Benefits of machine learning and sampling frequency on phytoplankton bloom forecasts in coastal areas

Abstract

In aquatic ecosystems, anthropogenic activities disrupt nutrient fluxes, thereby promoting harmful algal blooms that could directly impact economies and human health. Within this framework, the forecasting of the proxy of chlorophyll a in coastal areas is the first step to managing these algal blooms. The primary goal was to analyze how phytoplankton bloom forecasts are impacted by different sampling frequencies, by using a machine learning model. The database used in this study was sourced from an automated system located in the English Channel. This device has a sampling frequency of 20 minutes. We considered 12 physicochemical parameters over a six-year period. Our forecast methodology is based on the random forest (RF) model and a sliding window strategy. The lag times for these sliding windows ranged from 12 hours to 3 months with four different sampling times until 1 day.The results indicate that the optimal forecast was obtained for a 20 minutes time step, with an average R² of 0.62. Moreover, the highest values of fluorescence were predicted when the water temperature was approximately 11.8°C. Consequently, we demonstrated that the sampling frequency directly impacts the forecast performance of an RF model. Furthermore, this kind of model can recreate interactions that closely resemble biological processes. Our study suggests that the RF model can utilize the additional information contained in high-frequency datasets. The methodology presented here lays the foundation for the development of a numerical decision-making tool that could help mitigate the impact of these algal blooms.

The dependence of forecasts on sampling frequency as a guide to optimizing monitoring in community ecology

“…Sampling frequency has widely been recognized as having significant impacts on various analyses (e.g., Lehtiniemi et al, 2022;Ma et al, 2022), and yet sampling frequencies are commonly too low (Estes et al, 2018). The implications for ecological forecasting are not well known, as we are aware of few and contrasting findings regarding the effects of sampling frequency on forecasting (Derot et al, 2020;Wauchope et al, 2019), and which potentially were influenced by sample sizes. We found that lowering the sampling frequency worsened the abundance forecasts for 11 out of the 12 targets.…”

Machine learning-based estimation of riverine nutrient concentrations and associated uncertainties caused by sampling frequencies

“…Waterbodies must maintain a good chemical and ecological status to protect human health and safeguard natural ecosystems. Nutrients are important indicators that affect water quality, watershed health, and biological processes [ 1 , 2 ]. As key constituents of riverine nutrients, high concentrations of nitrogen (N) and phosphorus (P) may lead to eutrophication and anoxia in coastal waters [ 3 ], thereby not only affecting the living environment of human beings but also the biodiversity [ 4 ].…”

“…Not just hyperparameter and model structure, data input associated with different sampling frequencies might also induce uncertainties and influence estimation accuracy [ 37 ]. Derot et al [ 2 ] demonstrated that the different sampling frequency datasets directly impact the forecast performance of an RF model. According to their findings, the accuracy of phytoplankton bloom forecasts for a 20-min time step was higher than that of the 1-day time step.…”

Machine Learning Methods Applied to the Prediction of Pseudo-nitzschia spp. Blooms in the Galician Rias Baixas (NW Spain)

“…In case of binary classification (bloom/no bloom), the final result is the most "popular" class, i.e., the output for the majority of trees [41]. RF has been extensively used in ecological modelling [42][43][44], and during the last years, it has been successfully employed to HAB prediction [15,26,28,[45][46][47][48].…”