{"method_version":"analog-v1.0.0","feature_version":"features-v1.0.0","weight_version":"weights-v1.0.0","question":"Given the time of year, the current state of the major teleconnections, and how they evolved into that state, which previous year most closely resembled the present?","family_weights":{"enso":0.3,"nh_extratropical":0.3,"mjo":0.125,"sh_extratropical":0.1,"decadal":0.1,"stratosphere":0.075},"component_weights":{"state":0.35,"trajectory":0.4,"persistence":0.15,"background":0.1},"trajectory_windows":[30,60,90],"default_lookback":60,"calendar":{"default_tolerance_days":7,"max_tolerance_days":15,"penalty_per_day":0.012,"leap_rule":"Feb 29 maps to Feb 28 in non-leap years"},"eligibility":{"min_coverage":0.7,"min_families":3,"current_year_excluded":true,"one_entry_per_year":true},"modes":{"modern":"Full modern field including MJO (from 1991) and AAO (from 1979).","long":"Long-record core reaching 1951: AO, NAO, PNA, ONI, RONI, SOI, PDO, IPO, QBO."},"vintage_modes":{"realtime":"Uses only information whose period ended and whose modeled release date had passed at the historical comparison date. Public default.","final":"Uses currently preferred final values. Exploratory only; unsuitable for strict real-time hindcast evaluation."},"release_lag_model":{"daily":3,"weekly":6,"monthly":8,"seasonal":8},"mjo_handling":"Compared as an RMM1/RMM2 vector, so phase 8 and phase 1 are adjacent rather than maximally different, and weak amplitude naturally carries less influence than a strong event.","native_frequency_rule":"Monthly and seasonal indices never produce daily samples. Trajectory features exist only for genuinely daily series.","score_transform":"Component similarities exp(-0.5 * d^2) on robust-scaled differences, weighted by component, multiplied by sqrt(coverage) and a calendar-displacement penalty, expressed 0-100.","limitations":["An analog year is a historical similarity, not a forecast.","The similarity score is an index from this algorithm, not a probability.","Teleconnection relationships are seasonally dependent.","Indices are mutually correlated; family weighting mitigates but does not remove this.","Index definitions and normalization periods change over time.","Historical observations may be revised.","Similar large-scale states can produce different regional weather.","Agricultural effects depend on geography, crop, growth stage and local weather, and are not implied by an analog."]}