Associations of anomalous water temperature, salinity, and pH with change in water color of fish farming ponds

Abstract

Abstract Maintaining proper green water is important for a fish farming pond. However, it remains unclear how to translate this important rule into specific water quality characteristics to associate or even forecast changes in water color. To address this issue, we conducted a study based on daily monitoring of six grouper ponds in Fangliao Township, Pingtung, Taiwan, from March to December 2018. We investigated the relationships between changes in water color and anomalies in water temperature, salinity, and pH, as these three parameters had the most complete records. We employed a long‐short‐term memory model to detect water quality anomalies by computing residual values. Our findings indicate that changes in water color were associated with anomalies in these parameters, with water temperature anomalies being the best indicator for early detection. In fact, the top 5% of water temperature anomalies could predict over 40% of the water color changes. Additionally, pH anomalies occurred immediately after the color changes. This rule‐to‐physicochemical‐parameter paradigm that we developed for grouper ponds could be applied to other aquaculture farms. We anticipate that with the help of advanced environmental surveillance models, such as the one we used, the prospect of autonomous fish farming will be realized.

Keywords

• anomaly detection
• fish farming ponds
• green water
• long short‐term memory
• water quality