Advances in METplus Verification for Subseasonal-to-Seasonal Model Evaluation

Model evaluation and validation is an important part of model improvement. The METplus system combines multiple verification options in a consistent framework and is designed to facilitate quick setup. METplus contains several components, including the Model Evaluation Tools (MET) for computing verification statistics, METcalcpy, which contains python versions of statistics and process-oriented verification, and METplotpy for plotting. In addition, python scripts can also allow for combination of the different METplus components. Recently, a set of process-oriented diagnostic and verification metrics have been added to the METplus system to examine the predictability of phenomena on subseasonal to seasonal time scales.

Specifically, atmospheric blocking events are identified by first locating reversals in the 500 hPa geopotential height gradient as blocked longitudes, followed by applying spatial and temporal thresholds to ensure the large-scale, quasi-stationary characteristics of blocking anticyclones are met. Additionally, weather regime patterns and their frequency of occurrence are computed using K-means clustering and optionally empirical orthogonal functions. Finally, the Real-Time Multivariate MJO (RMM) Index and OLR-based MJO Index (OMI) to evaluate the signal of MJO, as well as phase diagrams have been added to METplus. This presentation will provide a description of these new capabilities in the METplus system, and show examples with the GFS model and ERA reanalysis.