In most previous works dealing with the modelling and numerical simulation of salinity gradient solar ponds, the heat diffusion is 1D or 2D, the diffusion of salt is very slow and that there are small convection movements, which obviously supposes that the wind speed is negligible. The studies are often limited to sodium chloride-based ponds where only relatively clear water is considered. The present work comes to fulfil the lack of 3D thermal behaviour prediction where, in one hand, we simulated solar ponds by a mixture of three different techniques, which are stable and well adapted to this type of problem using the method of finite differences in conjunction with the schemes of ADI-Gunn, Adams-Bashford and Schneider and al, which finds its first application in this type of problems. On the other hand, in order to validate our numerical results, we conducted an experiment on three identical parallelepiped ponds with {2.3040}$\,\text{m}^2$ of area and {1.200}$\,\text{m}$ depth; each one containing the same fungicide salt namely \ce{KH2PO4}, that has never been used before.