Mohammad Nazeri Tahrudi; Farshad Ahmadi; Keivan Khalili
Abstract
Introduction: Given the fact that Iran is located in the center of the dryland of earth and is significantly influenced by the deserts of Central Asia and hot dry deserts of Arabia and Africa, is one of the most arid and low rainfall land areas.So is the proper management of water resources is of critical ...
Read More
Introduction: Given the fact that Iran is located in the center of the dryland of earth and is significantly influenced by the deserts of Central Asia and hot dry deserts of Arabia and Africa, is one of the most arid and low rainfall land areas.So is the proper management of water resources is of critical importance. The first step in the proper management of water resources is studying the factors that affected these resources including climate change. In fact climate change is a dynamic process in terms of time and place. Large parts of the Earth's climate as part of their normal variability in short-term and long-term experience. Short-term climate changes due to the difference in terms of average annual values of specific climate variables in average periods such as 30 years. Causes and effects of regional climate change in several parts of the world have been widely studied from various aspects. Among hydrological parameters, precipitation is the most important parameter in the complex hydrologic cycle. Follow the phenomenon of global warming on the Earth's surface, the rainfall pattern has changed.Trends of rainfall in different parts of the world have been studied by many researchers. Due to climate change in Iran and climate change in the Basin of Urmia Lake it seems that evaluation the trend of monthly and annual precipitation and its time of change point in the basin of Urmia Lake changes is important. The goal of this study is evaluatingthe trend and time of the change point trend of monthly and annual precipitation of rain gage stations in Urmia Lake basin.
Material and methods: Lake Urmia is the focus of surplus accumulation of surface currents all the rivers of the basin, with an area of approximately 5750 square kilometers and the average elevation of 1276 m above sea level and is located in the middle of the northern basin. Around of Lake Urmia there are 16 wetlands with an area of 5 to 120 hectares (some have dried up) that mostly have sweet or salty and fresh water and a high value of ecosystems.Urmia Lake Basin is situated in eastern of 44-14 to 47-53 and north of 40-35 to 30-38 coordinates. Urmia Lake Basin rainfall changes is 220 to 900 mm and have mean precipitation about 263 mm that added in central parts of the basin to the highlands.
Trend analysis: The aim of process test is to specify whether an ascending or a descending trend exists in data series. Since parametric tests have some assumptions including normality, stability, and independence of variables, where most of these assumptions do not apply to hydrologic variables, the nonparametric methods are more preferred in meteorological and hydrological studies. The nonparametric methods are less sensitive to extreme values compared to parametric tests in the examination of trends. Nonparametric tests can also be utilized for data time series regardless of linearity or nonlinearity of the trend (Khalili et al. 2014). One of the most well-known nonparametric tests is Mann-Kendall test (Mann 1945; Kendall 1975).
The modified Mann-Kendall test (MMK): The main assumption of Mann-Kendall test is that the sample data has no significant autocorrelation. However, some hydrological series might have a significant autocorrelation coefficient. When a series has a positive autocorrelation coefficient, there is an increased chance for Mann-Kendall test to reveal the existence of a trend in this series. In this case, the null hypothesis i.e. lack of trend is rejected, yet this hypothesis should not actually be. The modified Mann-Kendall test was presented by Hamed and Rao (1998) and has been used by Kumar et al (2009) for the analysis of the trend of Indian rivers. In this method, the effect of all significant autocorrelation coefficients is removed from the time series and is appliesto a series whose autocorrelation coefficients are significant in one or more cases.
Change point test: Pettittest is a non-parametrictest that was developedin 1979byPettit. Themethod is used in order tofind change points ina time series(Salarijazi et al 2012).In this study,thestatisticwas usedtofind asudden change intemperaturedata.Thisstatistic isatest with rank basis and without a distributionin orderto detectsignificantchangesin the mean of the time seriesanditis importantwhenthereis noassumptionabout the change time.
Results and discussion: In this study the trend of monthly and annual precipitation of rain gage stations that located in Urmia Lake basin were investigated using modified Mann-Kendall test. Z values of case study were calculated in two monthly and annual scales. The results of evaluation the trend of precipitation of rain gage stations of Urmia Lake basin showed that in October, December, January, February and March (five months of the year) the trend of precipitation is decreasing and the mean of Z values showed the less than zero values. In April and May there is no sensible changing in precipitation trend. Also the results showed that the March, April and May have a low failure rate and February, December and July have a most of change point of monthly precipitation data. About 60 percentages of the time of change point in precipitation trend are between 1992 and 1998. Also the results showed that two months of May and November there is no changing point in west Urmia Lake rain gage stations. In annual scale the time of changing trend is between 1992 and 1998.
Conclusion: The results of evaluation the trend of Lake Urmia precipitations showed that the Urmia Lake basin has a combination of decreasing and increasing trend in studied time period. The decreasing trend in precipitation often seen in west stations of the basin and west and south-west of Urmia Lake. The increasing trend also seen in south and north-east of Urmia Lake basin. Also the results of zoning the Z values of Mann-Kendall test showed that in annual scale the regions that influenced by polar-continental air mass that they entered Iran have a decreasing trend.
Keyvan Khalili; Mohammad Nazeri Tahrudi; Rasoul Mirabbasi Najaf Abadi; Farshad Ahmadi
Abstract
Introduction: Climate change in the current era is a very important environmental challenge. Our understanding of the impacts of human activities on the environment, especially those related to global warming caused by increased greenhouse gases indicates that, most probably, a number of hydro-climatic ...
Read More
Introduction: Climate change in the current era is a very important environmental challenge. Our understanding of the impacts of human activities on the environment, especially those related to global warming caused by increased greenhouse gases indicates that, most probably, a number of hydro-climatic parameters are changing. Based on the scientific reports, the average temperature of the earth has increased about 0.6 degrees centigrade over the 20th century and it is expected that the amount of evaporation continues to rise. In this case, the atmosphere would be able to transport larger amounts of water vapor, influencing the amount of atmospheric precipitations (21). Low precipitation and its severe fluctuations in the daily, seasonal and annual time scales are the intrinsic characteristics of Iran’s climates. Based on the research background, it seems that no comprehensive study has been conducted on concentration of winter precipitation in Iran. The aim of this study is to calculate the concentration and Trend of precipitation of Iranian border stations over the last half-century.
Materials and Methods: Iran with an area of over16480000 square kilometers is situated in the northern hemisphere and southwest of Asia. Almost all parts of Iran have four seasons. In general, a year can be divided into two warm and cold seasons. In this study, 18 stations were selected among more than 200 synoptic stations existing in the country, for investigating the concentration and precipitation trend.
PCI Index The PCI index has been proposed as an index of precipitation concentration. The seasonal scales of this index are calculated as equation 1(18):
(1)
Where Pi is the amount of monthly precipitation in the ith month. Based on the proposed formula, the minimum value of theoretical PCI is 8.3, indicating absolute uniformity in the precipitation concentration (i.e. the same amount of precipitation occurs every month).
Trend analysis The aim of process test is to specify whether an ascending or a descending trend exists in data series. Since parametric tests have some assumptions including normality, stability, and independence of variables, where most of these assumptions do not apply to hydrologic variables, the nonparametric methods are more preferred in meteorological and hydrological studies.
Results and Discussion: The PCI index was calculated using the monthly precipitation of the selected stations at seasonal and winter time scales over a 50-year period. This period was then divided into two 25-year sub-periods for the investigation of changes in average values of PCI (7). In the first 25-year span, the irregular precipitation distribution has been observed in the Bandarabbas station and its surroundings in winter season. In none of the studied stations, highly irregular precipitation occurred. The highest share of PCI was relatedto the precipitation average distribution class, and the northern, northwestern, and northeastern parts of the country have a uniform precipitation distribution. In winter, within the first 25-year period, the country had ideal conditions in terms of precipitation and its concentration in the mentioned regions. Within the second 25-year period, the intensity of irregular precipitation concentration decreased, as the regions that had confronted strong precipitation irregularities wereadded to regions with uniform concentration. At the seasonal scale and in winter, the country’s share of uniform distribution diminished in the second 25 years, and overall most parts of Iran have been covered by average precipitation distribution. The uniform precipitation distribution in recent years (second 25 years) has decreased in winter in such a way that no uniform distribution has been observed in the northeast of the country and uniform distribution belongedto the Caspian sea border strip, southern regions of west and east Azerbaijan stations (Urmia, Khoy and Tabriz stations) along with Kermanshah, Sanandaj, and Zanjan stations.
Trend analysis of the PCI In winter the Abadan, Ahwaz, Bandarabbas, Birjand, Kermanshah, Sanandaj, Urmia and Zahedan stations experienced an insignificant decreasing trend in PCI. At other stations, an insignificant increasing trend was observed in the PCI series. Overall, 9 out of 18 considered stations, witnessed increasing PCI trend implying increased irregularities in winter precipitation.
The results of Mann-Kendall trend test for precipitation Based on the results it can be observed that in winter Ahwaz, Gorgan, Khoramabad, Kermanshah, Ramsar, Rasht and Sanandaj experienced an insignificant decreasing trend in precipitation. In Khoy, Sanandaj, Tabriz, Urmia, Zahedan, and Zanjan stations, the decreasing precipitation trend in winter was significant. Overall, 12 out of 18 studied stations have been afflicted with a decreasing precipitation trend in winter.
Conclusion: Precipitation Concentration Index (PCI) is an index for determining the precipitation variations in a certain region and PCI analysis can reveal the accessibility to water in an environment. In this study, the PCI was used to analyze the precipitation concentration at two annual and seasonal time scales throughout the Iran (from 1961 to 2010). The PCI zoning results at the seasonal scale demonstrated that precipitation concentration had the same trend within the two 25-year sub-periods. These results also revealed a high PCI in provinces with low precipitation such as Zahedan. These stations, according to Oliver (18) classification, have irregular and sporadic precipitation duringwinter. Overall, the PCI analysis at the seasonal scale indicated that the regions covered by polar-continental, Europe-originated polar-continental and North Atlantic ocean-originated polar-continental have the best precipitation concentration throughout the country. The results of this index provided valuable information for water resources managers in regions with low-precipitation, consistent with research by Gozzini et al (7). The results of modified Mann-Kendall (MMK) test for PCI in Iran revealed a decreasing trend over the last 50 years. Based on the obtained results in winter, the Khoy, Sanandaj, Tabriz, Urmia, Zahedan, and Zanjan stations experienced a significant decreasing trend. The existence of an increasing trend in PCI albeit insignificant reveals changes in Iran's winter precipitations confirmed by Mann-Kendall test for precipitations in 18 studied stations. Overall, it can be concluded that the decreasing trend in Iran's winter precipitation has resulted in increasing PCI and thereby increased irregularities in winter precipitations, especially in winter season.
Mohammad Nazeri Tahrudi; Keivan Khalili; Javad Behmanesh; Kamran Zeinalzadeh
Abstract
Introduction: Drought from the hydrological viewpoint is a continuation of the meteorological drought that cause of the lack of surface water such as rivers, lakes, reservoirs and groundwater resources. This analysis, which is generally on the surface streams, reservoirs, lakes and groundwater, takes ...
Read More
Introduction: Drought from the hydrological viewpoint is a continuation of the meteorological drought that cause of the lack of surface water such as rivers, lakes, reservoirs and groundwater resources. This analysis, which is generally on the surface streams, reservoirs, lakes and groundwater, takes place as hydrological drought considered and studied. So the data on the quantity of flow of the rivers in this study is of fundamental importance. This data are included, level, flow, river flow is no term (5). Overall the hydrological drought studies are focused on annual discharges, maximum annual discharge or minimum discharge period. The most importance of this analysis is periodically during the course of the analysis remains a certain threshold and subthresholdrunoff volume fraction has created. In situations where water for irrigation or water of a river without any reservoir, is not adequate, the minimum flow analysis, the most important factor to be considered (4). The aim of this study is evaluatingthe statistical distributions of drought volume rivers data from the Urmia Lake’s rivers and its return period.
Materials and Methods: Urmia Lake is a biggest and saltiest continued lake in Iran. The Lake Urmia basin is one of the most important basins in Iran region which is located in the North West of Iran. With an extent of 52700 square kilometers and an area equivalent to 3.21% of the total area of the country, This basin is located between the circuit of 35 degrees 40 minutes to 38 degrees 29 minutes north latitude and the meridian of 44 degrees 13 minutes to 47 degrees 53 minutes east longitude. In this study used the daily discharge data (m3s-1) of Urmia Lake Rivers.
Extraction of river drought volume The drought durations were extracted from the daily discharge of 13 studied stations. The first mean year was calculated for each 365 days using the Eq 1 (14).
(1) (For i=1,2,3,…,365)
That Ki is aith mean year, Yijis ith day discharge in jth year and n is number of period years. After the extraction the 1 to ndays drought duration, the years with no data were complete with Regression or interpolation methods. After the extraction, data initial evaluation (Trend, Independence and Stationarity) and completed the drought volume data, these data were fitted by the common distribution functions and select the best function based on Kolmogorov-Sminnov test. To read more information about the data initial evaluations see the NazeriTahroudi et al (15).
Log Pearson type 3 distribution Log Pearson type 3 distribution and its parameters is (7 & 12):
(2)
After selectingthe best distribution function based on Kolmogorov - Smirnov test, estimated the selected function parameter to evaluate the return period. For this purpose, there are many methods such as moments, Sundry Average method (SAM), Logarithm of applied moments observations and maximum likelihood that in this study were compared.
Results and Discussion: In this study, using daily flow data fromstations studied; the drought volume of days 1 to 60 was extracted, corrected, and completed. Before fitting the extraction drought volume river data with distribution function, the mentioned data were investigated with Wald-Wolfowitz (Independence and Stationary), Kendall (Trend) and Wilcoxon (Homogeneity) tests and the results of these tests were accepted in two significant levels of 1 and 5 percentages. Before estimatingthe Log Pearson type III parameters, first the drought volume river data were modeled by the Easy Fit software and common distribution functions and Log Pearson type III was selected by the Kolmogorov – Smirnov test as the best function. Results of two Anderson Darling and Chi Squared tests foraccurate evaluation were added. After initial evaluation of data and statistic tests, the time series of drought Volume River data of the studyarea were fitted by log Pearson type III. To estimatethe Log Pearson type III parameters used the sundry average method and to investigatethe accuracy of this method, 3 methods (moment, maximum likelihood and Logarithm of applied moment observations) were used and 4 mentioned methods for all of rivers were calculated. The most river drought relating to Gadar-Chai river with 1742 million cubic meters low volume and the lowest of it relating to Mardoq-Chai river with 68 million cubic meters low volume in 10000 year return period. After Gadar-Chai river the most low volume of discharge relating the Zarineh-rood river. Two Zarineh-rood and Gadar-Chai rivers among other rivers have a higher average discharge. Log Normal III, Gamma, Wikeby and GEV distributions have a good fitting on river flows data and no difference in investigation models that corresponded with Mosaedi et al (13) and NazeriTahroudi et al (15). The results of Grifits (7) also introduced the Wikeby distribution has a better than Beta distribution. Lee (12) also with evaluation the rainfall frequency in the study the rainfall concentration properties in Chia-Nan (Taiwan) introduced the Log Pearson type III as the best distribution function between the common distribution function. Results of Chi-Squared test in methods of parameter estimation showed that all methods are acceptable.
Conclusion: Drought occurrence can be estimated bythe analysis of historical data for different regions and using the results of predicting problems can be reduced. In this research daily river flow of Lake Urmia basin applied to calculate drought volume of rivers. Log Pearson III distribution selected among current hydrological distribution functions for fitting drought volume of rivers. Using selected distribution function and Sundry Average Moment method for estimating parameters return period of drought from 2 to 10000 years extracted. Results showed that volume of drought for Shahar-chai , Barandoz-chai, Nazlu-Chai, Mahabad-Chai, Rozeh-Chai, Gadar-chai, Simineh-rood, Zola-chai, Aji-chai, Sofi-chai, Leilan-chai and Mardoq-chay rivers in the return period of 10000 years will be 92, 125, 228, 150, 110, 1742, 90, 77, 690, 280, 65, 68 Mm3 respectively.
Mohammad Nazeri Tahrudi; Keivan Khalili; Farshad Ahmadi
Abstract
Introduction: Climate change has been one the most important subject in studies in the recent decades. Precipitation is an effective climatic parameter in the municipal and rural studies and in the industry, architecture, agriculture, climate and other fields. Trend analysis of average monthly and yearly ...
Read More
Introduction: Climate change has been one the most important subject in studies in the recent decades. Precipitation is an effective climatic parameter in the municipal and rural studies and in the industry, architecture, agriculture, climate and other fields. Trend analysis of average monthly and yearly rainfall investigated in many studies, but less researches probe regional rainfall analysis. In this study average yearly precipitation data measured at 31 synoptic stations of Iran in the period of 1961 to 2010 used to study regional variations of precipitation. In this order station divided to five regions by fuzzy clustering. Then, using the regional Kendall method, trend of precipitation investigated at five regions and all of Iran.
Materials and Methods: Iran with an area of over 16480000 square kilometers is situated in the northern hemisphere and southwest of Asia. Almost all parts of Iran have four seasons. In general, a year can be divided into two warm and cold seasons. Iran with range annual precipitation of 62.1-344.8 mm is located between two meridians of eastern 44° and 64° and two orbits of northern 40° and 25°. In order to investigate trend of precipitation two Mann-Kendall and Regional Kendall tests used. Also to evaluate the regional trends the Fuzzy method applied to clustering the studied region. The classic form of Mann-Kendall test has been used in many studies. The null hypothesis (no trends) is accepted when , otherwise H0 is rejected and its opposite hypothesis, i.e. the existence of a trend is accepted (5, 13). To estimate regional trend, the mean S statistic of Regional Mann-Kendall introduced that was presented by Douglas et al (7). Fuzzy Clustering: Clustering the studied area was done using the Fuzzy clustering method. One of the first clustering methods that were based on the objective function and Euclidean distance was presented by Dunn in 1974 and then was generalized by Bezdak in 1981.The FCM clustering algorithm is modified type of K-Means clustering algorithm. This algorithm minimizes the variance of clusters (1). The assumption of this algorithm is that data are in a vector space and the objective of this algorithm is to minimize the sum of variance in the D v cluster.
Results and Discussion: In this section the results of decreasing and increasing trend of annual precipitation of Iran can be observed in order to the data that recorded at provinces synoptic stations in the 1 and 5 percentage significance levels. Isfahan Synoptic station detected an increasing trend insignificant level of 5 percentages and the East Azerbaijan synoptic station followed a significant and severe decreasing trends. In order to investigate regional trend it is needed to use the clustering methods. After investigation the trend of mean annual precipitation at each station, the studied area was clustered using the Fuzzy clustering method and then the regional trend of Iran’s precipitation was evaluated. At first the number of different clusters investigated using the geographic properties and mean annual precipitation of the studied area and then with attention to the correlation of precipitation series in each cluster, five clusters selected to investigate the regional trend of precipitation. Overall the results showed that about 67 percentages of synoptic stations in center of provinces detected decreasing trend in the recent half century. Increasing the precipitation almost accrued in the center and northern part of Iran and other areas detected a decreasing precipitation trend in the studied data period that this subject is corresponded with Azerakhshi and et al (2). The observed trends over Iran and almost all stations and provinces were downward trend. This decreasing trend of precipitation also observed in Iran in the two past decades by Khalili et al (13).
Conclusion: Result showed decreasing trend in the west, north of Iran at each station and regional scale. Results indicated also a significant downward trend at northwest, central and south-west of the country, non-significant downward trend in western of Iran and non-significant upward trends in northern regions and Caspian Sea margins in the regional analysis. The most decreasing trend of precipitation observed at the north west of Iran because of increasing temperature and climate changes in the recent years.
Majid Montaseri; Babak Amirataee; Keyvan Khalili
Abstract
Introduction: Droughts are natural extreme phenomena, which frequently occur around the world. This phenomenon can occur in any region, but its effects will be more severe in arid and semi-arid regions. Several studies have highlighted the increasing of droughts trend around the world. The majority of ...
Read More
Introduction: Droughts are natural extreme phenomena, which frequently occur around the world. This phenomenon can occur in any region, but its effects will be more severe in arid and semi-arid regions. Several studies have highlighted the increasing of droughts trend around the world. The majority of studies in assessing the trend of time series are based on basic Mann-Kendall or Spearman's methods and no serious attention has been paid to the impact of autocorrelation coefficient on time series. However, limited numbers of studies have included the lag-1 autocorrelation coefficient and its impacts on the time series trend. The aim of this study was to investigate the trend of dry and wet periods in northwest of Iran using Mann-Kendall trend test with removing all significant autocorrelations coefficients based on SPI and RAI drought indices.
Materials and Methods: Study area has a region of 334,000 square kilometers, with wet, arid and semiarid climate, located in the northwest of Iran. The rainfall data were collected from 39 synoptic stations with average rainfall of 146 mm as the minimum of Gom station, and the highest annual rainfall of 1687 mm, in the Bandaranzali station. In this study, Standardized Precipitation Index (SPI) and Rainfall Anomaly Index (RAI) were used for trend analysis of dry and wet periods. SPI was developed by McKee et al. in 1993 to determine and monitor droughts. This index is able to determine the wet and dry situations for a specific time scale for each location using rainfall data. RAI index was developed by Van Rooy in 1965 to calculate the deviation of rainfall from the normal amount of rainfall and it evaluates monthly or annual rainfall on a linear scale resulting from a data series. Then, correlation coefficients of time series of these drought indices with different lags were determined for check the dependence or independence of the SPI and RAI values. Finally, based on dependence or independence of the time series values, trend analysis of wet and dry periods was conducted in different stations using one of the basic or modified Mann-Kendall tests. Also, the magnitude of the trends was derived from the Theil- Sen’s slope estimator.
Results and Discussion: Time series of SPI and RAI drought indices for a given annual rainfall as an example for three stations of Marivan, Gom and Maku show that during 1991 to 1994 and from 2002 to 2007 are in wet period and during 1987 to 1990 and 1998 to 2001 are in the dry period. It is clearly show that, dry and wet periods in RAI index are more severe than SPI. Comparison the correlation between Lag-1 autocorrelation coefficients values of SPI and RAI time series and Lag-1 autocorrelation coefficients of annual rainfall data indicate that these correlations are high and about 0.97 and 0.99, respectively. This difference is due to the different classification of SPI and RAI drought indices. The results of trend analysis indicate a decreasing trend in most of stations. Also, Mann-Kendall statistic has been declining while eliminating the effect of all significant correlation coefficients of dry and wet periods. This result in both SPI and RAI indices are similar and have a high correlation with R = 0.99. According to results, west of the study area have a significant decreasing (negative) trend. The spatial distribution of dry and wet periods showed that the difference between Mann-Kendall statistics of SPI and RAI indices is minimal. Also, The results show that, the slope of the trend line based on the SPI and RAI drought indices is negative in most of stations and correlation between these two indices in determining the slope of the trend line is high. But, this correlation compared with the trend statistics of SPI and RAI time series is less.
Conclusions: In this study, first the time series of SPI and RAI time series based on annual precipitation and common quantitative classification of mentioned two drought indices were determined. Then, trends of dry and wet periods of selected stations in northwest of Iran were evaluated based on these indices using the Mann-Kendall trend test with removing all significant autocorrelation coefficients. The results from this study indicate that using Mann-Kendall test with removing all significant autocorrelation coefficients effects are essential in assessing trend in time series. Although, according to various studies available in the literature, SPI is known as more accurate than RAI in drought mitigation, but according the results of this study, can solely be used both RAI and SPI index for trend detection.
F. Ahmadi; S. Ayashm; K. Khalili; J. Behmanesh
Abstract
Introduction Crop evapotranspiration modeling process mainly performs with empirical methods, aerodynamic and energy balance. In these methods, the evapotranspiration is calculated based on the average values of meteorological parameters at different time steps. The linear models didn’t have a good ...
Read More
Introduction Crop evapotranspiration modeling process mainly performs with empirical methods, aerodynamic and energy balance. In these methods, the evapotranspiration is calculated based on the average values of meteorological parameters at different time steps. The linear models didn’t have a good performance in this field due to high variability of evapotranspiration and the researchers have turned to the use of nonlinear and intelligent models. For accurate estimation of this hydrologic variable, it should be spending much time and money to measure many data (19).
Materials and Methods Recently the new hybrid methods have been developed by combining some of methods such as artificial neural networks, fuzzy logic and evolutionary computation, that called Soft Computing and Intelligent Systems. These soft techniques are used in various fields of engineering.
A fuzzy neurosis is a hybrid system that incorporates the decision ability of fuzzy logic with the computational ability of neural network, which provides a high capability for modeling and estimating. Basically, the Fuzzy part is used to classify the input data set and determines the degree of membership (that each number can be laying between 0 and 1) and decisions for the next activity made based on a set of rules and move to the next stage. Adaptive Neuro-Fuzzy Inference Systems (ANFIS) includes some parts of a typical fuzzy expert system which the calculations at each step is performed by the hidden layer neurons and the learning ability of the neural network has been created to increase the system information (9).
SVM is a one of supervised learning methods which used for classification and regression affairs. This method was developed by Vapink (15) based on statistical learning theory. The SVM is a method for binary classification in an arbitrary characteristic space, so it is suitable for prediction problems (12).
The SVM is originally a two-class Classifier that separates the classes by a linear boundary. In this method, the nearest samples to the decision boundary called support vectors. These vectors define the equation of the decision boundary. The classic intelligent simulation algorithms such as artificial neural network usually minimize the absolute error or sum of square errors of the training data, but the SVM models, used the structural error minimization principle (5).
Results Discussion Based on the results of performance evaluations, and RMSE and R criteria, both of the SVM and ANFIS models had a high accuracy in predicting the reference evapotranspiration of North West of Iran. From the results of Tables 6 and 8, it can be concluded that both of the models had similar performance and they can present high accuracy in modeling with different inputs. As the ANFIS model for achieving the maximum accuracy used the maximum, minimum and average temperature, sunshine (M8) and wind speed. But the SVM model in Urmia and Sanandaj stations with M8 pattern and in other stations with M9 pattern achieves the maximum performance. In all of the stations (apart from Sanandaj station) the SVM model had a high accuracy and less error than the ANFIS model but, this difference is not remarkable and the SVM model used more input parameters (than the ANFIS model) for predicting the evapotranspiration.
Conclusion In this research, in order to predict monthly reference evapotranspiration two ANFIS and SVM models employed using collected data at the six synoptic stations in the period of 38 years (1973-2010) located in the north-west of Iran. At first monthly evapotranspiration of a reference crop estimated by FAO-Penman- Monteith method for selected stations as the output of SVM and ANFIS models. Then a regression equation between effective meteorological parameters on evapotranspiration fitted and different input patterns for model determined. Results showed Relative humidity as the less effective parameter deleted from an input of the model. Also in this paper to investigate the effect of memory on predict of evapotranspiration, one, two, three and four months lag used as the input of model. Results showed both models estimated monthly evapotranspiration with the high accuracy but SVM model was better than ANFIS model. Also using the memory of evapotranspiration time series as the input of model instead of meteorological parameters showed less accuracy.
M. Moravej; K. Khalili; J. Behmanesh
Abstract
Introduction: Studying the hydrological cycle, especially in large scales such as water catchments, is difficult and complicated despite the fact that the numbers of hydrological components are limited. This complexity rises from complex interactions between hydrological components and environment. Recognition, ...
Read More
Introduction: Studying the hydrological cycle, especially in large scales such as water catchments, is difficult and complicated despite the fact that the numbers of hydrological components are limited. This complexity rises from complex interactions between hydrological components and environment. Recognition, determination and modeling of all interactive processes are needed to address this issue, but it's not feasible for dealing with practical engineering problems. So, it is more convenient to consider hydrological components as stochastic phenomenon, and use stochastic models for modeling them. Stochastic simulation of time series models related to water resources, particularly hydrologic time series, have been widely used in recent decades in order to solve issues pertaining planning and management of water resource systems. In this study time series models fitted to the precipitation, evaporation and stream flow series separately and the relationships between stream flow and precipitation processes are investigated. In fact, the three mentioned processes should be modeled in parallel to each other in order to acquire a comprehensive vision of hydrological conditions in the region. Moreover, the relationship between the hydrologic processes has been mostly studied with respect to their trends. It is desirable to investigate the relationship between trends of hydrological processes and climate change, while the relationship of the models has not been taken into consideration. The main objective of this study is to investigate the relationship between hydrological processes and their effects on each other and the selected models.
Material and Method: In the current study, the four sub-basins of Lake Urmia Basin namely Zolachay (A), Nazloochay (B), Shahrchay (C) and Barandoozchay (D) were considered. Precipitation, evaporation and stream flow time series were modeled by linear time series. Fundamental assumptions of time series analysis namely normalization and stationarity were considered. Skewness test applied to evaluate normalization of evaporation, precipitation and stream flow time series and logarithmic transformation function executed for in order to improve normalization. Stationarity of studied time series were evaluated by well-known powerful ADF and KPSS stationarity tests. Time series model's order was determined using modified AICC test and the portmanteau goodness of fit test was used to evaluate the adequacy of developed linear time series models. Man-Kendall trend analysis was also conducted for the precipitation amount, the number of rainy days, the maximum precipitation with 24 hours duration, the evaporation and stream flow in monthly and annual time scales.
Results and Discussion: Inferring to the physical base of ARMA models provided by Salas et al (1998), the precipitation has been considered independently and stochastically. If this assumption is not true in a given basin, it is expected that the MA component of stream flow discharge model be eliminated or washed out. This case occurred in basins A, B and C. In these basins, the behavior of precipitation and evaporation was autoregressive. It was observed that the stream flow discharge behavior also follows autoregressive models that had greater lags than precipitation and evaporation lags. This result proved that the precipitation, evaporation, and stream flow processes in the basin were regular processes. In basin D, the behavior of precipitation was stochastic and followed the MA model, which was related to the stochastic processes. In this basin, the stochastic behavior of precipitation affected the stream flow behavior, and it was observed that the stochastic term of MA also appeared in the stream flow. Thus, this leads to decrease the memory of stream flow discharge. The fact that the MA component in the stream flow discharge was greater than the MA component in precipitation indicated that during the process of producing stream flow discharge from precipitation, the stochastic factors performed an important role.
Conclusion: A comprehensive investigation on hydrological time series models of precipitation, evaporation and stream flow were investigated in this study. The framework of the study consists of trend analysis using Mann-Kendall test and time series. Trend analysis results indicate the significant changes of water resources in the studied area. It means that sustainable development in studied area is greatly threatened. The results of parallel modeling of precipitation, evaporation and stream flow time series showed that the behavior of stream flow models are greatly affected by precipitation models. In other words, this study evaluate the physical concept of ARMA models in real-world monthly time scale for three main hydrologic cycle components and suggest that considering parallel hydrological time series modeling could increase the accuracy to select a model for simulation and prediction of stream flow time series. In addition, it suggested that there is a relation between climate pattern and hydrological time series models.
Keywords: ARMA models, Stationarity, Trend analysis, Water cycle components
Mohammad Nazeri Tahrudi
Abstract
The application of statistical theory and probability analysis of hydrologic time series is assumed that the variables are normally distributed. Since many time series are not normal, it is required prior to any analysis and modeling, they looked normal. This conversion is done by Const. In this study, ...
Read More
The application of statistical theory and probability analysis of hydrologic time series is assumed that the variables are normally distributed. Since many time series are not normal, it is required prior to any analysis and modeling, they looked normal. This conversion is done by Const. In this study, using 12 common function to convert the normalized data, the average monthly rainfall in different regions of Iran into the data were normally distributed and the skewness coefficient, superior functions in each climate zone was Iran. The results showed that the data used in hot and dry regions with a square, as well as normal and the rest of the climate zones are likely to become a tropical area, Johnson and temperate regions selected for the inverse transform
H. Naveh; K. Khalili; M. T. Alami; J. Behmanesh
Abstract
One of the important tools in modeling and forecasting of hydrological processes, is using and analysis of time series. The generated river flow series by using time series models have been used in many researches such as drought, flood periods, reservoir systems design and other purposes. The use of ...
Read More
One of the important tools in modeling and forecasting of hydrological processes, is using and analysis of time series. The generated river flow series by using time series models have been used in many researches such as drought, flood periods, reservoir systems design and other purposes. The use of nonlinear time series is very useful in river flow forecasting because of nonlinear river flow behavior in different spatial and time scales. The purpose of this study is to investigate the efficiency of bilinear nonlinear time series model in river flow forecasting. In this research monthly flow of Shahar-Chai and Barandouz-Chai rivers located in West Azarbaijan for duration of 31 and 39 years respectively were used. Despite of simplicity of bilinear nonlinear model, the results showed that this model had high efficiency in modeling and forecasting of two rivers and presented best results from ARMA model. The error of fitted model of Barandouz-Chai (1.605) was less than the model fitted for Shahar-Chai river (1.920). The reason may be due to longer data period for Barandouz-Chai river or it’s recharge from springs and ground waters.