WaterGPT: Training a Large Language Model to Become a Hydrology Expert DOI Open Access
Yi Ren, Tianyi Zhang,

Xurong Dong

et al.

Water, Journal Year: 2024, Volume and Issue: 16(21), P. 3075 - 3075

Published: Oct. 27, 2024

This paper introduces WaterGPT, a language model designed for complex multimodal tasks in hydrology. WaterGPT is applied three main areas: (1) processing and analyzing data such as images text water resources, (2) supporting intelligent decision-making hydrological tasks, (3) enabling interdisciplinary information integration knowledge-based Q&A. The has achieved promising results. One core aspect of involves the meticulous segmentation training supervised fine-tuning phase, sourced from real-world annotated with high quality using both manual methods GPT-series annotations. These are carefully categorized into four types: knowledge-based, task-oriented, negative samples, multi-turn dialogues. Additionally, another key component development multi-agent framework called Water_Agent, which enables to intelligently invoke various tools solve field resources. handles data, including images, allowing deep understanding analysis environments. Based on this framework, over 90% success rate object detection waterbody extraction. For extraction task, Dice mIoU metrics, WaterGPT’s performance high-resolution 2013 2022 remained stable, accuracy exceeding 90%. Moreover, we have constructed high-quality resources evaluation dataset, EvalWater, covers 21 categories approximately 10,000 questions. Using highest date reaching 83.09%, about 17.83 points higher than GPT-4.

Language: Английский

Multi-Agent Large Language Model Frameworks: Unlocking New Possibilities for Optimizing Wastewater Treatment Operation DOI

Samuel Rothfarb,

Mikayla Friday,

Xingyu Wang

et al.

Environmental Research, Journal Year: 2025, Volume and Issue: unknown, P. 121401 - 121401

Published: March 1, 2025

Language: Английский

Citations

0
Urban Flood Modelling: Challenges and Opportunities - A Stakeholder-Informed Analysis DOI Creative Commons
Muhammad Qasim Mahmood, Xiuquan Wang, Farhan Aziz

et al.

Environmental Modelling & Software, Journal Year: 2025, Volume and Issue: unknown, P. 106507 - 106507

Published: April 1, 2025

Language: Английский

Citations

0
Artificial Intelligence in Aquatic Biodiversity Research: A PRISMA-Based Systematic Review DOI Creative Commons
Tymoteusz Miller, Grzegorz Michoński, Irmina Durlik

et al.

Biology, Journal Year: 2025, Volume and Issue: 14(5), P. 520 - 520

Published: May 8, 2025

Freshwater ecosystems are increasingly threatened by climate change and anthropogenic activities, necessitating innovative scalable monitoring solutions. Artificial intelligence (AI) has emerged as a transformative tool in aquatic biodiversity research, enabling automated species identification, predictive habitat modeling, conservation planning. This systematic review follows the PRISMA framework to analyze AI applications freshwater studies. Using structured literature search across Scopus, Web of Science, Google Scholar, we identified 312 relevant studies published between 2010 2024. categorizes into assessment, ecological risk evaluation, strategies. A bias assessment was conducted using QUADAS-2 RoB 2 frameworks, highlighting methodological challenges, such measurement inconsistencies model validation. The citation trends demonstrate exponential growth AI-driven with leading contributions from China, United States, India. Despite growing use this field, also reveals several persistent including limited data availability, regional imbalances, concerns related generalizability transparency. Our findings underscore AI’s potential revolutionizing but emphasize need for standardized methodologies, improved integration, interdisciplinary collaboration enhance insights efforts.

Language: Английский

Citations

0
Deep Learning Prediction of Streamflow in Portugal DOI Creative Commons
Rafael Francisco, José Pedro Matos

Hydrology, Journal Year: 2024, Volume and Issue: 11(12), P. 217 - 217

Published: Dec. 19, 2024

The transformative potential of deep learning models is felt in many research fields, including hydrology and water resources. This study investigates the effectiveness Temporal Fusion Transformer (TFT), a neural network architecture for predicting daily streamflow Portugal, benchmarks it against popular Hydrologiska Byråns Vattenbalansavdelning (HBV) hydrological model. Additionally, evaluates performance TFTs through selected forecasting examples. Information provided about key input variables, precipitation, temperature, geomorphological characteristics. involved extensive hyperparameter tuning, with over 600 simulations conducted to fine–tune performances ensure reliable predictions across diverse conditions. results showed that outperformed HBV model, successfully several catchments distinct characteristics throughout country. not only provide trustworthy associated probabilities occurrence but also offer considerable advantages classical frameworks, i.e., ability model complex temporal dependencies interactions different inputs or weight features based on their relevance target variable. Multiple practical applications can rely made TFT models, such as flood risk management, resources allocation, support climate change adaptation measures.

Language: Английский

Citations

1
WaterGPT: Training a Large Language Model to Become a Hydrology Expert DOI Open Access
Yi Ren, Tianyi Zhang,

Xurong Dong

et al.

Water, Journal Year: 2024, Volume and Issue: 16(21), P. 3075 - 3075

Published: Oct. 27, 2024

This paper introduces WaterGPT, a language model designed for complex multimodal tasks in hydrology. WaterGPT is applied three main areas: (1) processing and analyzing data such as images text water resources, (2) supporting intelligent decision-making hydrological tasks, (3) enabling interdisciplinary information integration knowledge-based Q&A. The has achieved promising results. One core aspect of involves the meticulous segmentation training supervised fine-tuning phase, sourced from real-world annotated with high quality using both manual methods GPT-series annotations. These are carefully categorized into four types: knowledge-based, task-oriented, negative samples, multi-turn dialogues. Additionally, another key component development multi-agent framework called Water_Agent, which enables to intelligently invoke various tools solve field resources. handles data, including images, allowing deep understanding analysis environments. Based on this framework, over 90% success rate object detection waterbody extraction. For extraction task, Dice mIoU metrics, WaterGPT’s performance high-resolution 2013 2022 remained stable, accuracy exceeding 90%. Moreover, we have constructed high-quality resources evaluation dataset, EvalWater, covers 21 categories approximately 10,000 questions. Using highest date reaching 83.09%, about 17.83 points higher than GPT-4.

Language: Английский

Citations

0