Progresses in integrated application methods of geographic analysis models for virtual geographic environment construction

Geographic analysis models serve as crucial resources for tackling geo-problems, simulating geographic environments, and supporting decision-making analysis. They play a key role in constructing virtual geographic environments (also known as digital twin geographic environments). Integrating geographic analysis models with diverse simulation capabilities from various fields enables a comprehensive representation of real geographical environments. This integration supports the development of visual, usable, analyzable, and interactive virtual geographic environments. This study systematically examines the current research progress in the integration methods of geographic analysis models. In general, integrating physical and social processes and merging big data with deep learning have gained increased attention in recent years. Compared with using a single model, multimodel integration enhances the understanding of the objective geographical world by integrating multiple elements and processes. Integrating models within a single domain or across domains ultimately involves basic execution links, such as combining, nesting, and connecting multiple models in a specific computing environment. In response to this, customized integration and modular integration are the main approaches to current geographical analysis model integration research. Customized integration offers tailored solutions for specific models considering cognitive and technical characteristics. However, it is heavily influenced by individual model characteristics and may become unsustainable during the integration of numerous models, leading to frequent revision activities. Modular integration relies on standardized components with clear association logic and supports component replacement. However, it faces strong technical dependencies and constraints, such as those that involve programming language, data structure, and operating environment. In practice, current model integration research often varies between customization and modularity, with the former lacking sustainability and the latter being limited by compatibility issues. In this study, the idea of a “geographic analysis model container” is proposed, and the methods are implemented to address the problems mentioned. With the assistance of the concept of containerization, the data, programs, and computing resources upon which geographical models rely are loaded into a “container.” Then, the model is developed in a customized manner within the container, while the modular model integration work occurs outside the container. Striking a balance between customized and modular approaches is anticipated to enhance the overall effectiveness of geographic modeling and simulation efforts. Conducting a multilevel, multigranularity, and multiscale integration across elements, processes, and functional relationships is necessary for addressing specific geographical problems. The development of “geographic analysis model containers” tailored to express geographical system evolution laws and driving mechanisms has become crucial. Accumulating and improving model integration capabilities based on containers is expected to establish an effective bridge between model construction, transformation, and integration. This advancement can promote geographical models from experimental tools to essential components of social service infrastructure. Decision-making analysis, solution simulation, and customized planning capabilities can be provided for various digital twin geographical scenarios (such as digital twin cities, digital twin river basins, and digital twin water conservancies) by integrating geographical analysis models. This integration drives advancements in digital twins, metaverses, and digital human technology, spanning visual experiences, operation, and maintenance management. It also incorporates geographical knowledge, enhances interactive functions, and promotes a comprehensive understanding. © 2024 Science Press. All rights reserved.