Abstract:
This research is motivated by the imperative to develop robust computational methods for the
control of pollution in water and the enhancement of quality in water of the Burnett River, Aus tralia. Recognizing the intricate interplay of spatial and temporal dynamics in water quality, we
propose hybrid model, denoted as "CNN-LSTM." The amalgamation of (CNN) Convolutional
Neural Networks and (LSTM) Long Short-Term Memory networks addresses the unique chal lenges posed by this complex system. Dissolved Oxygen is identified as a pivotal parameter for
prediction, and meticulous feature engineering techniques are employed to refine its role within
the model.
The empirical results of our investigation unveil a notable enhancement in prediction perfor mance when employing the "CNN-LSTM" hybrid model in comparison to the AT-LSTM model.
This improvement underscores the efficiency of combining CNN for spatial data and LSTM for
temporal data, aligning with the inherent characteristics of water quality time series. The hybrid
model demonstrates its capability to capture the intricate relationships between various environ mental factors, leading to more accurate predictions.
Additionally, the study highlights the importance of spatial and temporal considerations when
predicting future impacts of dissolved oxygen in the Burnett River. The combined method
of CNN and LSTM not only uses the distribution of negative water but also collects the ex pected time to better understand the performance of the system. In addition to providing a high performance and effective method for predicting water in water bodies, this research contributes
to the expansion of environmental management by providing good practices in decision-making
and management of fixed assets.
