Integration of Hydrological Model and Time Series Model for Improving the Runoff Simulation: A Case Study on BTOP Model in Zhou River Basin, China

Abstract

Improving the accuracy of runoff simulations is a significant focus of hydrological science for multiple purposes such as water resources management, flood and drought prediction, and water environment protection. However, the simulated runoff has limitations that cannot be eliminated. This paper proposes a method that integrates the hydrological and time series models to improve the reliability and accuracy of simulated runoffs. Specifically, the block-wise use of TOPMODEL (BTOP) is integrated with three time series models to improve the simulated runoff from a hydrological model of the Zhou River Basin, China. Unlike most previous research that has not addressed the influence of runoff patterns while correcting the runoff, this study manually adds the hydrologic cycle to the machine learning-based time series model. This also incorporates scenario-specific knowledge from the researcher’s area of expertise into the prediction model. The results show that the improved Prophet model proposed in this study, that is, by adjusting its holiday function to a flow function, significantly improved the Nash–Sutcliffe efficiency (NSE) of the simulated runoff by 53.47% (highest) and 23.93% (average). The autoregressive integrated moving average (ARIMA) model and long short-term memory (LSTM) improved the runoff but performed less well than the improved Prophet model. This paper presents an effective method to improve the runoff simulation by integrating the hydrological and time series models.

Forecasting and Anomaly Detection in BEWS: Comparative Study of Theta, Croston, and Prophet Algorithms

“…Prophet outperformed two other algorithms, ARIMA and ThymeBoost, in predicting monthly rainfall data in India [53]. Xiao et al [54] found that Prophet improved runoff modeling in the Zhou River Basin. Bolick et al [55] used the Prophet method to predict hourly water level changes in an urban stream (South Carolina, USA) and obtained very accurate estimates, with coefficient of determination values greater than 0.9.…”

Evaluating Urban Stream Flooding with Machine Learning, LiDAR, and 3D Modeling

“…For example, Prophet outperformed two other algorithms, ARIMA, and ThymeBoost, in predicting monthly precipitation data in India [40]. Xiao et al [41] found that Prophet improved runoff model simulations in the Zhou River Basin. However, to our knowledge, no study has applied Prophet to predicting changes in water levels in streams or rivers.…”

Ungauged Basin Flood Prediction Using Long Short-Term Memory and Unstructured Social Media Data

“…Recently, various studies using neural network models for flood prediction have been conducted [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]. An artificial neural network (ANN) model is a data-driven model that can make predictions rapidly, owing to fewer computational requirements than existing physical models.…”

“…Liu et al [19] used the Global Flood Awareness System (GloFAS) and ERA5-Land hydro-meteorological data with a piecewise random forest to produce more accurate hydrological simulation results. Furthermore, Xiao et al and Zhu et al [20,21] employed the BTOP model for ungauged basins, resulting in a notable increase in the Nash-Sutcliffe efficiency (NSE). However, deep learning models, such as LSTM models, surpass accuracy stochastic (e.g., autoregressive integrated moving average; ARIMA) and shallow learning models [22].…”