Forecast Verification of the Polar Ice Prediction System (PIPS) Sea Ice Concentration Fields

Abstract

The National Ice Center relies upon a coupled ice?ocean model called the Polar Ice Prediction System (PIPS) to provide guidance for its 24?120-h sea ice forecasts. Here forecast skill assessments of the sea ice concentration (C) fields from PIPS for the period 1 May 2000?31 May 2002 are presented. Methods of measuring the sea ice forecast skill are adapted from the meteorological literature and applied to locations where the forecast or analysis sea ice fields changed by at least ±5%. The forecast skill referenced to climatology was high (>0.85, relative to a maximum score of 1.0) for all months examined. This is because interannual variability in the climatology, which is used as a reference field, is much greater than the day-to-day variability in the forecast field. The PIPS forecasts were also evaluated against persistence and combined climatological?persistence forecasts. Compared to persistence, the 24-h forecast was found to be skillful (>0.2) for all months studied except during the freeze-up months of December 2000 and January 2001. Relative to the combined reference field, the 24-h forecast was also positive for the non-freeze-up months; however, the skill scores were lower (?0.1). During the poorly performing freeze-up months, a linear combination of persistence (?95% weight) and climatology (?5% weight) appears to provide the best available sea ice forecast. To examine the less restrictive question of whether PIPS can forecast sea ice concentration changes, independent of the magnitude of the changes, ?threat indexes? patterned after methods developed for tornado forecasting were established. Two specific questions were addressed with this technique. The first question is: What is the skill of forecasting locations at which a decrease in sea ice concentration has occurred? The second question is: Does PIPS correctly forecast melt-out regions? Using the more relaxed criterion of a threat index for defining correct forecasts, it was found that PIPS correctly made 24-h forecasts of decreasing sea ice concentration ?10%?15% of the time (it also correctly forecast increasing sea ice concentration an additional ?10%?15% of the time). However, PIPS correctly forecast melt-out conditions <5% of the time, suggesting that there may be deficiencies in the PIPS parameterization of marginal ice zone processes and/or uncertainties in the atmospheric?oceanic fields that force PIPS.