Run-Off temperature in pkg/exf

[jm-c]

RunOff and internal energy associated with are just forcing that are external to the ocean, either provided as input files or from an other component (e.g., land in coupled set-up), but should not depend on internal ocean-model parameter.

The problem with current pkg/exf implementation is that the time-integrated internal energy associated with RunOff ( = Cp x RunOff x Temp_RO ) depends on the ocean-model time-step, because of time-interpolation, whereas the time-integrated fresh-water input from RunOff does not (at least when using linear time-interpolation, i.e., the default).

With linear time-interpolation, the issue can be fixed by interpolating in time the internal energy flux instead of the temperature (as currently done).
This will ensure that the RunOff contribution to the energy budget is entirely determined by the input data-set and will not change if I change my model time-step.

[mjlosch]

@jm-c is right, and surface heat should not depend on the time step. What would be the natural solution? Instead to prescribing runoff and runoftemp, one could prescribe runoff and heatflux=Cp * runoff*runoftemp, or did you think of doing the multiplication and (linear) interpolation online, which would require some change?