...
| Panel | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Additional plots for further analysis:
|
...
| borderColor | red |
|---|---|
| borderStyle | solid |
| title | Notes from Frederic 22/5/2018 |
...
If the dot is not in the right location, change it and replot.
Probabilities
Using the plotted probability map for 10mm precipitation threshold, use the cursor data icon to read the probability at the chosen location for +96 hours. Make a note of this value.
Edit prob_tp_compare.mv, and change the threshold value to 20mm:
| Code Block |
|---|
prob=20 |
Replot the map and make a note of the probability at your chosen location.
Finally change the threshold probability to 30mm and replot:
| Code Block |
|---|
prob=30 |
At your chosen location, using the cursor data icon, make a note of the probability for the 30mm threshold values.
You should now have the probability values that total precipitation will exceed 10mm, 20mm and 30mm, for both the 2012 and 2016 ensembles, for forecast time +96 hours.
Task 2: Plot the CDF
This exercise uses the cdf.mv icon.
...
| Code Block | ||
|---|---|---|
| ||
param="tp" station=[44.0,4.0] # !use your own values! expID="ens_oper" |
...
Do the same for the 2016 operational ensemble reforecast:
| Code Block |
|---|
expID="ens_2016" |
...
expID="ens_2016" |
Compare the CDF from the different forecast ensembles.
| Panel | ||
|---|---|---|
| ||
Q. What can you say about the spread? Q. Why does the CDF not look like Figure 2 above? |
Compare with probability map values
Using the CDF graph for the 2012 ensemble, read the probability that total precipitation will exceed 10mm. For example, see what percentile value, p, is indicated on the y-axis for x=10mm. The probability that total precipitation exceeds this value is then 100-p.
The value read from the CDF graph in this way should agree with the value you obtained by reading the probability value from the map in Task 1.
Check your probabilities for 20mm and 30mm total precipitation.
| Panel | |||
|---|---|---|---|
| |||
Q. What can you say about the spread? Q. Why does the CDF not look like Figure 2 above? |
Compare with probability map values
Using the CDF graph for the 2012 ensemble, read the probability that total precipitation will exceed 10mm. For example, see what percentile value, p, is indicated on the y-axis for x=10mm. The probability that total precipitation exceeds this value is then 100-p.
The value read from the CDF graph in this way should agree with the value you obtained by reading the probability value from the map in Task 1.
Check your probabilities for 20mm and 30mm total precipitation.
| Panel | ||||
|---|---|---|---|---|
| ||||
Q. Do your probabilities read from the 2012 and 2016 maps of total precipitation in Task 1, agree with values from the CDF curves? |
| Panel | ||
|---|---|---|
| ||
Q. Using these two macros, compare the 2012 and 2016 forecast ensemble. Which was the better forecast for HyMEX flight planning? |
Task 3. Plot percentiles of total precipitation
To further compare the 2012 and 2016 ensemble forecasts, plots showing the percentile amount and probabilities above a threshold can be made for total precipitation.
Use these icons:
Both these macros will use the 6-hourly total precipitation for forecast steps at 90, 96 and 102 hours, plotted over France.
Edit the percentile_tp_compare.mv icon.
Set the percentile for the total precipitation to 75%:
| Code Block | ||
|---|---|---|
| ||
#The percentile of ENS precipitation forecast
perc=75 |
...
| |
Q. Do your probabilities read from the 2012 and 2016 maps of total precipitation in Task 1, agree with values from the CDF curves? |
The values may not match exactly as the number of samples (ensembles forecasts in this case) is limited.
Task 3. Plot percentiles of total precipitation
To further compare the 2012 and 2016 ensemble forecasts, plots showing the percentile amount above a threshold can be made for total precipitation.
These can also be compared to the CDF curves from Task 2.
As before, this will use the 6-hourly total precipitation for forecast steps at 90, 96 and 102 hours, plotted over France.
Edit the percentile_tp_compare.mv icon.
Set the percentile for the total precipitation to 70% and specify the location as in Task 1 & 2:
| Code Block | ||
|---|---|---|
| ||
#The percentile of ENS precipitation forecast
perc=70
location=[44.0,4.1] # [ lat, lon ] -- use your own values! |
Plot the map. It is very similar to the probability map but now shows precipitation values (in mm) for the specified percentile.
From the CDF graph, read the percentile value of 70% on the y-axis and find the total precipitation value indicated on the x-axis.
Use the cursor data icon on the map, as before, and confirm the CDF value agrees with the value at the location on the map (shown by the purple dot).
Repeat this by setting the percentile to 80% and 95%
| Panel | ||||
|---|---|---|---|---|
| ||||
Q. From the CDF and probabilities maps, which ensemble forecast shows increased probability of precipitation higher than 10mm? |
Exercise 5: Cluster analysis
...