Ali Talebi

Development of a Hybrid Wavelet Packet- Group Method of Data Handling (WPGMDH) Model for Runoff Forecasting

Effective runoff prediction is one of the main aspects of successful water resources management. One of the most important problems in the modeling of such hydrological processes is the non-stationarities in the data. Several data mining models have deficiencies in handling non-stationary data particularly when signal variations are highly non-stationary. The main objective of this study was to develop a robust model to estimate daily runoff quantities. Firstly, Group Method of Data Handling (GMDH) was used in its single form to model the rainfall-runoff process. Then, the discrete wavelet and wavelet packet transforms were used to decompose the original data to their corresponding components. Thereafter, hybrid models were developed using the wavelet-based analyzed data. Three different rivers were selected to perform these modeling approaches. Results showed that GMDH model had a moderate performance (R2xa0≈xa00.84, RMSExa0≈xa02.17xa0m3/s and Max. Errorxa0≈xa024xa0m3/s for Ghale Chay River). Wavelet transform enhanced the ability of the GMDH model to some extent (R2xa0≈xa00.90, RMSExa0≈xa01.7xa0m3/s, and Max. Errorxa0≈xa016xa0m3/s for Ghale-Chay River). However, it was shown that wavelet packet transform significantly enhanced the ability of the single GMDH model with R2 of 0.94, RMSE of 1.37m3/s, and Maximum Error of about 9.8m3/s for Ghale-Chay River. The results were similar in the other two rivers. It was confirmed that the wavelet packet transform can be effectively used to deal with the non- stationarities in the data and can efficiently enhance the performance of GMDH model.

Estimation of suspended sediment load using regression trees and model trees approaches (Case study: Hyderabad drainage basin in Iran)

Abstract Estimation of suspended sediment load is one of the important topics in river engineering. Different methods are used for estimating the sediment rate. In recent years, different artificial intelligence (AI) methods, such as artificial neural network (ANN), have been used for the estimation of sediments in rivers. In this research, the suspended sediment load has been studied by using regression trees (RTs) and model trees (MTs). The study area has been located in Hyderabad watershed in west of Iran. The input data included the flow discharge, sum of three days discharge, sum of five days precipitation and the suspended sediment discharge were considered as output in the models. The numbers of total data of sediment discharge was 223 records. The obtained results were compared with ANN method (feed forward back propagation algorithm) and sediment rating curve (SRC). Results showed that RT and MT outperformed ANN method in the study area. The method of SRC had high accuracy for daily sediment discharge less than 100 ton per day in comparison with AI models, while the AI models had higher accuracy for high sediment discharge. Moreover, the combination of artificial intelligent models had high accuracy regarding to each model lonely.

Estimation of spatially enhanced soil moisture combining remote sensing and artificial intelligence approaches

ABSTRACT The main objective of this study is to combine remote-sensing and artificial intelligence (AI) approaches to estimate surface soil moisture (SM) at 100 m spatial and daily temporal resolution. The two main variables used in the Triangle method, that is, land-surface temperature (LST) and vegetation cover, were downscaled and calculated at 100 m spatial resolution. LSTs were downscaled applying the Wavelet-Artificial Intelligence Fusion Approach (WAIFA) on Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat imageries. Vegetation fractions were also estimated at 100 m spatial resolution using linear spectral un-mixing and Wavelet–AI models. Vegetation indices (VIs) were replaced with the vegetation fractions obtained from sub-pixel classification in the Ts–VI triangle space. The downscaled data were then used for calculating the evaporative fraction (EF), temperature-vegetation-dryness index (TVDI), vegetation temperature condition index (VTCI), and temperature-vegetation index (TVX) at 100 m spatial resolution. Thereafter, surface SM modelling was performed using a combination of Particle Swarm Optimization with Adaptive Neuro Fuzzy Inference System (PSO-ANFIS) and Support Vector Regression (PSO-SVR) modelling approaches. Results showed that the best input data set to estimate SM includes EF, TVDI, Ts, Fvegetation, Fsoil, temperature (T), precipitation at time t (Pt, Pt – 1, Pt – 2), and irrigation (I). It was also confirmed that PSO-SVR outperformed the PSO-ANFIS modelling approach and could estimate SM with a coefficient of determination (R2) of 0.93 and a root mean square error (RMSE) of 1.29 at 100 spatial resolution. Range of error was limited between −2.64% and 2.8%. It was also shown that the method proposed by Tang et al., (2010) improved the final SM estimations.

Identification and Analysis of Drought Propagation of Groundwater During Past and Future Periods

Analysis of the characteristics and propagation behaviors of groundwater drought at different aquifer sites during past and future periods needs a proper understanding regarding its relation with meteorological droughts. Use of a robust technique of modeling (stochastic models) allowed groundwater level and precipitation to be forecasted and then the droughts were computed and analyzed using Standardized Precipitation Index (SPI). In this research, Aleshtar Plain was selected as a case study. Analysis in this region was carried out by hierarchy and K-means clustering (5 clusters), because of the regional investigation of groundwater drought and large number of boreholes. The performance results of models showed that best forecasting models in cluster 1, 2, 3, 4 and 5 were Auto-Regressive (AR)(1), AR(2), Moving Average (MA)(2), Mixed Autoregressive–Moving Average (ARMA)(2,2) and AR(2), respectively. Furthermore, the most appropriate model for precipitation within the study plain was ARMA(1, 2). Investigation of the relationship between meteorological and groundwater drought indicated that the strongest correlation between two types of droughts was for clusters 4 and 1 with a correlation coefficient of 0.76 and 0.63, respectively. Also, the lowest correlation was for cluster 2 with a correlation coefficient of 0.51. The results of cumulative periods related to the maximum correlation between SPI and Standardized Groundwater Level Index (SGI) showed that clusters 1 to 3 corresponded with cumulative 24-month periods of SPI and this magnitude for clusters 4 and 5 were 18 and 12xa0months, respectively. Moreover, results of maximum drought severity showed there was low variability between clusters considering the extreme droughts (SGIu2009≤u2009−2) during the study period. For the future period, drought severity results showed that groundwater drought of 2019 may happen with moderate value in cluster 5, severe values in clusters 2, 1 and 4, respectively, and extreme value in cluster 3. Hydrogeological evidence of the sites and results of autocorrelation structure of SGI series confirmed the time taken by meteorological drought for propagation into groundwater. Furthermore, results showed that the aquifer is controlled more by hydraulic diffusivity factor. so it would be expected that drought propagation time into groundwater would be long for the Western part and relatively short for sites located in the East, South tending to center and partially north of the aquifer. In general, these results represent an early warning system for groundwater drought preparation and mitigation of its impacts in a future time.

A Modified LAPSUS Model to Enhance the Effective Rainfall Estimation by SCS-CN Method

In this study, LAPSUS model is modified to enhance the effective rainfall estimation by SCS curve number method. The LAPSUS model calculates discharge based on effective rainfall and routs it towards lower neighbouring grid cells following the multiple flow direction principle. Then, the sediment transport capacity and sediment transport rate are calculated in each grid cell. Finally, erosion or sedimentation is calculated by comparing the sediment transport rate with the sediment already in the transport of each grid cell. The amount of rainfall, curve number, convergence factor, discharge exponent, slope exponent, erodibility factor, and sedimentation ability factor are inputted to the application page of the modified model that was created in the C++ programming. The outputs of the model are runoff and erosion maps in ASCII format. Evaluating performance of the modified model showed a high accuracy of its results. The value of the coefficient of determination (R2) calculated 0.99 for runoff and 0.97 for erosion. The Nash-Sutcliffe efficiency was 0.96 for runoff and 0.97 for erosion. The value of the precision index calculated 0.81 for both runoff and erosion. Also, the nRMSE calculated 3% for both runoff and erosion. The result showed that the modified model capable to estimate the runoff and erosion on a landscape in a micro sub-catchment scale.