Verifications of CPS3 for one-month prediction using its Hindcast

Hindcast Verification

• Deterministic score Maps
• Variables to be Assessed: T2m, SST, RAIN
• Diagonostic Measures:
• Mean Square Skill Score(MSSS)

• Probabilistic score Diagrams
• Variables to be Assessed: T2m, SST, RAIN
• Diagonostic Measures:
• Reliability diagrams (Aggregated verification)
• Relative Operating Characteristics(ROC) curve (Aggregated verification)
• Event: Anomaly > 0, Below Normal, Near Normal, Above Normal, All
• The 3 categories(above-normal, near-normal, below-normal) are defined by standard deviations of the analyzed multi-years time series at each grid point.

• Probabilistic score Maps
• Variables to be Assessed: T2m, SST, RAIN
• Diagonostic Measures:
• Relative Operating Characteristics(ROC) areas (Grid point verification)
• Event: Anomaly > 0, Below Normal, Near Normal, Above Normal, All
• The 3 categories(above-normal, near-normal, below-normal) are defined by standard deviations of the analyzed multi-years time series at each grid point.

Specifications of the hindcast

• Model: GSM2003C + MRI.COMv4.6
• Ensemble size: 5
• Target period for verification: 1991-2020(T2m and SST), 1998-2020(RAIN)

Verification data

• Precipitation (RAIN) : GPCP (Global Precipitation Climatology Project) precipitation dataset (version 1.3)
• Atmospheric Analysis : JRA-3Q
• Sea surface temperature : MGDSST