Solutions for Accommodating Temporally Limited Subseasonal-to-Seasonal Data: Using Harmonics to Generate a Climatology

As part of the development of a new database of fire weather variables, a period normal is required to convert the data into an anomaly space. The most common method for generating anomaly space data is to take a simple 30-year mean of the data’s time series at each grid point and then subtract that result from the original time series at each grid point. However, this data set, like many other subseasonal-to-seasonal data sets, is temporally limited to a time period of fewer than 10 years. As such, the typical climatology method cannot be used. Removing a simple time mean over the entire time period available is not suitable, as the temporal mean within this time period may not be fully representative of the actual background state that we aim to remove. To better capture the actual background state, we implement a harmonics method, standardized in our publicly assessable code package for broader use. Here, we present the results of applying the harmonics method to the new fire weather database, namely how and why it can be used to more effectively remove the background state. We also show how the coefficients and frequency terms that make up each harmonic have scientific value. Not only does the harmonics method provide a ‘climatology’, or ‘period normal’ for an anomaly forecast space, it also provides useful diagnostics for analysis and comparison of model patterns and oscillations. This harmonic approach may reduce future data generation and storage needs, simplify post-processing efforts, and expand the usability of specialized forecast variables.