Hrvatske vode, Vol. 31 No. 125, 2023.
Izvorni znanstveni članak
Model of chlorophyll-a for the Butoniga reservoir in Istria
Goran Volf
; Sveučilište u Rijeci, Građevinski fakultet Radmile Matejčić 3, Rijeka, Hrvatska
*
Petar Žutinić
; Sveučilište u Zagrebu, Prirodoslovno-matematički fakultet, Horvatovac 102a, Zagreb, Hrvatska
Marija Gligora Udovič
; Sveučilište u Zagrebu, Prirodoslovno-matematički fakultet, Horvatovac 102a, Zagreb, Hrvatska
Antonija Kulaš
; Sveučilište u Zagrebu, Prirodoslovno-matematički fakultet, Horvatovac 102a, Zagreb, Hrvatska
Perica Mustafić
; Sveučilište u Zagrebu, Prirodoslovno-matematički fakultet, Horvatovac 102a, Zagreb, Hrvatska
* Autor za dopisivanje.
Sažetak
The Butoniga reservoir is relatively small and shallow, and, as such, very sensitive to degradation and eutrophication processes caused both by climate change and by anthropogenic activities in its basin. Chlorophyll-a is a widely used ecological indicator of primary production, i.e. phytoplankton biomass and the process of eutrophication in reservoirs, depending on the availability of nutrients, pH value, light, temperature, as well as their interrelations. The machine learning technique for creating models in the form of regression trees and rules was applied to a set of measurement data in order to better understand the eutrophication process and the ecological state of the Butoniga reservoir. Using machine learning tools, two models were created: (1) a descriptive model of chlorophyll-a in the form of regression trees, which describes changes in chlorophyll-a concentrations and determines the most significant factors causing these changes; and (2) a simulation model of chlorophyll-a in the form of rules that serves to predict the chlorophyll-a concentration in relation to the observed values of the measured environmental factors. The created models contribute to a better understanding of the ecosystem of the Butoniga reservoir, and can be used for the reservoir management as well.
Ključne riječi
Butoniga reservoir, chlorophyll-a, machine learning, models, phytoplankton
Hrčak ID:
317910
URI
Datum izdavanja:
28.9.2023.
Posjeta: 73 *