This page has only limited features, please log in for full access.
The objective of the current study is to present a comparison of techniques for the forecasting of low frequency climate oscillation indices with a focus on the Great Lakes system. A number of time series models have been tested including the traditional Autoregressive Moving Average (ARMA) model, Dynamic Linear model (DLM), Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model, as well as the nonstationary oscillation resampling (NSOR) technique. These models were used to forecast the monthly El Niño-Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO) indices which show the most significant teleconnection with the net basin supply (NBS) of the Great Lakes system from a preliminary study. The overall objective is to predict future water levels, ice extent, and temperature, for planning and decision making purposes. The results showed that the DLM and GARCH models are superior for forecasting the monthly ENSO index, while the forecasted values from the traditional ARMA model presented a good agreement with the observed values within a short lead time ahead for the monthly PDO index.
Taesam Lee; Taha B.M.J. Ouarda; Ousmane Seidou. Characterizing and Forecasting Climate Indices Using Time Series Models. 2021, 1 .
AMA StyleTaesam Lee, Taha B.M.J. Ouarda, Ousmane Seidou. Characterizing and Forecasting Climate Indices Using Time Series Models. . 2021; ():1.
Chicago/Turabian StyleTaesam Lee; Taha B.M.J. Ouarda; Ousmane Seidou. 2021. "Characterizing and Forecasting Climate Indices Using Time Series Models." , no. : 1.
Nonstationarity is one major issue in hydrological models, especially in design rainfall analysis. Design rainfalls are typically estimated by annual maximum rainfalls (AMRs) of observations below 50 years in many parts of the world, including South Korea. However, due to the lack of data, the time-dependent nature may not be sufficiently identified by this classic approach. Here, this study aims to explore design rainfall with nonstationary condition using century-long reanalysis products that help one to go back to the early 20th century. Despite its useful representation of the past climate, the reanalysis products via observational data assimilation schemes and models have never been tested in representing the nonstationary behavior in extreme rainfall events. We used daily precipitations of two century-long reanalysis datasets as the ERA-20c by the European Centre for Medium-Range Weather Forecasts (ECMWF) and the 20th century reanalysis (20CR) by the National Oceanic and Atmospheric Administration (NOAA). The AMRs from 1900 to 2010 were derived from the grids over South Korea. The systematic errors were downgraded through quantile delta mapping (QDM), as well as conventional stationary quantile mapping (SQM). The evaluation result of the bias-corrected AMRs indicated the significant reduction of the errors. Furthermore, the AMRs present obvious increasing trends from 1900 to 2010. With the bias-corrected values, we carried out nonstationary frequency analysis based on the time-varying location parameters of generalized extreme value (GEV) distribution. Design rainfalls with certain return periods were estimated based on the expected number of exceedance (ENE) interpretation. Although there is a significant range of uncertainty, the design quantiles by the median parameters showed the significant relative difference, from −30.8% to 42.8% for QDM, compared with the quantiles by the multi-decadal observations. Even though the AMRs from the reanalysis products are challenged by various errors such as quantile mapping (QM) and systematic errors, the results from the current study imply that the proposed scheme with employing the reanalysis product might be beneficial to predict the future evolution of extreme precipitation and to estimate the design rainfall accordingly.
Dong-Ik Kim; Dawei Han; Taesam Lee. Reanalysis Product-Based Nonstationary Frequency Analysis for Estimating Extreme Design Rainfall. Atmosphere 2021, 12, 191 .
AMA StyleDong-Ik Kim, Dawei Han, Taesam Lee. Reanalysis Product-Based Nonstationary Frequency Analysis for Estimating Extreme Design Rainfall. Atmosphere. 2021; 12 (2):191.
Chicago/Turabian StyleDong-Ik Kim; Dawei Han; Taesam Lee. 2021. "Reanalysis Product-Based Nonstationary Frequency Analysis for Estimating Extreme Design Rainfall." Atmosphere 12, no. 2: 191.
Deep learning models have been applied for hydrometeorological applications. In this chapter, a few of them are explained. In the field of hydrometeorology, time-series deep learning models are mainly employed. In this chapter, the development procedure of a time series deep learning model for stochastic simulation producing a long sequence that mimics historical series is explained. Furthermore, the case study for daily maximum temperature with an LSTM model is presented.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Hydrometeorological Applications of Deep Learning. Climate Change Impacts on Water Resources 2021, 163 -190.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Hydrometeorological Applications of Deep Learning. Climate Change Impacts on Water Resources. 2021; ():163-190.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Hydrometeorological Applications of Deep Learning." Climate Change Impacts on Water Resources , no. : 163-190.
In this chapter, recently developed neural network algorithms are introduced, including extreme learning machine and autoencoder. These algorithms are popularly adopted in deep learning models. Particularly, autoencoder can be used in shrinking the dimension of inputs and outputs as well as hidden nodes.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Advanced Neural Network Algorithms. Climate Change Impacts on Water Resources 2021, 87 -106.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Advanced Neural Network Algorithms. Climate Change Impacts on Water Resources. 2021; ():87-106.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Advanced Neural Network Algorithms." Climate Change Impacts on Water Resources , no. : 87-106.
Before applying a neural network model, the data must be preprocessed in advance. Data normalization must be made to avoid the difference of data variability for inputs and outputs. It also simplifies the parameter range of a network model. Furthermore, data should be split for different purposes such as training, validation and testing. In this chapter, these data normalization and data split are explained in detail.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Data Preprocessing. Climate Change Impacts on Water Resources 2021, 21 -25.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Data Preprocessing. Climate Change Impacts on Water Resources. 2021; ():21-25.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Data Preprocessing." Climate Change Impacts on Water Resources , no. : 21-25.
In this current chapter, the fundamental mathematical background is presented for a deep learning model. Linear simple and multiple regression models are explained, including the definition of error terms and parameter estimation procedure, since they are similarly used in deep learning models. Also, the basic concept of the time series model is also explained and this part is mainly referred to in the LSTM model chapter.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Mathematical Background. Climate Change Impacts on Water Resources 2021, 5 -19.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Mathematical Background. Climate Change Impacts on Water Resources. 2021; ():5-19.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Mathematical Background." Climate Change Impacts on Water Resources , no. : 5-19.
A neural network is composed of a number of parameters, also called weights. The neural network training is the estimation procedure of those parameters. In this chapter, the training procedure of a neural network is described based on the gradient descent method and the backpropagation algorithm.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Training a Neural Network. Climate Change Impacts on Water Resources 2021, 47 -62.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Training a Neural Network. Climate Change Impacts on Water Resources. 2021; ():47-62.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Training a Neural Network." Climate Change Impacts on Water Resources , no. : 47-62.
Among recent developments of deep learning models, the availability of spatial datasets or images with deep learning is the most significant contribution. In this chapter, a convolutional neural network (CNN) that can analyze the spatial datasets is described. The training procedure of CNN is described with a simple example.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Deep Learning for Spatial Datasets. Climate Change Impacts on Water Resources 2021, 133 -150.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Deep Learning for Spatial Datasets. Climate Change Impacts on Water Resources. 2021; ():133-150.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Deep Learning for Spatial Datasets." Climate Change Impacts on Water Resources , no. : 133-150.
Deep learning model is composed of several layers of neural networks. Therefore, the basic concepts and terminology of a neural network are introduced. The simplest neural network model is introduced and used in the latter part of this book. Then, a full neural network model is described. The parameter estimation procedure employing backward propagation is also explained.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Neural Network. Climate Change Impacts on Water Resources 2021, 27 -46.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Neural Network. Climate Change Impacts on Water Resources. 2021; ():27-46.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Neural Network." Climate Change Impacts on Water Resources , no. : 27-46.
Deep learning has been popularly employed for analysis and forecasting in various fields. In this chapter, a brief introduction of deep learning is presented, including the definition and pros and cons of deep learning, followed by the recent applications of deep learning models in hydrological and environmental fields. The structure of the remaining chapters for this book is also explained.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Introduction. Climate Change Impacts on Water Resources 2021, 1 -4.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Introduction. Climate Change Impacts on Water Resources. 2021; ():1-4.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Introduction." Climate Change Impacts on Water Resources , no. : 1-4.
Tensorflow is an end-to-end open-source platform for machine learning containing a comprehensive, flexible ecosystem of tools, libraries, and community resources (https://www.tensorflow.org/). It provides multiple levels of abstractions to choose the right one. The high-level Keras API can be used to build and train models by easily getting started with Tensorflow. Keras allows employing Tensorflow without losing its flexibility and capability. In the following, two applications (i.e., temporal and spatial deep learning) are presented to illustrate how to use Keras with python.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Tensorflow and Keras Programming for Deep Learning. Climate Change Impacts on Water Resources 2021, 151 -162.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Tensorflow and Keras Programming for Deep Learning. Climate Change Impacts on Water Resources. 2021; ():151-162.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Tensorflow and Keras Programming for Deep Learning." Climate Change Impacts on Water Resources , no. : 151-162.
Training a neural network is to update the weights to minimize a specified loss function and the gradient descent method has been employed. However, the number of weights exponentially grows, especially in a deep learning machine. In recent years, several methods updating weights have been developed to improve the speed of convergence and to find the best trajectory to reach the optimum of the employed loss function for a network. In this chapter, those methods for updating weights are explained.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Updating Weights. Climate Change Impacts on Water Resources 2021, 63 -78.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Updating Weights. Climate Change Impacts on Water Resources. 2021; ():63-78.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Updating Weights." Climate Change Impacts on Water Resources , no. : 63-78.
In order improve the performance of a neural network model, a number of ways have been studied. In this chapter, minibatch and k-fold cross-validation are explained. The basic idea of these two methods is on controlling the dataset, since repeated usage of the same dataset for training and validation might result in overfitting. Furthermore, regularization of the neural network model training by L-norm regularization and dropout of hidden nodes are explained in this chapter to avoid overfitting.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Improving Model Performance. Climate Change Impacts on Water Resources 2021, 79 -86.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Improving Model Performance. Climate Change Impacts on Water Resources. 2021; ():79-86.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Improving Model Performance." Climate Change Impacts on Water Resources , no. : 79-86.
One of the major applications in deep learning models is to forecast the future. In recent years, time series forecasting with deep learning models has been developed and applied in a number of fields. Recurrent neural network models can allow forecasting future better, and long short-term memory (LSTM) is a breakthrough to overcome the shortages of the previous RNN model. These algorithms are explained in detail in this chapter.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Deep Learning for Time Series. Climate Change Impacts on Water Resources 2021, 107 -131.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Deep Learning for Time Series. Climate Change Impacts on Water Resources. 2021; ():107-131.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Deep Learning for Time Series." Climate Change Impacts on Water Resources , no. : 107-131.
A case study of CNN, a spatial deep learning model, is presented in this chapter for analyzing water quality with remote sensing data. Airborne remote sensing of cyanobacteria with multispectral/hyperspectral sensors is employed for input data and its water quality as output data.
Taesam Lee; Vijay P. Singh; Kyung Hwa Cho. Environmental Applications of Deep Learning. Climate Change Impacts on Water Resources 2021, 191 -204.
AMA StyleTaesam Lee, Vijay P. Singh, Kyung Hwa Cho. Environmental Applications of Deep Learning. Climate Change Impacts on Water Resources. 2021; ():191-204.
Chicago/Turabian StyleTaesam Lee; Vijay P. Singh; Kyung Hwa Cho. 2021. "Environmental Applications of Deep Learning." Climate Change Impacts on Water Resources , no. : 191-204.
Weather forecasting, especially that of extreme climatic events, has gained considerable attention among researchers due to their impacts on natural ecosystems and human life. The applicability of artificial neural networks (ANNs) in non-linear process forecasting has significantly contributed to hydro-climatology. The efficiency of neural network functions depends on the network structure and parameters. This study proposed a new approach to forecasting a one-day-ahead maximum temperature time series for South Korea to discuss the relationship between network specifications and performance by employing various scenarios for the number of parameters and hidden layers in the ANN model. Specifically, a different number of trainable parameters (i.e., the total number of weights and bias) and distinctive numbers of hidden layers were compared for system-performance effects. If the parameter sizes were too large, the root mean square error (RMSE) would be generally increased, and the model’s ability was impaired. Besides, too many hidden layers would reduce the system prediction if the number of parameters was high. The number of parameters and hidden layers affected the performance of ANN models for time series forecasting competitively. The result showed that the five-hidden layer model with 49 parameters produced the smallest RMSE at most South Korean stations.
Trang Tran; Taesam Lee; Jong-Suk Kim. Increasing Neurons or Deepening Layers in Forecasting Maximum Temperature Time Series? Atmosphere 2020, 11, 1072 .
AMA StyleTrang Tran, Taesam Lee, Jong-Suk Kim. Increasing Neurons or Deepening Layers in Forecasting Maximum Temperature Time Series? Atmosphere. 2020; 11 (10):1072.
Chicago/Turabian StyleTrang Tran; Taesam Lee; Jong-Suk Kim. 2020. "Increasing Neurons or Deepening Layers in Forecasting Maximum Temperature Time Series?" Atmosphere 11, no. 10: 1072.
Streamflow estimation plays a significant role in water resources management, especially for flood mitigation, drought warning, and reservoir operation. Hence, the current study examines the prediction capability of three well-known machine learning algorithms (Support Vector Regression (SVR), Artificial Neural Network with backpropagation (ANN-BP), and Extreme Learning Machine (ELM)) for the monthly and daily streamflows of four rivers in the United States. For model development, three main predictor variables (P, Tmax, and Tmin) and their antecedent values were considered. The SVM-RFE feature selection method was used to select the most appropriate predictor variable.The performance of the developed models was tested using four evaluation statistics. The results indicate that (1) except some improvements, the accuracy of all models decreases at the daily scale compared to that at the monthly scale; (2) the SVR has the best performance among the three models at the monthly and daily scales, while the ANN-BP model has the worse performance; (3) the ELM has better generalization performance than the ANN-BP for streamflow simulation at the monthly and daily scales; and (4) all models fail to predict the streamflow for the Carson River as a snowmelt-dominated basin. Generally, findings of the current study indicate that the SVR model produces better results than the ELM and ANN-BP for streamflow simulation at the monthly and daily scales.
Peiman Parisouj; Hamid Mohebzadeh; Taesam Lee. Employing Machine Learning Algorithms for Streamflow Prediction: A Case Study of Four River Basins with Different Climatic Zones in the United States. Water Resources Management 2020, 34, 4113 -4131.
AMA StylePeiman Parisouj, Hamid Mohebzadeh, Taesam Lee. Employing Machine Learning Algorithms for Streamflow Prediction: A Case Study of Four River Basins with Different Climatic Zones in the United States. Water Resources Management. 2020; 34 (13):4113-4131.
Chicago/Turabian StylePeiman Parisouj; Hamid Mohebzadeh; Taesam Lee. 2020. "Employing Machine Learning Algorithms for Streamflow Prediction: A Case Study of Four River Basins with Different Climatic Zones in the United States." Water Resources Management 34, no. 13: 4113-4131.
Effective water quality monitoring of coastal areas through the measurement of Chlorophyll-a (Chl-a) has remarkably progressed by ocean color remote sensing. Among different sensors, Moderate Resolution Imaging Spectroradiometer (MODIS) Level 3 products provide reliable global representations of the Chl-a concentration. On the other hand, due to the coarse spatial resolution of MODIS data, its applicability is limited for spatially complex coastal regions. To overcome this limitation, a few downscaling techniques have been suggested based on the polynomial regression method. However, this type of regression has some restrictions, such as sensitivity to outliers, and nonlinear types of machine learning algorithms have not been tested in downscaling Chl-a datasets. Therefore, three machine learning (ML) techniques, support vector regression (SVR), random forest regression (RFR), and long short-term memory (LSTM), were developed using the Sentinel-2A/MSI bands as predictors and MODIS Chl-a as a predictand and compared their results with the results of multiple polynomial regression (MPR), to find the most suitable model for downscaling MODIS Chl-a in coastal area of South Korea. The obtained results showed that the 2nd degree MPR and SVR-Radial Basis Function (RBF) illustrate the best performance in the winter and summer days, respectively. In addition, LSTM is less sensitive to the changes in all variables (sensitivity index range from 0.31 to 0.48). Overall, we conclude that the downscaling approach based on ML models, especially SVR-RBF, can serve as a suitable alternative in some cases to produce high-resolution Chl-a maps, especially for coastal marine water quality monitoring.
Hamid Mohebzadeh; Taesam Lee. Spatial downscaling of MODIS Chlorophyll-a with machine learning techniques over the west coast of the Yellow Sea in South Korea. Journal of Oceanography 2020, 77, 103 -122.
AMA StyleHamid Mohebzadeh, Taesam Lee. Spatial downscaling of MODIS Chlorophyll-a with machine learning techniques over the west coast of the Yellow Sea in South Korea. Journal of Oceanography. 2020; 77 (1):103-122.
Chicago/Turabian StyleHamid Mohebzadeh; Taesam Lee. 2020. "Spatial downscaling of MODIS Chlorophyll-a with machine learning techniques over the west coast of the Yellow Sea in South Korea." Journal of Oceanography 77, no. 1: 103-122.
The prediction of a time series such as climate indices and the sunspot number (SSN) with long-term oscillatory behaviors has been a challenging task due to the complex combination of oscillations. Frequency extraction algorithms have been developed to separate a time series into different oscillation components according to frequency, such as empirical model decomposition (EMD) and wavelet analysis. In the current study, the deep learning long short-term memory (LSTM) model was employed to predict the oscillation components extracted using EMD. The SSN series was modeled with the hybrid EMD-LSTM model. The simulation study results indicate that the LSTM model reproduces the smooth cyclic pattern of the sine function, and only a few hidden units are needed to model it. The EMD-LSTM model achieves better performance than does the LSTM model for mid-range SSN predictions while the LSTM achieves better performance within the first few time lags. However, the cyclic prediction of the SSN requires mid-range lags; thus, the superior performance of the EMD-LSTM model for these lags cannot be ignored. Furthermore, the remaining components from the significant EMD signals can be modeled to reveal the variability (or uncertainty) in the prediction. The summed residual is fitted by k-nearest neighbor resampling. The final SSN prediction results show that the EMD-LSTM model predicts a later and larger SSN for Solar Cycle 25 than does the LSTM model. Overall, the results lead to the conclusion that the EMD-LSTM model might be a suitable alternative for modeling complex sunspot time series with cyclic patterns.
Taesam Lee. EMD and LSTM Hybrid Deep Learning Model for Predicting Sunspot Number Time Series with a Cyclic Pattern. Solar Physics 2020, 295, 1 -23.
AMA StyleTaesam Lee. EMD and LSTM Hybrid Deep Learning Model for Predicting Sunspot Number Time Series with a Cyclic Pattern. Solar Physics. 2020; 295 (6):1-23.
Chicago/Turabian StyleTaesam Lee. 2020. "EMD and LSTM Hybrid Deep Learning Model for Predicting Sunspot Number Time Series with a Cyclic Pattern." Solar Physics 295, no. 6: 1-23.
Time series forecasting of meteorological variables such as daily temperature has recently drawn considerable attention from researchers to address the limitations of traditional forecasting models. However, a middle-range (e.g., 5–20 days) forecasting is an extremely challenging task to get reliable forecasting results from a dynamical weather model. Nevertheless, it is challenging to develop and select an accurate time-series prediction model because it involves training various distinct models to find the best among them. In addition, selecting an optimum topology for the selected models is important too. The accurate forecasting of maximum temperature plays a vital role in human life as well as many sectors such as agriculture and industry. The increase in temperature will deteriorate the highland urban heat, especially in summer, and have a significant influence on people’s health. We applied meta-learning principles to optimize the deep learning network structure for hyperparameter optimization. In particular, the genetic algorithm (GA) for meta-learning was used to select the optimum architecture for the network used. The dataset was used to train and test three different models, namely the artificial neural network (ANN), recurrent neural network (RNN), and long short-term memory (LSTM). Our results demonstrate that the hybrid model of an LSTM network and GA outperforms other models for the long lead time forecasting. Specifically, LSTM forecasts have superiority over RNN and ANN for 15-day-ahead in summer with the root mean square error (RMSE) value of 2.719 (°C).
Trang Thi Kieu Tran; Taesam Lee; Ju-Young Shin; Jong-Suk Kim; Mohamad Kamruzzaman. Deep Learning-Based Maximum Temperature Forecasting Assisted with Meta-Learning for Hyperparameter Optimization. Atmosphere 2020, 11, 487 .
AMA StyleTrang Thi Kieu Tran, Taesam Lee, Ju-Young Shin, Jong-Suk Kim, Mohamad Kamruzzaman. Deep Learning-Based Maximum Temperature Forecasting Assisted with Meta-Learning for Hyperparameter Optimization. Atmosphere. 2020; 11 (5):487.
Chicago/Turabian StyleTrang Thi Kieu Tran; Taesam Lee; Ju-Young Shin; Jong-Suk Kim; Mohamad Kamruzzaman. 2020. "Deep Learning-Based Maximum Temperature Forecasting Assisted with Meta-Learning for Hyperparameter Optimization." Atmosphere 11, no. 5: 487.