EphèseLe catalogue de la bibliothèque de l'ECAM

Enhancing Process-Based Environmental Models with Machine Learning in Data-Limited Scenarios / Nicolas SAMELSON 

Public ISBD Titre : Enhancing Process-Based Environmental Models with Machine Learning in Data-Limited Scenarios Type de document : Travail de fin d'études Auteurs : Nicolas SAMELSON, Auteur ; Joerg WICKER, ; Quentin DELHAYE, Editeur : ECAM Année de publication : 2024 Langues : Anglais (eng) Mots-clés : Intelligence artificielle Index. décimale : TFE - Informatique (ECAM) Résumé : Environmental modelling in New Zealand often faces challenges due to fragmented, location-specific projects that lack standardisation and adaptability to climate changes. Resource constraints limit the use of best practices like ensemble modelling, confining decisions to single models, and the exclusion of community input further complicates participatory processes. This thesis explores an innovative approach to embed empirical equations into a Graph Auto-Encoder, aiming to discover transferable and generalisable patterns across models. The proposed method is then applied on a real-world scenario of freshwater quantity modelling, specifically the water balance in soils, where data availability is limited. Through this research, we seek to enhance the accuracy and generalisation of environmental models, opening new avenues for improved environmental management and decision-making. Enhancing Process-Based Environmental Models with Machine Learning in Data-Limited Scenarios [Travail de fin d'études] / Nicolas SAMELSON, Auteur ; Joerg WICKER, ; Quentin DELHAYE,  . - ECAM, 2024. Langues : Anglais (eng) Mots-clés : Intelligence artificielle Index. décimale : TFE - Informatique (ECAM) Résumé : Environmental modelling in New Zealand often faces challenges due to fragmented, location-specific projects that lack standardisation and adaptability to climate changes. Resource constraints limit the use of best practices like ensemble modelling, confining decisions to single models, and the exclusion of community input further complicates participatory processes. This thesis explores an innovative approach to embed empirical equations into a Graph Auto-Encoder, aiming to discover transferable and generalisable patterns across models. The proposed method is then applied on a real-world scenario of freshwater quantity modelling, specifically the water balance in soils, where data availability is limited. Through this research, we seek to enhance the accuracy and generalisation of environmental models, opening new avenues for improved environmental management and decision-making.

Exemplaires

Code-barres	Cote	Support	Localisation	Section	Disponibilité
aucun exemplaire

Documents numériques

Ce document n'est visible qu'après identification SAMELSON_Nicolas_-_5MIN.pdf Adobe Acrobat PDF

---

1 (1 - 1 / 1) Par page : 25 50 100 200