مجله تحقیقات منابع آب ایران
سال نهم شماره 1 (پیاپی 26، بهار و تابستان 1392)

Hydrogeological parameters like transmissivity are among the important and money-consuming input parameters of ground water modeling. Fuzzy logic, Artificial Neural Networks, and Neurofuzzy has high capability in hydrogeological parameter estimation.In this research combination of these models applied to transmissivity estimation of Tasuj aquifer. Tasuj plain aquifer is one of the marginal plains of Lake Urmia which suffered more ground water declination in last decades and needs qualitative and quantitative management. To overcome the complexity of hydrogeologic systems, Hybrid Artificial Intelligence Modelis then presented as a committee machine. Based on the outputs, weights of each models optimized using particle swarm optimization which caused to form committee machine. Related geophysical and hydrogeological variables to transmissivity such as transverse resistivity (Rt), electric conductivity (EC), alluvium thickness (B), and geographic location were among inputs to this study. The obtained results from presented committee machine showed higher efficiency compared to the Fuzzy logic, Artificial Neural Network, and Neurofuzzy models, individually.

In this study, several precise models and methods are used for estimating annual and monthly evaporation rates for the reservoir of Saveh (Alghadir) Dam for 13 years (1995-2007). Water budget and evaporation pan, experimental methods, penman type equations, Bowen Ratio Energy Budget (BREB) method, and Complementary Relationship Lake Evaporation model (CRLE) were applied to this study. The maximum and minimum values for the mean annual evaporation rates are estimated between 145 cm (for Debruin-Keijman method) and 175 cm (CRLE model for shallow lakes). Also the long term monthly evaporation pattern from the applied methods are different from the standard method of BREB evaporation. The least difference in the mean monthly evaporation from the BREB results, are for the Mass Transfer, Papadakis, Penman, Brutsaert-Stricker, and Priestley-Taylor methods. However, only the standard deviation of Penman and Mass Transfer methods were very small and the other three methods have relatively large standard deviation from the BREB results. The qualification of the CRLE model have been rejected, and the Thornthwaite, Hamon and Ryan-Harleman methods have also large bias in mean and standard deviation for monthly evaporation rates from the BREB results.

Evaluation of infiltration and runoff in karstic areas is considerably important. These processes in karstic areas are more different from the other land coverages. Due to variation of factors affecting the infiltration and runoff in karstic areas, a specific formula cannot be set for these processes. In this study the amount of runoff and infiltration in a karstic catchment, named Delibajak in central Iran, were evaluated in both basin and plot scales. Precipitation and runoff components were continuously measured over the basin during research period and the runoff coefficient was estimated based on the collected data. Also, in plot scale, these components were evaluated on soil and rock coverages in 25 tests using sprinkler. Evaluated infiltration and runoff data in the basin-scale, show a very high infiltration capacity in Delibajak catchment, with very small runoff coefficients (1.5, 0.54 0.45 for events of 133, 130 and 95 millimeters of rain, respectively). Results of sprinkler data indicate a wide range of infiltration rate in Delibajak basin from almost zero to 40 percent.

This study aimed the analyses of the simultaneous effects of two phenomena MJO and ENSO on the occurrence of dry and wet spells in Iran. The study showed that the concurrence of El Niño with the MJO negative phase (El-N) and La Niña with the positive phase of the MJO (La-P) have the highest frequencies. Precipitation amounts and the incidence of dry and wet spells were, therefore, compared for these two distinct alternatives. The results indicated that, in February, November, and December, a nationwide dry condition is expected when La Niña is coincided with the MJO positive phase. On the other hand, the pervasive wet spell is anticipated if the El Niño events have concurred with the MJO negative phase. Although rainfall amount in dry zones of the southern Iran is less than corresponding values in the northern half of the country, the simultaneous impact of the considered Oscillations on the occurrence of dry or wet episodes are more significant for the southern half. For instance, Iranshahr station, in southeastern part of the country, has experienced 96 percent reduction or 90 percent increase in precipitation compared to the longterm mean values as the (La-P) or (El-N) was prevailed. Furthermore, for some stations located in very dry zone of the country including (Birjand, Torbatheidarieh, Tabas, and Yazd) the occurrence of dry or wet periods in April was significantly associated to the (La-P) or (El-N) episodes. For these stations about 105 percent increase or 51 percent decrease compared to the longterm mean precipitation values is anticipated as (El-N) or (La-P) events was prevailed, respectively.

In assessing the potential impacts of climate change on different sections including water, agriculture, and urban drainage management, projection of changes in climate extremes as the results of climate change and global warming are essential. To have an outlook on future projections of climate extremes particularly precipitation, the outputs derived from three coupled general circulation models (HadCM3, NCCCSM, and CGCM3T47) contributing to the Fourth Assessment Report of the IPCCAR4, have been downscaled for Mashhad station under A1B emission scenarios by LARS- WG during three period 2011-2030, 2046-2065 and 2080-2099. The extremes are described by seven indices based on precipitation including CDD, R10mm, R20mm, RX5day, SDII, R95T, and R99T. The results showed that heavy precipitation events for pentads increase the maximum five-day precipitation and their intensity in three sequential periods. In addition, a larger fraction of the total annual precipitation is projected to occur during heavy precipitation events, i.e. events that exceed the 95th and 99th percentile. Increases are found for these indices indicated the more frequent future occurrence of floods in the twenty first century.

For water resources management in dry areas that rely on dams and surface water storage, the use of Drought Early Warning System (DEWS) with the hydrological indicator that capability to deal with the drought and water shortage are very useful and it also prevents the reducing of water reserves.In this research, it is tried to develop a drought early warning system, relying on effective component in reservoir management. The developed drought early warning system consists of five essential elements, namely, (1) Drought monitoring, (2) Prediction and uncertainty analysis of water consumption in the future, (3) Calculation of an index for drought alert (4) Risk and uncertainty analysis and (5) Policy Making for Drought Management that used in Zayandeh-Rud dam. To design this system, at the first stage the inflow was predicted by using Artificial Neural Networks (ANNS) in a period of 6-months with considering the relevant uncertainty and difference of probability levels. Also drought conditions were categorized in five levels by using of historical data (1983-2005) of reservoir water storage and using Self Organizing Feature Map (SOFM). The levels arenone, slightly severe, fairly severe, severe and very severe. Then a drought alert index was calculated with current drought monitoring conditions of reservoir and water consumption measuring in a 6-month forecast period. Based on the results of calculated index, warning of different levels of green status (normal condition) to red status (severe condition) with relevant uncertainty and different confidence levels was determined. In the next step, a nonlinear optimization model was used to determine optimum reduction of demands for maximum reservoir incomes. Finally, the performance of this system and its role in reducing the severity of the drought has been studied in the period of 1998-2001 as a severe drought in the study region.Results showed that the developed DEWS can alert droughts with overall accuracy of about 75%. Furthermore, it can determine the optimal water release in different management scenarios. So this system can be an effective tool for water resources management in areas that rely on dams.

In order to simulate time series, various methods are presented such as times series models (AR, ARMA and ARMAX), multi-linear regression (MLR), and nonparametric regression (K-NN). In this research, performance of these models for estimation of missing values and prediction of future values of evaporation series (from open water) were assessed. ARMAX model with standardized input time series of Tmin, Tmax, Tav, Wind, RH, and sunshine hours, outperformed the other models and the K-NN and MLR were in the next ranks, respectively. Also after the principal component analysis, ARMAX model showed noticeable deviation for estimating missing values and MLR and K-NN in calibration and MLR in validation stage performed the best. For short-term predictions, ARMAX model has the best performance, but MLR performed better in long-term predictions, Time series models were not robust for long term predictions.

In cases that the gauging station in the downstream is destroyed for some reasons, and it is necessary to know the stream flow in the downstream, it is possible to forecast stream flow in the downstream station using the available data in the upstream station. In this research, the peak discharge of Gelinak station has been forecasted at outlet of the Taleghan watershed using artificial neural network in two states. In the first state, historic data of the Gelinak station including the maximum daily mean discharges, corresponding rainfall, one day antecedent rainfall and five days antecedent rainfall, sum of the five days antecedent rainfall and monthly mean temperature. In the second state, these data for the hydrologic units of Gatehdeh, Mehran, Alizan, Joestan were extracted and the physiographic parameters area, average height, main waterway length, and the average river slope were added into the artificial Neural Network model. The model is feed forward with two layers and the back-propagation algorithm. Data were trained, validated, and tested in three stages. Results showed that the forecast of peak discharge using the upstream station and the physiographic parameters are better than the peak discharge forecast using data from the last year in the downstream station.