We now have the ability to extract dates from monthly time-series to use in creating composites. The steps are

1. Select a time series from those provided.
2. Select a month or a season.
3. Select a criteria. These are
   • Number: The code will return the number of events that are the highest (lowest) of the time range it looks at.
   • Percentile. The code counts the valid year year range and then calculates percentile cutoffs based on the number of values. So for 24 years, below the 25 percentile would be 6 values and above the 75th would be 6 as well (>18).
   • Value. All values great/less than the input values are used. You may need to look at the time-series for this.
   • Standardized Value (standardized over year range you select). Enter a sigma value (e.g. >2 sigma) and the cutoffs are based on that.
4. Select greater than or equal, less than or equal, or equal.
5. Optionally enter a year range if you would like to limit the number of years looked at for a long time-series. For example, entering 1950-2025 when requesting the 5 highest Nino 3.4 values from the Nino 3.4 time-series that runs from 1870-2025 will only look at those later years.
6. Optionally use you own label for the part of the plot with years. So if the label might be "Jan 2016,1983,1998,1992,1973" by default you can change it to
   "Jan: The 5 Strongest ENSO years from 1950-2024"

The code should only look in the part of the data set that exists. So if your data set covers 1950-2020 and your time-series is 1952-2024, then the time-series should only use the years 1952-2020 to determine the years that match the criteria.

Ties are ignored, so there is some randomness. Missing index time-series count as far as the number of values.

Years chosen and time-series values are available at the bottom of the page