North American Climate Extremes Monitoring

Methodology

A description of the methodology employed in the calculation of the monthly, seasonal, and annual index values are presented below.

Number of Frost Days, Summer Days, Icing Days, and Tropical Nights

The indices for the number of frost days, summer days, icing days, and tropical nights are computed using a standard reference temperature, which are the same for all stations used in the analysis.

• Number of Frost Days: days when the daily minimum temperature is less than 0°C (32°F).
• Number of Summer Days: days when the daily maximum temperature is greater than 25°C (77°F).
• Number of Icing Days: days when the daily maximum temperature is less than 0°C (32°F).
• Number of Tropical Nights: days when the daily minimum temperature is greater than 20°C (68°F).

A minimum amount of data is required in order to produce a meaningful index. If 70 percent of data are available, then the specific index can be calculated. If more than 30 percent of data is missing, the index is set to missing.

Much Below Average Maximums & Minimums and Much Above Average Maximums & Minimums

There is a broad range of climates across North America and conditions can vary greatly from location to location. For this reason a reference temperature for determining extreme weather conditions should differ from location to location. The computation of percentiles allows us to select a reference temperature unique for each station used in the analysis. The reference temperature for each station is determined by performing a distribution of observed frequencies of occurrences of all temperature values for a specific time period (either monthly, seasonal, or annual) and computing the top and bottom 10 percent of all occurrences.

1. Computation of Percentiles

The 10th percentile (bottom 10 percent of all occurrences) and 90th percentile (top 10 percent of all occurrences) for maximum and minimum temperatures are calculated for each station. These percentiles use all available data for a specified time period (month, season, or year). Available daily data from 1955 through present is sorted in ascending order. If 70 percent of data are available, then the computation continues; otherwise, the reference temperature and index values are set to missing. The reference temperature for the 10th percentile is computed by determining which temperature marks off the lowest 10 percent of the observations from the rest, while the temperature threshold for the 90th percentile is computed by determining which temperature exceeds all but the highest 10 percent of the values.

2. Computation of Days Above or Below Percentiles

After computing the reference temperature for the 10th and 90th percentile, for each station, and for maximum and minimum temperatures for a particular time period, the number of days above the 90th percentile value or below the 10th percentile value is totaled for a given month, season, or year.

3. Computation of the Percentage of Days Above or Below Percentiles

After computing the number of days above or below percentile for each station, that number is divided by the total number of days in that particular month/season/year. Multiplying this value by 100 provides the percentage of days above or below percentiles.

Computation of the 1961–90 Average for All Indices

A 30-year average is computed for each index at all stations where adequate data is available. The 30-year average is computed if the index has at least 21 years of data available. A straight average is computed with the data available from 1961–90. If more than 30 percent of data is not available, the 30-year average will be set to missing.

Computation of Anomalies with respect to the 1961–1990 Average

An anomaly is the deviation of a single value from the average. If a 30-year average temperature is computed for a station, an anomaly is computed. If a 30-year average temperature is not computed due to lack of data, anomalies are not computed.