Daily data for visualisation
Hi @vagudets,
Expected Behavior
In a Sunset workflow, I want to use the visualization module to plot monthly data in a seasonal horizon. In my recipe, I download daily data and then convert it to monthly data using an indicator.
Current Behavior
When I run the visualization module, I get an error saying that visualization only works for monthly data, because the function reads the frequency from the recipe instead of the actual dataset.
Example:
> Visualization(recipe = recipe, data = HI_cal,
+ skill_metrics = HI_metrics,
+ probabilities = HI_cal$probs,
+ significance = TRUE)
ERROR [2025-12-01 11:34:14] Visualization functions not yet implemented
for daily data.
Error in plot_metrics(recipe = recipe, data_cube = data$hcst, metrics = skill_metrics, :
Possible Solutions?
Currently, I have to modify the recipe to use monthly data before running the visualization, which works. Is there a way to bypass this check? For example, in sunset/modules/Visualization/R/plot_most_likely_terciles_map.R, the check is:
if (recipe$Analysis$Variables$freq %in% c("daily", "daily_mean")) {
stop("Visualization functions not yet implemented for daily data.")
}
Could this check be modified to look directly at the dataset rather than the recipe?
Steps To Reproduce
- Script:
recipe_file <- "/esarchive/scratch/abojaly/GitLab/sunset/dev_Indicator_HI/recipe_HI.yml"
recipe <- prepare_outputs(recipe_file)
# Load datasets
data_raw <- Loading(recipe)
# Change units
data_units <- Units(recipe, data_raw)
# Indicator
HI <- Indicators(recipe = recipe, data = data_units) #List of all indicators
# Calibration
HI_cal <- Crossval_calibration(recipe, HI)
# Skill Metrics
HI_metrics <- Crossval_metrics(recipe, HI_cal)
# Plot data
recipe$Analysis$Variables$freq <- 'monthly_mean'
Visualization(recipe = recipe, data = HI_cal,
skill_metrics = HI_metrics,
probabilities = HI_cal$probs,
significance = TRUE)
-
Branch/SUNSET Version: dev_HeatIndex
-
Environment: hub06