Slow cooling down after peaking

[minxuehe]

We used Livneh dataset (daily Tmax, Tmin, and precipitation) to generate hourly tempeature data at a location in California (40.03N, 122.16W). We chose tdf=0.8 (Tmax timing) as the default parameter value (0.67) yields earlier tempearature peaks. However, the recession is slower (slow cooling down after the peak) compared to the observed hourly tempeature (referring to the figure attached). Are there any other parameters that can be tuned to resovle the slow colling down issue?

[arbennett]

Currently there is not a way to control where Tmin occurs. This could be solved via implementing another parameter, maybe tmin_night_fraction for consistency. Then in metsim.disaggregate.set_min_max_hour it would be used as:

t_t_min = (-params['tmin_night_fraction'] * (set_times - rise_times)) + rise_times - ts

So a value of 0 would be the current behavior and then as tmin_night_fraction increases it moves the time at which Tmin occurs back.

Alternatively, different interpolators could be implemented. The current interpolator is a monotonic Hermite cubic spline. The key is the monotonicity, which guarantees the bounds of Tmin and Tmax are enforced. This ultimately just means that the derivative of the interpolator must be 0 at Tmin and Tmax.