Prediction of monthly dry days with machine learning algorithms: a case study in Northern Bangladesh.

Dry days at varied scale are an important topic in climate discussions. Prolonged dry days define a dry period. Dry days with a specific rainfall threshold may visualize a climate scenario of a locality. The variation of monthly dry days from station to station could be correlated with several climatic factors. This study suggests a novel approach for predicting monthly dry days (MDD) of six target stations using different machine learning (ML) algorithms in Bangladesh. Several rainfall thresholds were used to prepare the datasets of monthly dry days (MDD) and monthly wet days (MWD). A group of ML algorithms, like Bagged Trees (BT), Exponential Gaussian Process Regression (EGPR), Matern Gaussian Process Regression (MGPR), Linear Support Vector Machine (LSVM), Fine Trees (FT) and Linear Regression (LR) were evaluated on building a competitive prediction model of MDD. In validation of the study, EGPR-based models were able to better capture the monthly dry days (MDD) over Bangladesh compared to those by MGPR, LSVM, BT, LR and FT-based models. When MDD were the predictors for all six target stations, EGPR produced highest mean R2 of 0.91 (min. 0.89 and max. 0.92) with a least mean RMSE of 2.14 (min. 1.78 and max. 2.69) compared to other models. An explicit evaluation of the ML algorithms using one-year lead time approach demonstrated that BT and EGPR were the most result-oriented algorithms (R2 = 0.78 for both models). However, having a least RMSE, EGPR was chosen as the best model in one year lead time. The dataset of monthly dry-wet days was the best predictor in the lead-time approach. In addition, sensitivity analysis demonstrated sensitivity of each station on the prediction of MDD of target stations. Monte Carlo simulation was introduced to assess the robustness of the developed models. EGPR model declared its robustness up to certain limit of randomness on the testing data. The output of this study can be referred to the agricultural sector to mitigate the impacts of dry spells on agriculture.

Integrating fuzzy logic with Pearson correlation to optimize contaminant detection in water distribution system with uncertainty analyses.

An effective detection algorithm, supervising an online water system, is expected to monitor changes in water quality due to any contamination. However, contemporary event detection methods are often criticized for their high false detection rates as well as for their low true detection rates. This study proposes two new event detection methods for contamination that use multi-objective optimization by investigating the correlation between multiple types of conventional water quality sensors. While the first method incorporates non-dominated sorting genetic algorithm II (NSGA-II) with the Pearson correlation Euclidean distance (PE) method in order to maximize the probability of detection (PD) and to minimize the false alarm rate (FAR), the second method introduces fuzzy logic in order to establish a degree of correlations ranking that replaces the correlation relationship indicator threshold. Optimization is performed by using NSGA-II in the second method. The results of this study show that the incorporation of fuzzy logic with NSGA-II in event detection method have produced better results in event detection. The results also show that both methods detect all true events without producing any false alarm rates. Moreover, an uncertainty analysis on input sensor signals is performed to test the robustness of the fuzzy logic-based event detection method by employing the widely used Monte Carlo simulation (MCS) technique. Four different scenarios of uncertainty are analyzed, in particular, and the findings suggest that the proposed method is very effective in minimizing false alarm rates and maximizing true events detection, and hence, it can be regarded as one of the novel approaches to demonstrate its application in the development of an event detection algorithm.