Physical parameters-driven models for predicting shrimp growth in aquaculture

Abstract

As of today, more than half of the world’s shrimp production comes from farmed shrimp. Shrimp farming (or shrimp aquaculture) is most common in China and is also practiced in Thailand, Indonesia, India, Vietnam, Brazil, Ecuador, and Bangladesh. Growth models for shrimps in open pond aquaculture systems are being discussed in the paper. The Water quality of the pond and the nature of stocking are proposed as the proxy to determine the growth rates of the cultured organisms. The devised shrimp growth model, when tested on a publicly available dataset obtained from a conducted experiment, with recorded farm parameters like temperature, stocking density, pH and salinity, gave the mean error of 0.1 g and Root Mean Squared Error (RMSE) of 3.76g, in the final weights of the shrimp. This paper demonstrates the first such approach to bring reasonable accuracy growth models for the shrimp without needing any extra ground setup and using data from the various literature sources. Considering the approach to predict shrimp growth through physical parameters and also the feasibility of the framework to be implemented by deploying IoT sensors in the pond water, this study aims to bring the notion of digital transformation in the context of Indian aquaculture systems of open ponds shrimp farms.