Given that the input concentration varies periodically, reaching a maximum at 5:00 PM and a minimum at 5:00 AM, a good first approximation for the input would be the sinusoidal input function shown below. (What would the period be in this case?)
We might also consider further simplifying the model by representing the input as a step function if the difference between the maximum and minimum is relatively small or the lake is large enough, but a sinusoidal model is probably the best initial guess.