THE NWMP tools uses the term "frequencies". They express the probalitity that a certain value will not be exceeded.
One standard fequeny is the 50%-frequency, also called median. Due to the skewness of most samples observed in hydrology (at least in rainfall and floodflow), the median is often lower than the average.
Other frequencies commonly used are the "dry year" and the "wet year". The "dry year" means in the NWMP tools a frequeny of 20%. This means, in 20% of all years (one out of five) this value will not be exceeded (it may be lower), in the remaining four out of five years it will be exceeded (the resource will be larger).
Consequently, the value for the "80% wet year" will only be reached or exceeded in one out of five years; in the other years the ressource will be less.
For sufficiently large samples, the DVS permits also to calculate the 10% and 90% probability.
For frequencies like dry year and wet year, usually so-called empirical frequency distributions are used for interpolation. For extreme values (like the one-in-hundred-years rainfall or the one-in-thousand-years flood) extrapolations are necessary, they require that first a mathematical function is selected and its parameters estimated. Typical excamples for that are the Gumbel and the Pearson-III distributions.
In both cases, the cumulative probability for each observed value of the available sample is calculated. First, all values are sorted ascendingly by size. Than, for each of these sorted (or ranked) values the cumulative probability is calculated as a function of the rank i and the total sample size n. For this, different formulas (so-called plotting formulas) exist. typically, the NWMP tools including the DVS uses the formula i/(n+1), while the tool bfff used the the formula (i-0.375)/(n+0.25). The latter formula may be more accurate, particularly in case of small samples.
The values for the defined frequencies like 0.2, 0.5 and 0.8 are then obtained using linear interpolation. In the illustration using an example of 13 observed mean monthly baseflow values, it becomes clear that the plotting formula of the bfff results in slightly higher values for the dry year (20%) and in considerably lower values for the wet year (80%), when the curves are less steep.