Ensuring a good ecological status of water bodies is one of the key challenges of communities and one of the objectives of the European Water Framework Directive. Although recent works identified the most significant stressors affecting the ecological quality of rivers, the ability to predict the overall ecological status of rivers based on a limited amount of easily accessible geospatial data has not been investigated so far. Most of the analyses focus on detailed local modelling and measurements which cannot be systematically applied at regional scales for the purposes of water resources management. The aim of this work is to understand the capabilities of five supervised machine learning classifiers of predicting the ecological status of rivers based on land use, climate, morphology, and water management parameters extracted over the river catchments corresponding to the ecological monitoring stations. Moreover, the performances of machine learning classifiers are compared to the results of the canonical correlation analysis. The method is applied to 360 catchments in Tuscany (central Italy) with a median size of 33.6 km2 and a Mediterranean climate. The results show (i) a significant correlation of ecological status with summer climate (i.e., maximum temperatures and minimum precipitation), land use and water exploitation, (ii) an 80 % precision of Random Forest algorithm to predict ecological status and (iii) higher capability of all classifiers to predict at least good ecological status. In perspective, such predictive capabilities can support decision making in the land and water resources management and highlight strategies for river eco-hydrological conservation.

Prediction of ecological status of surface water bodies with supervised machine learning classifiers / Arrighi, Chiara; Castelli, Fabio. - In: SCIENCE OF THE TOTAL ENVIRONMENT. - ISSN 0048-9697. - ELETTRONICO. - (2023), pp. 0-0. [10.1016/j.scitotenv.2022.159655]

Prediction of ecological status of surface water bodies with supervised machine learning classifiers

Arrighi, Chiara
;
Castelli, Fabio
2023

Abstract

Ensuring a good ecological status of water bodies is one of the key challenges of communities and one of the objectives of the European Water Framework Directive. Although recent works identified the most significant stressors affecting the ecological quality of rivers, the ability to predict the overall ecological status of rivers based on a limited amount of easily accessible geospatial data has not been investigated so far. Most of the analyses focus on detailed local modelling and measurements which cannot be systematically applied at regional scales for the purposes of water resources management. The aim of this work is to understand the capabilities of five supervised machine learning classifiers of predicting the ecological status of rivers based on land use, climate, morphology, and water management parameters extracted over the river catchments corresponding to the ecological monitoring stations. Moreover, the performances of machine learning classifiers are compared to the results of the canonical correlation analysis. The method is applied to 360 catchments in Tuscany (central Italy) with a median size of 33.6 km2 and a Mediterranean climate. The results show (i) a significant correlation of ecological status with summer climate (i.e., maximum temperatures and minimum precipitation), land use and water exploitation, (ii) an 80 % precision of Random Forest algorithm to predict ecological status and (iii) higher capability of all classifiers to predict at least good ecological status. In perspective, such predictive capabilities can support decision making in the land and water resources management and highlight strategies for river eco-hydrological conservation.
0
0
Goal 6: Clean water and sanitation
Goal 11: Sustainable cities and communities
Arrighi, Chiara; Castelli, Fabio
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2158/1286732
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact