Advancing Large Language Models for Spatiotemporal and Semantic Association Mining of Similar Environmental Events
Authors:
Yuanyuan Tian,
Wenwen Li,
Lei Hu,
Xiao Chen,
Michael Brook,
Michael Brubaker,
Fan Zhang,
Anna K. Liljedahl
Abstract:
Retrieval and recommendation are two essential tasks in modern search tools. This paper introduces a novel retrieval-reranking framework leveraging Large Language Models (LLMs) to enhance the spatiotemporal and semantic associated mining and recommendation of relevant unusual climate and environmental events described in news articles and web posts. This framework uses advanced natural language pr…
▽ More
Retrieval and recommendation are two essential tasks in modern search tools. This paper introduces a novel retrieval-reranking framework leveraging Large Language Models (LLMs) to enhance the spatiotemporal and semantic associated mining and recommendation of relevant unusual climate and environmental events described in news articles and web posts. This framework uses advanced natural language processing techniques to address the limitations of traditional manual curation methods in terms of high labor cost and lack of scalability. Specifically, we explore an optimized solution to employ cutting-edge embedding models for semantically analyzing spatiotemporal events (news) and propose a Geo-Time Re-ranking (GT-R) strategy that integrates multi-faceted criteria including spatial proximity, temporal association, semantic similarity, and category-instructed similarity to rank and identify similar spatiotemporal events. We apply the proposed framework to a dataset of four thousand Local Environmental Observer (LEO) Network events, achieving top performance in recommending similar events among multiple cutting-edge dense retrieval models. The search and recommendation pipeline can be applied to a wide range of similar data search tasks dealing with geospatial and temporal data. We hope that by linking relevant events, we can better aid the general public to gain an enhanced understanding of climate change and its impact on different communities.
△ Less
Submitted 19 November, 2024;
originally announced November 2024.
Detection of Critical Events in Renewable Energy Production Time Series
Authors:
Laurens P. Stoop,
Erik Duijm,
Ad J. Feelders,
Machteld van den Broek
Abstract:
The introduction of more renewable energy sources into the energy system increases the variability and weather dependence of electricity generation. Power system simulations are used to assess the adequacy and reliability of the electricity grid over decades, but often become computational intractable for such long simulation periods with high technical detail. To alleviate this computational burd…
▽ More
The introduction of more renewable energy sources into the energy system increases the variability and weather dependence of electricity generation. Power system simulations are used to assess the adequacy and reliability of the electricity grid over decades, but often become computational intractable for such long simulation periods with high technical detail. To alleviate this computational burden, we investigate the use of outlier detection algorithms to find periods of extreme renewable energy generation which enables detailed modelling of the performance of power systems under these circumstances. Specifically, we apply the Maximum Divergent Intervals (MDI) algorithm to power generation time series that have been derived from ERA5 historical climate reanalysis covering the period from 1950 through 2019. By applying the MDI algorithm on these time series, we identified intervals of extreme low and high energy production. To determine the outlierness of an interval different divergence measures can be used. Where the cross-entropy measure results in shorter and strongly peaking outliers, the unbiased Kullback-Leibler divergence tends to detect longer and more persistent intervals. These intervals are regarded as potential risks for the electricity grid by domain experts, showcasing the capability of the MDI algorithm to detect critical events in these time series. For the historical period analysed, we found no trend in outlier intensity, or shift and lengthening of the outliers that could be attributed to climate change. By applying MDI on climate model output, power system modellers can investigate the adequacy and possible changes of risk for the current and future electricity grid under a wider range of scenarios.
△ Less
Submitted 1 February, 2024; v1 submitted 31 January, 2024;
originally announced January 2024.
