Derive Temperature Spikes

[joesimeone]

One comment will be for one of the definitions in the emailed table. In general, we want to convert to farenheit, then use weather-metrics to derive heat index. Using daily heat index measures, we can spin up the requested definitions.

Qs

2. Off mean or max / min ?? I guess I'll just do both, but worth thinking through what's better given grant.

These are more or less finished now, but will run some additional checks just to make sure everything is working right. There are some outstanding questions that need to be sorted. Since the grant entries were in Fahrenheit, I produced the metrics in that scale. See entries below

[joesimeone]

1. Anomaly: Deviations from 30-year summer/winter average temperature

Nothing | Should be good

[joesimeone]

2. Temperature shocks: Counts of days in a year that exceeded critical temperature thresholds

Nutshell

Although I calculated these from both average temperatures and minimum / maximum temperatures, I think it makes more sense to use the minimums and maximums. (Kind of like an acute vs prolonged exposure). However, if you use minimum and maximums, you can't use Heat Index for cold temperature spikes.

1. Cold Shock Issue

So, using heat index for cold spikes is problematic. When dew-point temperatures exceed dry-bulb temperatures, weather-metrics throws a warning and creates NAs for that Heat Index value. This is an issue with minimums specifically because, often times the daily average dry-bulb temperature will be larger. If we want to define cold spikes using minimum temperatures on any given day, we'll need to use the minimum dry bulb temp rather than heat index value.

Warning & console output

2. Heat Shock Issue

Sort of similar as above: On days when dry-bulb average temperature is the same as dewpoint - Missings get throw in about .5% of cases. With this definition, I just excluded them because there aren't a ton, but there may be a better systemic solution to the problem.

Warning & console output

Nutshell

[joesimeone]

3. Heat shock: Days across temperature distribution categories

Just hit prism variables with dplyr's percent_rank function. Do we have specific cut points in mind?