جستجوی مقالات مرتبط با کلیدواژه « پیش پردازش » در نشریات گروه « آب و خاک »

human intervention in nature and the land-use change of rivers and floodplains, and climate change, the risk of flood has increased in the world. Controlling and reducing the damages flood is done in two methods, structural and non-structural. Research conducted in different parts of the world shows that the use of non-structural methods such as flood forecasting and warning systems along with structural methods reduces flood damage. Flood forecasting and flood notification are logical tools to reduce flood risks in flood-prone areas. Precipitation forecasting is very important due to its use in flood and water resources studies. Numerical weather predictions have been extensively developed in recent years. Today, the world's leading meteorological centers have used these models in their meteorological forecasts. The ECMWF is one of the centers in global meteorological forecasts. The present study was conducted to evaluate the performance of the ECMWF center for precipitation forecasting in the Poldokhtar watershed. Numerical prediction models have systematic errors. Therefore, the purpose of this study was to investigate the effect of 7 bias correction methods (Delta, EQM, EZ, QM, LS, STB, TVSV) on the ECMWF rainfall forecasting model.

The studied area is the Poldokhtar watershed from the subbasins of the Karkheh basin. This watershed is located in Lorestan province. The information from 7 rain gauges station was used in this research. In this study, the comparison of ECMWF center forecasted rainfall with rain gauge station data was made point by point. 7 rain gauge stations in the Poldokhtar watershed were selected for point evaluation. The distance of these stations from the center of gravity of the network points was less than 10 km. The period used in this study is from 2016 to 2020. In the first step, the predicted rainfall was downloaded from the TIGGE database website. The data of this database is in GRIB2 format, which was extracted with QGIS software in Excel format. In the next step, predictions were evaluated before bias correction. Then due to the bias in the predictions, corrections were made with 7 bias correction methods. Finally, the bias correction forecast was evaluated. Bias correction methods that were used in this study include the Delta method, Elevation Zone (EZ), Quantile Mapping based on Empirical Distribution (EQM), Quantile Mapping (QM), Linear Scaling (LS), Spatio-Temporal Bias correction (STB), and the TVSV method. In this study, the daily precipitation forecast of the ECMWF center was used. In the used methods, 70% of the data were considered for the control period and 30% of the data for the prediction period. In the Delta method, the changes between the average observations and the simulation are added to the daily observed precipitation. In the LS method, the correction of daily values is based on the difference between the observed control period and the uncorrected data. The elevation zone (EZ) bias correction method, corrects the forecasted precipitation at the high altitude. The EQM method is a statistical-empirical base-quantile method, which is based on the experimental transformation and bias correction of the simulated precipitation by the regional climate model. The QM method removes biases by using cumulative distribution functions (CDF) for observed and predicted values at any time scale. In the STB method, the predicted daily precipitation is calculated for the relevant time window, and the values of the ground stations and the corrected prediction values are replaced and calculated bias. The TVSV method is based on the 7-day time window. In this study, two types of statistical indicators continuous and classified were used to evaluate the numerical model of precipitation forecasting. The continuous index includes RMSE, ME, and MAE and the classified indicators include POD, FAR, and BIAS.

The results showed that the ECMWF model had a low RMSE in most parts of the Poldakhtar watershed and had acceptable accuracy. Also, the corrected precipitation evaluation showed that the Delta method had a good performance in all the rain gauge stations under study. The STB method in two stations corrected 90% of the precipitation data. The EZ and QM methods in about 4 stations corrected about 85% of the predictions. These two methods had similar performance. LS and EQM methods had poor performance in all 7 points studied. In other words, they could not correct the bias of the predictions. The TVSV method also did not have acceptable performance. The Delta method improved the predicted precipitation in most parts of the RMSE and the STB method was able to correct the low estimate of the predicted precipitation. The results of this study showed that altitude is involved in the accuracy of predicted precipitation. If the low altitude forecast model had the lowest RMSE and at high altitudes, the highest value of this index. Biased correction methods improved the predicted precipitation values to an acceptable level, which increases the application of the predicted precipitation model in the flood warning system. According to the ME index, the underestimation is higher in the upper elevations of the basin. The main reason for this difference can be not correct the effect of altitude on the value of precipitation. At the ECMWF center, no significant change was observed in the POD index after bias correction. The small change in the POD index at the ECMWF center can be due to the good performance and structure of the numerical model at this center. The POD index at high altitudes performed better than this index at low altitudes. The bias correction methods improved the predicted precipitation values to an acceptable level, therefore increasing the effectiveness of the precipitation forecasting model in the flood warning system.

Sediment transportation and accurate estimation of its rate is a significant issue for river engineers and researchers. So far, various and complex relationships have been proposed to predict the amount of suspended sediment transport rate, such as velocity and critical shear stress based equations. However, the complex nature of sediment transport and lack of validated models make it difficult to model the suspended sediment concentration and suspended sediment discharge carried by rivers. Although the developed models led to promising results in sediment transport prediction, due to the importance of sediment transport and its impact on hydraulic structures it is necessary to use other methods with higher efficiency. On the other hand, in recent years, the Meta model approaches have been applied in investigating the hydraulic and hydrologic complex phenomena. Hybrid models involving signal decomposition have also been shown to be effective in improving the prediction accuracy of time series prediction methods, as indicated in. Complementary Ensemble Empirical Mode Decomposition analysis is one of the widely used signal decomposition methods for hydrological time series prediction. Decomposition of time series reduces the difficulty of forecasting, thereby improving forecasting accuracy.In this study, due to the complexity of the sediment and erosion phenomenon and the effect of different parameters in estimating,  time series pre-processing methods along with support vector machine (SVM) and Gaussian process regression (GPR) kernel based approaches were used to estimate suspended sediment load of a natural river at two consecutive hydrometric stations. For this purpose, different models were defined based on hydraulic and sediment particles characteristics. Moreover, the capability of integrated pre-processing and post-processing methods in two states of inter-station and between-stations was investigated. First, the Wavelet Transform (WT) method was used for data pre-processing then, the high-frequency sub-series were selected and re-decomposed using the Empirical Mode Decomposition (EMD). Finally, the most effective sub-series were imposed as inputs for kernel-based models. In addition, to assess the reliability of the superior model, Monte Carlo uncertainty analysis was used.The results showed that the GPR model had a desirable degree of uncertainty in modeling.

In this study, data of two stations of Housatonic River was used. The distance between stations was approximately 50 km. The first station is located near Great Brighton, Massachusetts, and the second station is in Connecticut. The basin area for the stations is 282 and 634 square miles, respectively. The flow path is from the first station to the second station. SVM and GPR models are based on the assumption that adjacent observations should convey information about each other. Gaussian processes are a way of specifying a prior directly over function space. This is a natural generalization of the Gaussian distribution whose mean and covariance are a vector and matrix, respectively. Due to prior knowledge about the data and functional dependencies, no validation process is required for generalization, and GP regression models are able to understand the predictive distribution corresponding to the test input. Wavelet Transform (WT) uses a flexible window function (mother wavelet) in signal processing. The flexible window function can be changed over time according to the signal shape and compactness. After using WT, the signal will decompose into two approximations (large-scale or low-frequency component) and detailed (small-scale component) components. EEMD was proposed to solve the mode mixing issue of empirical mode decomposition (EMD) which specifies the true IMF as the mean of an ensemble of trials. Each trial consists of the decomposition results of the signal plus a white noise of finite amplitude. EMD can be used to decompose any complex signal into finite intrinsic mode functions and a residue, resulting in subtasks with simpler frequency components and stronger correlations that are easier to analyze and forecast. Another important feature of empirical model of decomposition is that it can be used for noise reduction of noisy time series, which can be effective in improving the accuracy of model predictions. In the uncertainty analysis method, two elements are used to test the robustness and to analyze the models uncertainty. The first one is the percentage of the studied outputs which are in the range of 95PPU and the next one is the average distance between the upper (XU) and lower (XL) uncertainty bands. In this regard, the considered model should be run many times (1000 times in this study), and the empirical cumulative distribution probability of the models be calculated. The upper and lower bands are considered 2.5% and 97.5% probabilities of the cumulative distribution, respectively.

In order to evaluate and review the performance of the tested models and determine the accuracy of the selected models, three performance criteria named Correlation Coefficient (CC), Determination Coefficient (DC), and Root Mean Square Errors (RSME) were used. The obtained results indicated that the accuracy of the applied integrated models was higher than the single SVM and GPR models. The use of integrated methods decreased the error criteria between 20 to 25 %. The obtained results for the uncertainty analysis showed that in suspended sediment load modeling the observed and predicted values were within the 95 PPU band in most of the cases. Moreover, it was found that the amount of d-Factors for train and test datasets were smaller than the standard deviation of the observed data. Therefore, based on the results, it could be induced that the suspended sediment modeling via integrated WT-EEMD-GPR model led to an allowable degree of uncertainty.

Comparison of the developed models’ accuracy revealed that integrated GPR and SVM models had higher performance compared with single GPR and SVM models in predicting the suspended sediment discharge. The use of these two methods approximately decreased the error criteria between 20 to 25 %. According to the results, for the models that were developed based on the station data, the model with the input parameters of Dwt, Dwt-1, and Dst-1 and in the case of investigating the relationship between the stations, the model with the input parameters of Dst-2, Dwt-1, and Dst-1 were superior models. Also, based on the uncertainty analysis, the integrated GPR model had an allowable degree of uncertainty in suspended sediment modeling. However, it should be noted that the used methods are data sensitive models. Therefore, further studies using data ranges out of this study and field data should be carried out to determine the merits of the models to estimate suspended sediment load in the real conditions of flow.

Hydrological time-series is a time-dependent hydrological variable that finding the model of changes and predicting is the most important goal of time-series analysis. The purpose of this study is to simultaneously study the characteristics of time series and their prediction and the important parameters of the GEP for high-precision predictions in the training and validation. In this study, groundwater depth time-series of Chamchamal plain station located in Kermanshah province with a 12-year period and mountainous climate and the monthly time-series of Alaska temperature with a 50-year period and cold and dry climate have been used. Genexprotools5.0 software has been used to model time-series by GEP.The results of studying with GEP showed that the periodicity of data properties that existed in the time series of temperature caused correlation results above 90% in different stages of training and validation. So that the effect of different parameters of GEP is less than 10% in improving results. On the other hand, by examining the time-series of groundwater depth, which lacks periodicity and has a descending ACF shape, the prediction results of the GEP with any effective expression parameter, R more than 44% in the validation wasn't obtained. This means that the time-series preprocessing has a greater impact on the prediction results. So that by eliminating the semester, the prediction results in all stages of modeling are significantly reduced. In this case, the best R for the validation is 50%.

For sustainable land management, it is necessary to understand the functions and characteristics of soils, and their spatial and temporal changes. Visible and near-infrared spectroscopy has a specific capability to identify and determine soil properties due to high accuracy and high-performance speed. The purpose of this study is to evaluate the accuracy of visible and near-infrared spectroscopy method in estimating soil organic matter and total neutralizing value. Therefore, 110 soil samples were collected from Khuzestan, Yazd and Tehran provinces, and spectral reflectance was performed using ASD FieldSpec3. The spectra obtained from the spectrometer were pre-processed using five methods including Savitzky-Golay filter (SG), the first derivative with the Savitzky-Golay filter (FD-SG), the second derivative with the Savitzky-Golay filter (SD-SG), the standard normal variate (SNV), and Multiplicative scatter correction (MSC). Also, the performance of PLSR and SVMR methods was compared in terms of soil organic carbon and total neutralizing value estimation. The results indicated that the PLSR model in estimating both organic carbon properties and total neutralizing value had higher accuracy compared to the SVR model. In estimation of soil organic carbon, PLSR method and MSC preprocessing method had the best performance (R2VAL=0.59, RMSEVAL=0.19 and PRDVAL=1.47) and the second derivative method had the least performance (R2VAL=0.15, RMSEVAL=0.27 and PRDVAL=0.52). Also for estimation of total neutralizing value, the first derivative preprocessing method had the best performance (R2VAL=0.78, RMSEVAL=5.70 and PRDVAL=2.01) and the second derivative method had the least performance (R2VAL=0.1, RMSEVAL=11.13, and PRDVAL=0.31). The key wavelengths were observed for soil organic matter in the range of 421- 612 nm and for total neutralizing value in the range of 2315- 2151 nm. This study showed that the Vis-NIR spectroscopy method, due to its physical basis and considering the influencing factors, as a large-scale model, makes it possible to evaluate and predict soil OC and TNV.

Threshold friction speed is an important factor for assessing the soil erodibility, but its measurement is time consuming and costly. Estimating threshold friction velocity by use of soil reflectance increases operating speed and reduces cost. The aim of this study was to compare the efficiency and accuracy of partial least squares regression (PLSR), support vector regression (SVR) and artificial neural network (ANN) regression models in estimating the threshold friction velocity in dust-prone areas of Khuzestan Province. For this purpose, 91 soil samples were collected from the central and southern parts of dust-prone areas of the province and the threshold friction velocity was determined using wind tunnels. Then, the reflectance spectra of soil samples were obtained with a spectrometer. Pre-processing methods were performed on the main spectrum and modeling was performed using, PLSR, SVR and ANN models. The results showed that the threshold friction velocity in the region was 9.7 m/s and the minimum was 5.25 m/s.  Also, the threshold friction velocity was significantly (p<0.05) correlated with dissolved sodium (r= -0.58) and sodium adsorption ratio (R= -0.48). The ANN model had the best estimation accuracy in the second derivative preprocessing (PRD = 2.52) and the SVR model had the lowest estimation accuracy in the main spectrum (PRD = 0.56). Finally, the key wavelength of the threshold friction velocity was in the range of 1850 and 1930 nm. Because of the soil reflectance correlation with threshold friction velocity (r=0.76), the spectroscopy method can be used to assess the soil erodibility in areas prone to dust production.

Forecasting of river discharge is a important aspect of efficient water resources planning and management. In this study, time series pre and post-processing methods along with support vector machine (SVM) and Gaussian process regression (GPR) kernel based approaches were used to estimate flow discharge of two natural river in the United States with two consecutive hydrometric stations. The first river contained about 2 years of data and in the second river 4 years of daily discharge data was used. Different models were defined based on hydraulic characteristics and the capability of integrated pre and post-processing methods in two states of inter-station and between-stations was investigated. For data pre-processing, the Discrete Wavelet Transform (DWT) method was first used. Then, the high-frequency sub-series were selected and re-decomposed using the Ensemble Empirical Mode Decomposition (EEMD). Finally, sub-series with higher energy were imposed as inputs for kernel-based models. Non-linear neural average (NNA) model was also used for data post-processing. The obtained results from the defined models showed the high accuracy of the integrated methods used in the research in estimating flow discharge. At both stations, the error percentage was reduced by approximately 20 to 25% using the integrated pre-post-processing methods compared to the intelligent kernel based models. It was observed that in the case of river flow prediction based on the station's own data, the RMSE error value of the model decreased from approximately 0.3 to 0.26 and in the case of using the previous station data decreased from 0.44 to 0.33. Due to the high capability and accuracy of the pre-processing methods used in this study, similar studies are recommended in other rivers of the country.

Forecasting stream flow is very important for planning and managing water resources systems and its precision has a significant role in accurately implementing the planned objectives. Besides, soft computing has a significant ability in modelling hydrologic processes. Therefore, in the present study, the Hybrid Wavelet-Gene Expression Programming model has been developed in comparison with its singular approach so that it forecasts the daily streamflow of Khoshkroud river located in Guilan province. For this purpose, in addition to the process of pre-processing hydrometric data, the effect of meteorological parameters on the model’s performance and efficiency has been studied. Also, pre-processing was performed with different properties and for four durations of one, two, three and six days. Correlation coefficient (R), index of agreement (Ig), Nash-Sutcliffe coefficient (NSE), mean absolute error (MAE), root-mean-square error (RMSE) and peak flow criteria(PFC) statistical indices were used to assess the models’ performances. The results show that using wavelet transform to pre-process hydrometric data will significantly improve the efficiency of the hybrid model in comparison with the singular model, such that the correlation coefficient of the validation data for three days has increased from 0.27 to 0.80 and similarly, the mean absolute error has decreased from 1.4 to 0.80 m3/Sec. On the other hand, meteorological parameters have caused the extreme values in the river’s flow rate time series to be well modelled and their efficiency in the extreme values to be significantly increased. The results obtained from this research express that the hybrid model alongside the meteorological parameters can be successfully and efficiently used in flow forecasting.

In recent years, due to the lack of surface coating and low soil resistance to wind erosion, the large area of Khuzestan province is sensitive to dust production. Among the soil characteristics, the organic matter by collecting soil particles, has important role in soil resistance to wind erosion and dust production. Since these areas are so wide, the use of traditional methods of soil analysis is really costly and time consuming. The spectroscopy approach, due to the advantage of speed and easy movement, can reduce the cost and time of measurement. The aim of this study is to investigate the spectral behavior of soil organic carbon in central and southern regions of Khuzestan province by using tow multivariate regression, Support Vector Regression) SVR( and neural network (PLS-ANN), and key wavelength determination of soil organic matter in these areas.

In this research, the study area was divided into 2 km square grids and systematic and random sampling Methods were performed. The soil organic matter in samples was measured in the laboratory. The Reflectance spectra of soil samples were determined using FildSpect setup in dark room. And spectral measurements were carried out with three types of detectors in range of visible to near infrared (3500-2500 nm). To eliminate the noise in normal reflectance spectra, the main spectra were preprocessed by four methods, including the first derivative with the Savitzky-Golay filter (FD-SG), the second derivative with the Savitzky-Golay filter (SD-SG), the standard normal variant method (SNV) and the continuum removed method (CR). Next, the performance of SVR and PLS-ANN models in main spectra preprocessed method were compared.

The results showed that the PLS-ANN model had better accuracy compared to SVR model in estimating organic carbon. In SVR models, the continuum removal method (CR) had the best performance (R2CAL=0.84, RMSECAL=0.06 and RPDCAL=1.82) and the main Spectra had the worst performance (R2CAL=0.74, RMSECAL=0.14 and RPDCAL=1.66). In PLS-ANN models, the best performance belonged to the second derivative (SD-SG), (R2CAL=0.92, RMSECAL=0.05 and RPDCAL=2.34) and the worst performance was related to the first derivative (FD-SG), (R2CAL=0.80, RMSECAL=0.1 and RPDCAL=1.86).

In this study, the preprocessing methods improved the overall accuracy of SVR and PLS-ANN models compared to the main spectrum. According to the second derivative method, in PLS-ANN witch had the best accuracy in estimating soil organic carbon, the Wavelength ranges around 800, 1800 and 2000 nm were identified as the key wavelength of the organic carbon in sensitive centers to dust production.

A broad area of saline and semi-saline lands of Khuzestan province have changed into centers susceptible to dust production due to eroded wind and lack of surface coating and low soil resistance. The objective of this study was to model the soil salinity of sensitive areas to dust production in Khuzestan Provenience usin spectrometry method of visible and near-infrared wavelengths (2500-350 nm). The least square multivariate regression model, artificial neural network and random forest model were used to estimate soil salinity. The main soil spectrum was determined using the FieldSpect machine. Also, preprocessing methods including Savitzky-Golay filter, the first derivative with the Savitzky-Golay filter (FD-SG), the second derivative with the Savitzky-Golay filter (SD-SG), the standard normalization method (SNV), and the continuum remove method (CR) were used to eliminate the noise and to increase the accuracy of the multivariate model. The results showed that the combined model partial least squares-artificial neural network model with assessment criteria (RPDcal = 3.40-2.65) has high accuracy for salinity estimation. In contrast, the combined model of least squares - random forest showed the lowest accuracy (RPDcal = 0.85-1.98). Preprocess of the main spectrum in two models (neural network and partial least squares regression) increased the relative accuracy of the model; while in the random forest model, preprocess reduced the accuracy of the model compared to the main spectrum. The ranges of 1800, 1900, 2000, 2300 and 1500 nm were recognized as "the key wavelengths" impressed by soil salinity. The key wavelengths can be used in remote sensing studies and mapping of soil salinity in areas sensitive to dust production in Khuzestan province.

Although the soil salinity as an effective factor on soil and water management is typically assessed by measuring the soil electrical conductivity (ECe), this conventional laboratory method is time-consuming and costly. Therefore, near-infrared spectroscopy (NIR) as a fast, cheap and non-destructive method to assess soil salinity level can be considered as a valuable alternative method. Reviews of literature on the application of NIR spectroscopy for soil salinity prediction have shown that there is no sufficient information about the effect of soil texture on results accuracy; therefore, in this study the soil salinity was predicted under different soil salinity levels and various soil textures. The effect of different pre-processing methods was also investigated to improve the predicted soil salinity. Twenty three surface soil samples were collected from different places in Fars province, then; some soil properties such as percentage of particles size and ECe were measured. These samples were artificially salted by adding the water in different salinity levels to the soil samples. The ECe of these soils were between 2.1 to 307.5 dS/m and then all samples dried to reach the field capacity level. Soil reflectance spectra were obtained in 350-2500 nm wavelength range. The absorbance and derivative of reflectance spectra were calculated based on the reflectance spectra. In order to determine the effect of smoothing technique, as a pre-processing method, 4 various methods (moving average, Gaussian, median and Savitzky-Golay filters) in 12 different segment sizes (3,5,7,9,11,13,15,17,19,21,23 and 25) were applied and the processed spectra introduced to Partial Least Square Regression (PLSR) model to predict soil salinity in two calibration and validation steps. At the first step, the soil salinity was predicted for all samples using of reflectance, absorbance and derivative of reflectance spectra under 4 pre-processing methods and 12 segment sizes. According to the R2 and RMSE indices, the best type of spectra, the effect of various pre-processing methods and the best segment size in prediction of soil salinity were determined as absorbance spectra, moving average and Savitzky-Golay filters for segment size of 25 and 15, respectively. In the second step, the effect of soil texture on prediction accuracy was investigated. For this purpose, soil samples were divided into the coarse and fine textures and soil salinity was predicted for each of these groups using different pre-processing methods and different segment sizes. In prediction of soil salinity by absorbance, reflectance and derivative of reflectance spectra, the R2 values in validation step were 0.742, 0.706 and 0.670; and RMSE values were 29.92, 31.96 and 33.9 (dS.m-1), respectively. The absorbance spectra were the best spectra type in prediction of soil salinity. Therefore, in next step, absorbance spectra were used only for predicting the salinity in fine and coarse soil textures. Results showed that the prediction in coarse texture was better than that of the fine texture (R2= 0.836 and R2=0.756, respectively). It was also revealed that the highest R2 occurred in coarse texture and the accuracy of prediction was reduced in fine textures. The results showed that the performance of different pre-processing methods is related to the spectrum type. Although the pre-processing methods had no positive effect in using of reflectance spectra, but it improved the predicted values which were obtained using of absorbance and derivative of reflectance spectra. The best results were occurred when the absorbance spectra were used. Moving average method increased the accuracy of prediction more than the other pre-processing methods, and according to the results this method, for the segment size of 25, was the best technique in soil salinity prediction. According to the R2 and RMSE indices, the prediction of soil salinity by absorbance spectra was more accurate than the prediction using reflectance and derivative of reflectance spectra (R2= 0.742, 0.706 and 0.670, respectively). Although the predicted soil salinity in coarse soils were more accurate than that in fine soils. Using of absorbance spectra to predict the soil salinity in all soil textures was efficient. The results showed that using of pre-processing methods improved the soil salinity prediction by absorbance and derivative of reflectance spectra, and the moving average and Savitzky-Golay filter were the best pre-processing methods.

Soil organic matter is considered as an indicator of soil quality and its fertility due to its impact on the physical, chemical and biological properties of soil. The purpose of this study was to investigate the effect of moisture on soil spectral behavior in order to estimate the amount of organic matter using visible-near infrared spectroscopy. Also, the performance of different preprocessing methods was compared in modeling. For this purpose, 50 soil samples from a surface layer (0-30 cm) were randomly collected from some cities in Tehran province. The samples were air-dried and passed through a 2 mm sieve. The amount of organic matter and soil texture were measured in the laboratory. To measure spectral reflectance, soil samples are dry in the oven and they were hydrated with levels of 5, 10, 15 and 20 wt.% (Grams of water / g of soil). Then, Soil samples were scanned using a FieldSpec Pro Spectroradiometer with a measurement range of 350–2500 nm. The results of the cross-evaluation indicated that SNV pre-processing method was better in predicting organic carbon content of soil. The best result for the validation group was obtained in dry oven moisture with R2 and RMSE values of 0.83 and 0.422 respectively. Also, R2 and RMSE were 0.75 and 0.543, respectively, at 5% moisture content, 0.70, 0.553 at 10% moisture, 0.605, 0.558 in moisture content of 15%.

Optimal use of irrigation water requires to the precise irrigation planning and the accurate crop coefficient estimation is the prerequisite of that particularly in the global scale. The aim of research was the comparison of preprocessing approaches of artificial neural network: regression and data reduction (principle component analysis and rotation) for crop coefficient estimation using NDVI, RI, TVI, MSAVI, SAVI, mTVI, EVI, MNDVI and TVX for wheat crop coefficient of East Azarbaijan Province. The performance of regression and data reduction approaches indicated the error criteria decreasing of data reduction approach, for example RMSE increasing from rotation to regression and from principle component analysis to regression was 11.8 and 22.7%, respectively. The used approaches of crop coefficient estimation has overestimation as the average increase of crop coefficient in the validation period showed 7.7, 6.13 and 4.6% increasing of crop coefficient from FAO to the regression, principle component and rotation approaches, respectively. Therefore, using the rotation in the data reduction analysis increased the accuracy of estimation. Decreasing of correlation coefficient-39.13% - from MSAVI to NDVI indicated that the improved indices basis on the study area condition increased the performance of crop coefficient estimation using satellite images.

Southeast watersheds of Iran including Great Karoon, Karkheh, Jarrahi and Zohreh have the most significant contribution in the water supply of the agriculture, industry, drinking water and hydroelectric power plants over Iran. 25 percent of the country’s electricity is produced from hydroelectric power plants located in this region. The existence of a monthly relatively high resolution gridded precipitation dataset is of the most important needs of water resources management for such as deciding on the suitable time of dewatering and discharge of dams, calibration of dynamical monthly forecasting models and drought early warning. Even considering all observation stations governed by Meteorological Administration and Ministry of Power, the density of stations is not so enough to use them for calibration of hydro-climate model outputs. To overcome this deficiency, one way to fill the gap is using bias corrected global gridded precipitation dataset such as APHRODITE, CMORPH, PRESIANN and other newly generated data.
Material and Watershed of Karkheh, great Karoon, Jarrahi and Zohreh are the area of study which covers southwest provinces of Khuzestan, Kermanshah, Ilam, Chaharmohal-Bakhtiari, Kohkiluyeh and Buyerahmad, Isfahan, Hamadan, Fars and Lorestan, which is shown in figure 2. There are 135 observation station in the area of study which governs by Iran Meteorological Organization and Ministry of Power. Area of study covers by 75 grids of 0.5×0.5 degree latitude and longitude. For each grid there is an APHRODITE precipitation data. In the 34% of grids, there is no observation station. The main goal of this study is to attribute a reliable monthly precipitation data to all grids without any observation station. Period of APHRODITE data set is 1987-2007, which is same to observation period. Firstly regional bias of APHRODITE data set has been computed by comparing observed precipitation with APHRODITE one. Then bias corrected APHRODITE precipitation (Composite APHRODITE Observation dataset) has been placed in non-observation grids. Efficiency of composite precipitation data has been determined by statistical parameters of bias, correlation and Nash-Sutcliff indices.
In this research the results have been evaluated at monthly and seasonal time scales. In the case of seasonal time scale, we found that the minimum APHRODITE’s bias of 1.2 mm has been occurring in summer, while the maximum bias has been occurring in winter by 40.9mm. It means that the bias is high in the rainy season. Seasonal correlations were statistically acceptable in 0.05 significant levels, showing same seasonal fluctuations in APHRODITE and rain gage data. To provide seasonal composite APHRODITE-Observed precipitation gridded data set, mean seasonal bias of APHRODITE has been removed, while preserving seasonal fluctuation. The highest spatial correlation of 0.8 was detected in autumn, while it was about 0.7 for spring and winter. The minimum seasonal correlation was in summer by 0.5. There were also a good agreement between area averaged observation and APHRODITE data, when considering statistical indices of bias, Nash-Sutcliff and relative percentage errors. Results show the cumulative distribution function of APRODITE data is behind of the observed cumulative distribution function data, meaning that APHRODITE reaches its maximum earlier than observation data. This implies that APHRODITE cannot capture well the extreme monthly precipitation. Monthly correlations are approximately greater than 0.9, but the only exception is September with a correlation coefficient of 0.52. All correlations are significant in 0.05 levels. The highest spatial correlation was occurred in Novembers. Monthly Nash-Sutcliff was 0.96 in monthly time series. The categorical percentage score was 94.1%. These results strongly confirm that APHRODITE precipitation data is a good option for replacement in grid cells without observations. The number of observation stations per cell is varied from 1 to 7. We found that the maximum monthly correlations occur in grid cells of 0.5×0.5 degree latitude and longitude which having at least 3 observation stations. The three-station bias has been applied to APHRODITE data, then bias-removed data has been replaced with grid cells without observations. Spatial patterns of new composite APHRODITE-observation data set has good agreement with observation in the areas having intense observation stations. They also can capture well the spatial precipitation distribution of rainy areas located in the center of basin and low rainfall areas located in the southwest of the region. The results of this research can be used in calibration of dynamical seasonal forecasting outputs, drought early warning and rain-runoff simulation.