Short-term forecasts of streamflow in the UK based on a novel hybrid artificial intelligence algorithm

Abstract

In recent years, the growing impact of climate change on surface water bodies has made the analysis and forecasting of streamflow rates essential for proper planning and management of water resources. This study proposes a novel ensemble (or hybrid) model, based on the combination of a Deep Learning algorithm, the Nonlinear AutoRegressive network with eXogenous inputs, and two Machine Learning algorithms, Multilayer Perceptron and Random Forest, for the short-term streamflow forecasting, considering precipitation as the only exogenous input and a forecast horizon up to 7 days. A large regional study was performed, considering 18 watercourses throughout the United Kingdom, characterized by different catchment areas and flow regimes. In particular, the predictions obtained with the ensemble Machine Learning-Deep Learning model were compared with the ones achieved with simpler models based on an ensemble of both Machine Learning algorithms and on the only Deep Learning algorithm. The hybrid Machine Learning-Deep Learning model outperformed the simpler models, with values of R2 above 0.9 for several watercourses, with the greatest discrepancies for small basins, where high and non-uniform rainfall throughout the year makes the streamflow rate forecasting a challenging task. Furthermore, the hybrid Machine Learning-Deep Learning model has been shown to be less affected by reductions in performance as the forecasting horizon increases compared to the simpler models, leading to reliable predictions even for 7-day forecasts.

Streamflow Predictions in Ungauged Basins Using Recurrent Neural Network and Decision Tree-Based Algorithm: Application to the Southern Region of the Korean Peninsula

“…Hybrid models combining machine learning and deep learning techniques have been applied in studies to forecast streamflow in gauged basins [21,22], but most studies that attempt to predict streamflow in ungauged watersheds apply machine learning techniques such as Random Forest or use deep learning techniques such as LSTM [23,24,[26][27][28]30]. In this study, deep learning techniques were used to initially predict the behavior of streamflow itself, and then machine learning techniques linked to meteorological data and watershed characteristics were used to finally predict streamflow.…”

“…They claimed that their model has a similar streamflow forecasting performance to bi-directional LSTM. Similarly, Di Nunno et al [22] proposed a NARX (Nonlinear AutoRegressive network with eXogenous inputs)-MLP (Multilayer Perceptron)-RF (Random Forest) hybrid model for streamflow forecasting.…”

An active learning convolutional neural network for predicting river flow in a human impacted system

“…If certain input data being used for a prediction is better suited to one of the aggregated ML methods than another, the weightings are not updated. Furthermore, hybrid methods have been developed (Di Nunno et al, 2023;Granata and Di Nunno, 2023) that combine both machine learning and deep learning methods to forecast streamflow. These also do not adapt to data on the fly.…”

Assessing the reliability of a physical-based model and a convolutional neural network in an ungauged watershed for daily streamflow calculation: a case study in southern Portugal

“…For instance, Si et al ( 2021) considered a graphical convolutional GRU model to predict the streamflow in the next 36 h hours, while Szczepanek (2022) used three different models, namely, XGBoost, LightGBM, and CatBoost, for daily streamflow forecast. Additionally, hybrid solutions considering different machine learning algorithms, such as Di Nunno et al (2023) and Yu et al (2023), are becoming widely used and with improved results.…”