Extreme indices are quantities in climate change studies aimed at tracking the occurrence and frequency of extreme events over time, such as droughts, heatwaves and cold spells.

Climate Explorer calculates indices related to temperature maxima, temperature minima and precipitation. Many of these are from the core set defined by the WMO's Expert Team on Sector-Specific Climate Indices (ET-SCI, see https://www.climdex.org/learn/indices/), however some are defined independently.

Some indices simply involve counting, e.g. the number of days where the minium temperature is below 0°C. These results are simple and robust to compute. Other indices depend on calculating percentiles (e.g. the fraction of days where precipitation is above the 95th percentile of wet days in a defined base period of years). These have to be calculated more carefully.

For indices including percentiles, Climate Explorer adopts methodology described in Zhang et al. (2005). A percentile is calculated for each day of the year. We define a "base period" between 1961 and 1990 inclusive. For each year in the base period, we gather data for all the other years, plus an extra year (either the first or the last in the base period). We then take data from each of these years for a 5-day window around the day being calculated. If there is sufficient valid (non-fill value) data in this set, percentiles are calculated by sorting the collected data, identifying the two closest points to the desired percentile and linearly interpolating between them. For years outside the base period, data from the whole base period is used.

The results of the Climate Explorer algorithm have been compared to those from Climpact. The values are mostly either identical or agree to within a small percentage difference. However, some small differences exist due to subtle differences in the calculation of quantiles, although both follow the Zhang et al. (2005) method. If desired, time series outputs from Climate Explorer can be used as inputs for Climpact.