Tiantian Yang

Developing reservoir monthly inflow forecasts using artificial intelligence and climate phenomenon information

Author(s): Yang, T; Asanjan, AA; Welles, E; Gao, X; Sorooshian, S; Liu, X | Abstract:

Simulating California reservoir operation using the classification and regression‐tree algorithm combined with a shuffled cross‐validation scheme

Author(s): Yang, T; Gao, X; Sorooshian, S; Li, X | Abstract:

An enhanced artificial neural network with a shuffled complex evolutionary global optimization with principal component analysis

Abstract The classical Back-Propagation (BP) scheme with gradient-based optimization in training Artificial Neural Networks (ANNs) suffers from many drawbacks, such as the premature convergence, and the tendency of being trapped in local optimums. Therefore, as an alternative for the BP and gradient-based optimization schemes, various Evolutionary Algorithms (EAs), i.e., Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Simulated Annealing (SA), and Differential Evolution (DE), have gained popularity in the field of ANN weight training. This study applied a new efficient and effective Shuffled Complex Evolutionary Global Optimization Algorithm with Principal Component Analysis – University of California Irvine (SP-UCI) to the weight training process of a three-layer feed-forward ANN. A large-scale numerical comparison is conducted among the SP-UCI-, PSO-, GA-, SA-, and DE-based ANNs on 17 benchmark, complex, and real-world datasets. Results show that SP-UCI-based ANN outperforms other EA-based ANNs in the context of convergence and generalization. Results suggest that the SP-UCI algorithm possesses good potential in support of the weight training of ANN in real-word problems. In addition, the suitability of different kinds of EAs on training ANN is discussed. The large-scale comparison experiments conducted in this paper are fundamental references for selecting proper ANN weight training algorithms in practice.

Assessment of the Influences of Different Potential Evapotranspiration Inputs on the Performance of Monthly Hydrological Models under Different Climatic Conditions

AbstractPotential evapotranspiration (PET), which determines the upper limit of actual evapotranspiration (AET), is a necessary input in monthly hydrological models. In this study, the sensitivities of monthly hydrological models to different PET inputs are investigated in 37 catchments under different climatic conditions. Four types of PET estimation methods (i.e., Penman–Monteith, Hargreaves–Samani, Jensen–Haise, and Hamon) give significantly different PET values in the 37 catchments. However, similar runoff simulations are produced based on different PET inputs in both nonhumid and humid regions. It is found that parameter calibration of the hydrological model can eliminate the influences of different PET inputs on runoff simulations in both nonhumid and humid regions. However, the influences of parameter calibration on the simulated water balance components, including AET and water storage change (WSC), are different in nonhumid and humid regions. In nonhumid regions, simulated runoff, AET, and WSC ar...

Evaluation of streamflow simulation results of land surface models in GLDAS on the Tibetan plateau

The Global Land Data Assimilation System (GLDAS) project estimates long-term runoff based on land surface models (LSMs) and provides a potential way to solve the issue of non-existent streamflow data in gauge-sparse regions such as the Tibetan Plateau (TP). However, the reliability of GLDAS runoff data must be validated before being practically applied. In this study, the streamflows simulated by four LSMs (CLM, Noah, VIC, and Mosaic) in GLDAS coupled with a river routing model are evaluated against observed streamflows in five river basins on the TP. The evaluation criteria include four aspects: monthly streamflow value, seasonal cycle of streamflow, annual streamflow trend, and streamflow component partitioning. The four LSMs display varying degrees of biases in monthly streamflow simulations: systematic overestimations are found in the Noah (1.74u2009≤u2009biasu2009≤u20092.75) and CLM (1.22u2009≤u2009biasu2009≤u20092.53) models, whereas systematic underestimations are observed in the VIC (0.36u2009≤u2009biasu2009≤u20090.85) and Mosaic (0.34u2009≤u2009biasu2009≤u20090.66) models. The Noah model shows the best performance in capturing the temporal variation in monthly streamflow and the seasonal cycle of streamflow, while the VIC model performs the best in terms of bias statistics. The Mosaic model provides the best performance in modeling annual runoff trends and runoff component partitioning. The possible reasons for the different performances of the LSMs are discussed in detail. In order to achieve more accurate streamflow simulations from the LSMs in GLDAS, suggestions are made to further improve the accuracy of the forcing data and parameterization schemes in all models.

Shuffled Complex-Self Adaptive Hybrid EvoLution (SC-SAHEL) Optimization Framework

Abstract Simplicity and flexibility of meta-heuristic optimization algorithms have attracted lots of attention in the field of optimization. Different optimization methods, however, hold algorithm-specific strengths and limitations, and selecting the best-performing algorithm for a specific problem is a tedious task. We introduce a new hybrid optimization framework, entitled Shuffled Complex-Self Adaptive Hybrid EvoLution (SC-SAHEL), which combines the strengths of different evolutionary algorithms (EAs) in a parallel computing scheme. SC-SAHEL explores performance of different EAs, such as the capability to escape local attractions, speed, convergence, etc., during population evolution as each individual EA suits differently to various response surfaces. The SC-SAHEL algorithm is benchmarked over 29 conceptual test functions, and a real-world hydropower reservoir model case study. Results show that the hybrid SC-SAHEL algorithm is rigorous and effective in finding global optimum for a majority of test cases, and that it is computationally efficient in comparison to algorithms with individual EA.