The map above is the forecast of the Spring Index anomaly (ΔSI-x) in units of days. Their negative values mean early onset for a climatology defined for the period 1982-2010, and late onset for positive values. The black solid lines indicate region where the spring onset has already occurred. The actual forecasted day of the Spring Index (SI-x) is indicated in the Table below in calendar days or Day-Of-the-Year (DOY) units. The SI-x is computed as in Ault et al. (2015) with current climatic conditions. These data sets are multiple simulations of daily Tmax and Tmin of the CFSv2 model. The SI-x product was post-processing using an ensemble model statistic output (EMOS), Non-homogeneous Gaussian Regression (Gneiting et al., 2005). The different ensemble members are defined with four daily runs with different initializations: 00Z, 06Z, 12Z, and 18Z. Thus, a total of 120 model realizations are used in each issued SI-x forecast. The EMOS product is calibrated with a re-forecasting dataset from 1982-2010 and further details can be found in Carrillo et al. (2018). The products related to the SI-x forecast are shown in four field maps in the Table below for different initializations (Dec 15, Dec 31, Jan 15, etc): (1) the Spring Index anomaly; (2) the Spring Index; (3) a global skill that is defined as CDF of centered anomaly correlation for the entire domain (Wilks, 2011); two thresholds (0.5 and 0.7) identify skill for different initializations; (4) a local skill defined using Pearson correlation.
Toby R. Ault
toby.ault@cornell.edu
Dept. of Earth & Atmos. Sciences, Cornell University
∆SI-x | SI-x | |||||
Jan | 15^{th} | 31^{th} | 15^{th} | 31^{th} | Global_Skill | Local_Skill |
Dec | 15^{th} | 31^{th} | 15^{th} | 31^{th} | Global_Skill | Local_Skill |
Dec | time_series | Global_Skill | Local_Skill |
Jan | time_series | Global_Skill | Local_Skill |
Feb | time_series | Global_Skill | Local_Skill |
Mar | time_series | Global_Skill | Local_Skill |