Temperature

BCRMSE

    • PS pair-wise differences occur during the fall and winter seasons, typically in the overnight hours; all favor the REF configuration.

Bias

    • Regardless of initialization time, a number of PS pair-wise differences favoring the REF configuration are noted during the overnight hours, most notably for the annual, fall and winter aggregations when comparing REF-TWIND1 and for all temporal aggregations for the REF-TWIND2 comparison.

Dew Point Temperature

BCRMSE

    • For both initializations, a majority of SS pair-wise differences favor TWIND1/TWIND2; however, SS differences favoring REF are seen at longer forecast lead times for the 00 UTC initializations mostly during the winter and fall aggregations. No differences are PS.

Bias

    • A majority of the SS pair-wise differences favor the TWIND1/TWIND2 configuration, especially during the overnight hours. PS pair-wise differences are predominantly noted in the fall and winter aggregations and are generally seen at the longer forecast lead times, with no distinct signal in which configuration is favored.

Wind

BCRMSE

    • For REF-TWIND1, a majority of forecast lead times are SS and favor the REF configuration for all temporal aggregations except summer, which exhibits the fewest SS pair-wise differences. No PS differences are noted.
    • For REF-TWIND2, several SS pair-wise differences are noted, with TWIND2 favored more frequently for the summer aggregation and REF for the other temporal aggregations. None are PS.

Bias

    • For both initializations and all temporal aggregations, all forecast lead times are SS with a majority of the differences being PS. TWIND1/TWIND2 is favored for overnight hours (i.e., 03 - 15 UTC) and REF is favored for day time hours (i.e., 18 - 00 UTC). TWIND1/TWIND2 has more PS pair-wise differences in the fall and winter aggregations than during the other temporal aggregations.

3-hour and Daily Precipitation Accumulation

GSS

    • Only a few SS pair-wise differences are noted between the REF-TWIND1 and REF-TWIND2 inter-comparisons when considering 3-hourly and daily QPF.

Frequency Bias

    • No SS pair-wise differences are observed for the 3-hourly and daily QPF.

All variables

All verification metrics

    • Inter-comparisons of the REF-TWIND1 and REF-TWIND2 configurations, regardless of forecast lead time or temporal aggregation, show very few SS pair-wise differences in either bias or BCRMSE for all upper air variables. When SS pair-wise differences are noted, tehy are mainly located inthe lowest vertical levels.