Deep learning speeds up ice flow modelling by several orders of magnitude

This paper introduces the Instructed Glacier Model (IGM) – a model that simulates ice dynamics, mass balance and its coupling to predict the evolution of glaciers, icefields or ice sheets. The novelty of IGM is that it models the ice flow by a Convolutional Neural Network, which is trained from data generated with hybrid SIA + SSA or Stokes ice flow models. By doing so, the most computationally demanding model component is substituted by a cheap emulator. Once trained with representative data, we demonstrate that IGM permits to model mountain glaciers up to 1000 × faster than Stokes ones on Central Processing Units (CPU) with fidelity levels above 90% in terms of ice flow solutions leading to nearly identical transient thickness evolution. Switching to the GPU often permits additional significant speed-ups, especially when emulating Stokes dynamics or/and modelling at high spatial resolution. IGM is an open-source Python code which deals with two-dimensional (2-D) gridded input and output data. Together with a companion library of trained ice flow emulators, IGM permits user-friendly, highly efficient and mechanically state-of-the-art glacier and icefields simulations.

DEVELOPMENT OF AN ONLINE EDUCATION COMMUNITY SERVICE (OECS) MODEL BASED ON SULAM (SERVICE LEARNING MALAYSIA – UNIVERSITY FOR SOCIETY)

This paper proposes the development of an Online Education Community Service (OECS) Model based on SULAM (Service Learning Malaysia - University For Society). It aims to fulfil the needs for student community service activities that could not be conducted in-person or face-to-face due to movement restrictions during the COVID-19 pandemic. The suggestions in this paper are based on qualitative findings through literature reviews pertaining to community service among students. Several research methodologies and sampling techniques are proposed for the development of the model. One of these suggestions is adopting the Design and Development Research approach that could be developed in phases namely the needs analysis phase, the model component design phase, and the model development phase. The needs analysis phase involves reviewing literature using the content analysis technique. The model component design phase includes semi-structured interviews with students, lecturers, and community members whereas the questionnaire survey will be conducted among students only. Interview data is to be analysed using NVivo while questionnaire data from the pilot study and actual study will be analysed using SPSS. The model development phase involves conducting expert consensus workshops using the Fuzzy Delphi Method based on verified findings from the first and second phases of the study. Findings from the model development phase are to be analysed using Excels software. This proposed development of the SULAM-based OECS is in line with the Ministry of Higher Education’s aspiration to combine course learning outcomes with the aspect of community service to produce holistic students. Hence, students will be able to share their knowledge and skills obtained in university with the community in the effort to turn institutions of higher learning into Centres of Excellence. It is also in line with the country’s vision of achieving mutual prosperity by 2030 and the Sustainable Development Goals, specifically the goal of Quality Education.

MODEL-ORIENTED PROGRAMMING

The problem of complex multi-component system processing arises in many fields of science and engineering. A system can be described in terms of its components, behavior, and interaction. This work proposes a new declarative Turing complete “model-oriented” programming paradigm based on the concept of “model-component” - a complex structure with well-defined characteristics and behavior, and no external methods. The set of model-components is closed under the union operation of model-components into “model-complex”. The proposed approach allows the program to describe the complex system and behavior of its components in a declarative way, possesses a higher level of encapsulation than the object-oriented paradigm, involves a reduced amount of imperative programming, and is naturally focused on parallel computations.

A Novel Multiphysics Multiscale Multiporosity Shale Gas Transport Model for Geomechanics/Flow Coupling in Steady and Transient States

Summary A novel multiphysics multiscale multiporosity shale gas transport (M3ST) model was developed to investigate shale gas transport in both transient and steady states. The microscale model component contains a kerogen domain and an inorganic matrix domain, and each domain has its own geomechanical and gas transport properties. Permeabilities of various shale cores were measured in the laboratory using a pulse decay permeameter (PDP) with different pore pressure and confining stress combinations. The PDP-measured apparent permeability as a function of pore pressure under two effective stresses was fitted using the microscale M3ST model component based on nonlinear least squares fitting (NLSF), and the fitted model parameters were able to provide accurate model predictions for another effective stress. The parameters and petrophysical properties determined in the steady state were then used in the transient-state,continuum-scale M3ST model component, which performed history matching of the evolutions of the upstream and downstream gas pressures. In addition, a double-exponential empirical model was developed as a powerful alternative to the M3ST model to fit laboratory-measured apparent permeability under various effective stresses and pore pressures. The developed M3ST model and the research findings in this study provided critical insights into the role of the multiphysics mechanisms, including geomechanics, fluid dynamics and transport, and the Klinkenberg effect on shale gas transport across different spatial scales in both steady and transient states.

Research on Runoff Simulations Using Deep-Learning Methods

Runoff simulations are of great significance to the planning management of water resources. Here, we discussed the influence of the model component, model parameters and model input on runoff modeling, taking Hanjiang River Basin as the research area. Convolution kernel and attention mechanism were introduced into an LSTM network, and a new data-driven model Conv-TALSTM was developed. The model parameters were analyzed based on the Conv-TALSTM, and the results suggested that the optimal parameters were greatly affected by the correlation between the input data and output data. We compared the performance of Conv-TALSTM and variant models (TALSTM, Conv-LSTM, LSTM), and found that Conv-TALSTM can reproduce high flow more accurately. Moreover, the results were comparable when the model was trained with meteorological or hydrological variables, whereas the peak values with hydrological data were closer to the observations. When the two datasets were combined, the performance of the model was better. Additionally, Conv-TALSTM was also compared with an ANN (artificial neural network) and Wetspa (a distributed model for Water and Energy Transfer between Soil, Plants and Atmosphere), which verified the advantages of Conv-TALSTM in peak simulations. This study provides a direction for improving the accuracy, simplifying model structure and shortening calculation time in runoff simulations.

Research on model component method based on OpenMI technology

Open Model Interface (OpenMI) technology can construct standardized components with unified timing interface and standardized data interaction. This provides a standardized method for integrating multiple models with different interfaces and functions. This paper briefly introduces the operation principle of OpenMI technology and the process of model standardization, and describes the seven key steps of model transplantation by taking the one-dimensional hydraulic model of the Yellow River as an example.