of
the time series is performed by representing the time series of volume and test application: autocorrelation, randomness
and prediction models. It was found that since the operation of the same; useful and total volumes have not been
overcome, so it never relieved design channels. It is evident that the storage volumes have fallen below the dead
volume. The ARIMA model (Autoregressive Integrated Moving Average) (2,2,1) has been found as the one that fits the
observations, however, the forecast model statisticals show significant differences between estimates and observations in
the validation stage.
1. INTRODUCTION
The Canoabo reservoir supplies water for human and
industrial consumption. The population benefited
includes: Ciudad de Morn (67,372 people), Municipality
of Puerto Cabello (181.905 people) and Moron
petrochemical complex as well as power plant "Planta
(1)
Centro". The Canoabo reservoir operation has been
The standard error for rk is calculated with the
influenced by the occurrence of environmental and
assumption that the autocorrelations have "disappeared"
structural of the dam problems. The quantity and quality
by the lag k; being equal to 0 at all lags greater than or
of water entering the Canoabo reservoir has decreased
equal to k. The standard error is calculated as follows:
due to the following anthropic actions: deforestation,
pollution and inappropriate land uses (livestock on
slopes), which have eroded and degraded large areas,
diversion of flow rates planting of crops, slash and burn,
poor management of solid waste and wastewater from
homes (Sevilla et al, 2009; Malpica,2012). (2)
In this study, the variation of the storage volume is
analyzed on Canoabo reservoir operation during the This standard error is used to calculate 100 (1-)%
period 1993-2009. The analysis of time series is made by probability limits around zero, using a critical value of
applying the following techniques: representation of the standard normal distribution:
volumes over time and its relationship to operating
activities; and an autocorrelation analysis of the volumes
in the time to detect whether there is a pattern in the data (3)
of operation and these can be fitted to a mathematical
model for prediction. ARIMA models
1
Table 1 Properties of Canoabo reservoir
4. RESULTS
2
runs is equal to 26, compared with an expected value of
3052,5 if the sequence was randomized. Since the P -
value for this test is less than 0,05, it can be rejected the
hypothesis that the series is random, with a confidence
level of 95,0%. The second test counts the number of
times the sequence ascended or descended. The number
of such runs is equal to 484, compared to an expected
value of 4069,67 when the sequence was randomized.
Since the P - value for this test is less than 0,05, it can be
rejected the hypothesis that the series is random, with a
confidence level of 95,0%. The third test is based on the
sum of squares of the first 24 autocorrelation coefficients.
Figure 2 Storage Volume vs Sediment Volume (SV) of Canoabo
Reservoir, Venezuela Since the P - value for this test is less than 0,05, it can be
rejected the hypothesis that the series is random, with a
confidence level of 95,0%.
In Figure 4 is shown the autocorrelation values estimated Since the three tests are sensitive to different types of
between storage volumes for different delays (days). The deviations from random behavior, failure to comply with
delayed autocorrelation coefficient k measures the these, suggests that the time series may not be completely
correlation between the values of volume at time t and at random.
time t-k. Probability limits of 95,0% are also shown
around 0. If the probability limits for a particular delay Forecast
do not contain the estimated coefficient; there is a
statistically significant correlation to that delay in the In an extended analysis of various models, it was selected
confidence level of 95,0%. In this case, 120 of the 180 the model of an integrated ARIMA (2, 2, 1)
autocorrelation coefficients are statistically significant at a autoregressive moving average. This model assumes that
confidence level of 95,0%, implying that the time series the best forecast available for future data is given by the
can not be completely random (white noise). parametric model relating the most recent value with the
previous values.
The statistical significance of the terms in the forecasting
model is summarized in Table 3. Terms with P-values
less than 0,05 are statistically different from zero with a
confidence level of 95,0%. P-values for terms AR (1), AR
(2) and MA (1) are lower than 0,05, so that are statistically
different from 0.
Test for Randomness The performance of the ARIMA (2, 2, 1) model selected to
adjust historical data is summarized in Table 4. Showing:
In Table 2, the results of the application of three tests are root mean square error (RMSE), mean absolute error
shown for determining whether the storage volume is a (MAE) and mean error (ME). Each is based on statistical
random sequence of numbers or not. A time series of forecast errors a-below, which are the differences
random numbers is often called white noise because it between the data at time t and the predicted value at time
contains an equal contribution to various frequencies. t -1. The first two statistics measure the magnitude of the
The first test counts the number of times the sequence errors. A better model would give a smaller value. The
was above or below the median. The number of such last statistic measures the bias. A better model would
3
give a value closer to 0. In this case, the model has been pattern in the operation thereof; due to the occurrence of
calculated based on the first 6100 data 5 data at the end of abrupt increases and decreases in storage volumes.
the time series have been retained to validate the model .
The forecasting model ARIMA (2,2,1) was successfully
In Table 3, the statistical error for both periods validation approached in the calibration step; however showed
estimation is shown. If the results are significantly higher significant differences in the stages of validation and
in the validation period, implies that the model does not prediction.
fit as well as expected otherwise, to predict the future.
7. ACKNOWLEDGEMENTS
Tabla 4 ARIMA Model Summary
The research was conducted at the Center for Water
Research and Environmental (CIHAM-UC) and
Estimation Validation
Statistical Period Period Laboratory of Water and Soil Quality of HidroLabToro
RMSE 319224 171928 Consults, with financial support from the Ministry of
MAE 87873,4 92425,1 Popular Power for Science and Technology (FONACIT)
ME 18,83 76684,5 and Ministry of Popular Power for the Environment.
In general it is observed in Figure 5, a significant CORPOCENTRO (2010). Dosier Estado Carabobo. Corpocentro.
variability in the values of the storage volumes of Disponible en
Canoabo reservoir due to the irregularity of the same [http://corpocentro.vicepresidencia.gob.ve/dossier/Dossier
%202010/DOSIER%20CARABOBO%202010.pdf]. Fecha de Consulta:
operation. Thus, for a period between 01/01/2001 and 22/06/2013. Valencia.
01/10/2001, a substantial reduction of storage volumes HIDROCENTRO, (1993) Balance Hdrico del Embalse Canoabo.
can be seen to below the dead volume. As also, a rapid Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
increase in storage volumes; depending of the HIDROCENTRO, (1994) Balance Hdrico del Embalse Canoabo.
implementation of a pumping system to transfer from Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (1995) Balance Hdrico del Embalse Canoabo.
Temerla River to Canoabo Reservoir, which is evidenced Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
by the high values of volumes in the 2005 period. From HIDROCENTRO, (1996) Balance Hdrico del Embalse Canoabo.
the above, it can be shown that Canoabo Reservoir has Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
been operated under conditions in which there is not a HIDROCENTRO, (1997) Balance Hdrico del Embalse Canoabo.
pattern; making it difficult to fit a model for the operation Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (1998) Balance Hdrico del Embalse Canoabo.
of this. Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
Regarding the ARIMA modeling method, the difference HIDROCENTRO, (1999) Balance Hdrico del Embalse Canoabo.
between the observed data and the predicted data show Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
significant residues as a result of physical quantities HIDROCENTRO, (2000) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
manipulated; finding ranges between 19.716,9 and
HIDROCENTRO, (2001) Balance Hdrico del Embalse Canoabo.
18.534,6 m3. Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2002) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2003) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2004) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2005) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2006) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2007) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
HIDROCENTRO, (2008) Balance Hdrico del Embalse Canoabo.
Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia
Figure 5 Storage Volume vs Sediment Volume of Canoabo HIDROCENTRO, (2009) Balance Hdrico del Embalse Canoabo.
Reservoir Gerencia de Captacin, Tratamiento y Mantenimiento. Valencia.
Malpica, F. (2012). Uso sustentable de las cuencas alta y media del ro
Canoabo. Tesis de Maestra. Universidad Yacambu. Vicerrectorado de
Investigacin y postgrado. Repblica Bolivariana de Venezuela.
6. CONCLUSIONS http://es.scribd.com/doc/102692317/Disenos-de-investigacion-
cuantitativa-Informe-de-tesis-de-grado
From the time series for modeling of Canoabo Sevilla, V., Comerma J.A., and Silva O. (2009) caracterizacin de la
Reservoir operation, confirmed that from the beginning cuenca del ro Canoabo en el Estado Carabobo, Venezuela. I. Anlisis
of its operation have not been reached the useful and climtico y de produccin de agua. Agronoma Trop. 59(1): 33-44. 2009