With high and sustained precipitation rates, I was hoping out of this event to ground truth evaporative cooling effects.. particularly for my preferred habit of skating the XC trails at Falls. it's often marginal so knowing whether you can bank an extra hundred metres or three of freezing level makes a pretty big difference to intepreting a positive or negative outcome for the trails at 1500-1600m.
I kept a log of forecast model temps, not from a single run but twice a day grabbing the latest set of numbers for the most current period in the model output.
I've then plotted that against BoM observed temps. For Perisher, I've also plotted it against rainfall rate, but unfortunately Falls and Buller (and Hotham) were not reporting precipitation through this weekend.
It's hard to interpret much at the peak of the event midnight Sunday with the airmass changing so quickly.
But the period from 0200-1400 Saturday is more relevant for what I was aiming to do. If evaporative cooling was not captured in the models, I would expect to see observed temp dip during the heavier precipitation, relative to model temp. I cannot see evidence of that here. Observed temps also don't track lower relative to models at Falls or Buller as the snow arrived and intensified during Sunday.
Clearly evaporative cooling associated with short sharp convective showers cannot be reflected in hourly timestep model output temperatures. But my amateur speculation is that evaporative cooling is these days captured through parameterisation in the models for these large-scale (temporal and spatial) events, so I will unfortunately not be banking on the freezing level and accumulation being any lower than the modelled level. Interested in what others make of this?
The dataset I ended up with is crude and not what I was hoping for, particularly with no precip data for Falls/Buller, but I hope you may still find these plots interesting.
What is clear is that EC does an impressive job of modelling temp in our mountains. ACCESS-C also good but of limited value with narrow time horizon. ACCESS-G and GFS not in the same league, presumably primarily due to poorly captured terrain (as well as being inferior more generally).
EC model is from Jane's old site (effectively yr.no), from the mid-mountain forecast, with the following offsets to bring it line with BoM observation elevations:
- Falls Creek: -0.7 (1650m forecast to 1750m obs)
- Buller: -0.7 (1600m forecast to 1700m obs)
- Perisher: +0.4 (1800m forecast to 1750m obs)
AXS-C has the following arbitrary (but fixed) offsets to bring it into best fit with observations:
- Falls Creek: -2.2
- Buller: -2.5
- Perisher Valley: no offset
AXS-G has the following offsets calculated (for this specific time period) to bring it into best fit with the new Jane's weather forecast page:
- Falls Creek: -3.4
- Buller: -4.6
- Perisher Valley: -3.4
GFS has the following offsets also calculated to bring it into best fit with the new Jane's weather forecast page:
- Falls Creek: -3.5
- Buller: -4.6
- Perisher Valley: -3.5