cold water reinjection - temperature rise under the well

I built a 3D heat transport model (transient flow/transient transport) to simulate the cold water plume due to reinjection into a deep aquifer. The reinjection well was realized by 9 nodes (well bore conditions), injection was set to the lowest node (time series). To each well node a Temperature BC (time series) was assigned.
After simulation (at first 60 days) can be seen that the environment of the well and Temperature BC nodes, respectively cools down, but the uppermost layer under the well warms up. Temperature rise in this layer amounts to about 10 °C relating to the initial temperature.
The layers (22 layers altogether in the model) in and under the aquifer have a thickness of 30 m. Element size in the vicinity of the well 0.15 m.
What is the reason for the temperature rise under the well? How can this obviously numerical error be eliminated?
Thanks for suggestions in advance.

The reason for the temperature rise is numerical oscillation due to a steep temperature gradient combined with a relatively large element size of 30 m. The ideal solution is a much finer vertical discretization below the injection well.

Thank you Peter for your answer to my question und your suggestion for solving the problem.