Comparison of Earth rotation excitation in data‐constrained and unconstrained atmosphere models

Changes in Earth rotation are strongly related to fluctuations in the angular momentum of the atmosphere, and therefore contain integral information about the atmospheric state. Here we investigate the extent to which observed Earth rotation parameters can be used to evaluate and potentially constrain atmospheric models. This is done by comparing the atmospheric excitation function, computed geophysically from reanalysis data and climate model simulations constrained only by boundary forcings, to the excitation functions inferred from geodetic monitoring data. Model differences are assessed for subseasonal variations, the annual and semiannual cycles, interannual variations, and decadal‐scale variations. Observed length‐of‐day anomalies on the subseasonal timescale are simulated well by the simulations that are constrained by meteorological data only, whereas the annual cycle in length‐of‐day is simulated well by all models. Interannual length‐of‐day variations are captured fairly well as long as a model has realistic, time‐varying SST boundary conditions and QBO forcing. Observations of polar motion are most clearly relatable to atmospheric dynamics on subseasonal to annual timescales, though angular momentum budget closure is difficult to achieve even for data‐constrained atmospheric simulations. Closure of the angular momentum budget on decadal timescales is difficult and strongly dependent on estimates of angular momentum fluctuations due to core‐mantle interactions in the solid Earth.