| count2prob {easyVerification} | R Documentation |
Using plotting positions as described in Wilks (2011), counts of occurrences per forecast category are converted to probabilities of occurrence. For ensembles of size 1 (e.g. verifying observations), the count vector is returned unaltered (corresponding to occurrence probabilities of 0 or 1).
count2prob(x, type = 3)
x |
input matrix of counts from |
type |
selection of plotting positions (default to 3, see Types) |
Matrix of probabilities per category
The types characterize the plotting positions as specified in Wilks (2011). The plotting positions are computed using the following relationship:
p(x_i) = (i + 1 - a)/(n + 1 - a)
where i is the number of ensemble members not exceeding x, and n is the number of ensemble members. The types are characterized as follows:
| type | description | a |
| 1 | Weibull | 0 |
| 2 | Bernard and Bos-Levenbach | 0.3 |
| 3 | Tukey | 1/3 |
| 4 | Gumbel | 1 |
| 5 | Hazen | 1/2 |
| 6 | Cunnane | 2/5 |
Wilks, D.S. (2011). Statistical methods in the atmospheric sciences (Third Edition). Academic press.
convert2prob for conversion of continuous forecasts to ensemble counts
tm <- toymodel() ## convert to tercile forecasts (only display first forecast and obs) count2prob(convert2prob(tm$fcst, prob=1:2/3))[1,] count2prob(convert2prob(tm$obs, prob=1:2/3))[1,]