Introduction
Design and planning of pond aquaculture require a unified, quantitative pond model that can be linked also with the simplified quantitative models of neighboring areas of different land use. Considering the labor-consuming and expensive experimentation of individual production ponds, in this study we present the reusability check-based improvement of a formerly validated production pond model (Varga et al., 2020), applying the modeling framework of Programmable Process Structures (Varga and Csukas, 2022).
Materials and methods
The testing and improvements include both the reduced and the extended use of the reference model (i.e., appropriate reduction and extension of a selected reference model to describe the various pond managerial cases e.g., natural fishpond with no feeding and manuring, manured pond, manured and foraged pond with optional artificial fertilizer, etc.), and its application for the model-based scaling-up.
To check and improve the reference model, data generated during the vegetative seasons of 2021 and 2022, from carp-rearing experiments conducted in frequently monitored pilot ponds of MATE AKI HAKI, Szarvas, Hungary were used. This included data on stocking and harvesting of fish (mainly carp); on manuring, fertilizing, and feeding strategy; on manually measured and sensor-based water quality; on the concentration of food web elements (zooplankton, Chl-A represented phytoplankton); as well as about meteorology (Sharma et al., 2023). In order to pre-process this data, Matlab® Data Cleaner was utilized, and the moving median smoothing approach with a smoothing factor of 0.25 was also applied. The normalized root mean square error (NRMSE, %) was used to calculate the average deviation of simulated and calculated data in the absence of knowledge of the obvious sampling errors.
Results & Discussion
In the first phase of the suggested stepwise reusability check, each step ended with the necessary refinement and testing of the refined parameters or prototype models, while the already made improvements were fixed for the further steps. A simplified structure of the investigated model was created showing the contribution of the various nutrients (e.g., from feed, manure, and, inorganic fertilizer) to the food web of the produced carp. Finally, a hypothesis-based extension was suggested to distinguish phytoplankton’s eukaryotes and cyanobacteria groups. After iterative testing and verifying computations in multiple steps, a unified model was created for a wider range of ponds, from natural lakes to heavily stocked and manured ones. The input files, describing the structure, the input data, and parameters, as well as the prototyped local programs of the applied Programmable Process Structures, can be found in the contributing Mendeley database (Sharma et al., 2023). The obtained large NRMSE values contain both sampling & measurement and model errors. The reusability checking procedure brought attention to the even more conscious planning of experiments. In order to develop the strategy for sampling and measurements, which must comprise a small but essential number of spatially distributed and simultaneous measurements, the experimental design must involve the model specialists and some preparatory modeling. The inclusion of the pond history and the comprehensive set of initial concentrations are quite crucial since food web models in particular are very sensitive to the initial conditions.
The resulting model was also tested for model-based scaling-up of production ponds in the knowledge of the very limited amount of case-specific input data. In addition to the estimated approximate fish production, the up-scaled simulation can give estimations also about the other characteristics and environmental impacts of the production pond (e.g., for the necessary water supply, nutrient emission, O2 production and consumption, CO2 sequestration, and emission, etc.).
Acknowledgments
This research has been performed in the scope of the European Union’s Horizon 2020 research project EATFISH (grant no. 956697).
References
Sharma, P., Gyalog, G., Berzi-Nagy, L., Tóth, F., Nagy Z., Halasi-Kovács, B., Fazekas, DL., Mezőszentgyörgyi, D., Csukas, B. C & Varga, M. (2023). “Data for reusability check-based refinement of a biophysical fishpond model”, Mendeley Data, V1, doi: 10.17632/837f4mvpmb.1
Varga, M., Berzi-Nagy, L., Csukas, B., & Gyalog, G. (2020). Long-term dynamic simulation of environmental impacts on ecosystem-based pond aquaculture. Environmental Modelling & Software, 134, 104755. https://doi.org/10.1016/j.envsoft.2020.104755
Varga, M., Csukás, B. (2022). Foundations of Programmable Process Structures for the unified modeling and simulation of agricultural and aquacultural systems. Information Processes in Agriculture, in the press, https://doi.org/10.1016/j.inpa.2022.10.001