Hydrologically Informed Machine Learning for Rainfall-Runoff Modelling: Towards Distributed Modelling

Abstract

Abstract. Despite showing a great success of applications in many commercial fields, machine learning and data science models in general, show a limited use in scientific fields including hydrology. The approach is often criticized for lack of interpretability and physical consistency. This has led to the emergence of new paradigms, such as Theory Guided Data Science (TGDS) and physics informed machine learning. The motivation behind such approaches is to improve the physical meaningfulness of machine learning models by blending existing scientific knowledge with learning algorithms. Following the same principles, in our prior work (Chadalawada et al., 2020), a new model induction framework was founded on Genetic Programming (GP) namely Machine Learning Rainfall-Runoff Model Induction Toolkit (ML-RR-MI). ML-RR-MI is cable of developing fully-fledged lumped conceptual rainfall-runoff models for a watershed of interest using the building blocks of two flexible rainfall-runoff modelling frameworks (FUSE and SUPERFLEX). In this study, we extend ML-RR-MI towards inducing semi-distributed rainfall-runoff models. This effort is motivated by the desire to address the decreasing meaningfulness of lumped models which tend to particularly deteriorate within large catchments where the spatial heterogeneity of forcing variables and watershed properties are significant. Henceforth, our machine learning approach for rainfall-runoff modelling titled Machine Induction Knowledge-Augmented System Hydrologique Asiatique (MIKA-SHA) captures spatial variabilities and automatically induces rainfall-runoff models for the catchment of interest without any subjectivity in model selection. Currently, MIKA-SHA learns models utilizing the model building components of FUSE and SUPERFLEX. However, the proposed framework can be coupled with any internally coherent collection of building blocks. MIKA-SHA’s model induction capabilities have been tested on the Red Creek catchment near Vestry, Mississippi, United States. The resulted model architectures through MIKA-SHA are compatible with previously reported research findings and fieldwork insights of the watershed and are readily interpretable by hydrologists.

The Data Synergy Effects of Time‐Series Deep Learning Models in Hydrology

“…This setup permitted simultaneous training and simulation over thousands of sites or more. However, in many other machine learning studies, following the conventional wisdom of stratification, geoscientists still tend to train separate models using data from each site (Duan et al., 2020; Herath et al., 2021; Petty & Dhingra, 2018), or each region composed of sites with similar environmental conditions (Abdalla et al., 2021; Sahoo et al., 2017).…”

Augmentation of the Urban Green Infrastructure Using Stormwater Surface Runoff as a Resource in the Nice Expressway, Karnataka, India

“…Conventional rainfall runoff models face problems in the fact that they do not look in depth into the latest problems arising with the effects of change in land use and land cover owing to agricultural practices, forestry, pollution and toxic wastes released into water bodies (Elaji & Ji, 2020). To overcome this issue, physically-distributed models like the European Hydrological System -Systeme Hydrologique European or SHE model and IHDM deal effectively with the changes in land cover and land use or in drainage patterns in catchments (Herath et al, 2020).…”

“…Advanced approaches in flexible monitoring and prediction can be associated in recent decades with a powerful machine learning technique called Genetic Programming (GP) proposed as a modelling and forecasting tool, although a recent introduction to the field (Chadalawada et al, 2020;Herath et al, 2020). GP being a latest addition to a highly emerging sub field in computer science called Evolutionary Computation, works on the lines of its parent's theories.…”

Data Assimilation for Rainfall-Runoff Prediction Based on Coupled Atmospheric-Hydrologic Systems with Variable Complexity

“…The hydrological model has a more comprehensive physical foundation including lumped, grid-based, and fully distributed setups [18,19]. Not only the above physically-based models are used, but machine learning models are also widely applied in rainfall-runoff forecasting (i.e., artificial neural networks (ANNs) [20], support vector machines (SVM) [21], and the recent emergence of theory-guided data science (TGDS) [22,23]. For flood forecasting, which is affected by the discretization construction method, different construction expressions determine variations between heterogeneity analysis and model calculation structure, and further influence the accuracy of physical expressions in the prediction processes of the hydrological model [24][25][26].…”