Preprint: Harvey and Katrina may have lengthened the day by about 2 microseconds — briefly

A preprint posted to arXiv by Lin Li and colleagues argues that landfalling tropical cyclones can transiently change Earth's length of day by 1–2 microseconds, on the same numerical scale as the permanent shift NASA attributed to the 2011 Tohoku earthquake. Using atmospheric reanalysis, they estimate Hurricane Harvey (2017) lengthened the day by 1.87 microseconds and Katrina (2005) by 2.29 microseconds. Numerical simulations are said to confirm the reanalysis estimate. The paper is not peer-reviewed.

The methods note matters more than the headline. This is not a direct measurement of Earth rotation with VLBI or GNSS. It is an angular-momentum budget: reanalysis winds and mass fields are converted into atmospheric angular momentum, then into an equivalent length-of-day anomaly, with simulations used as a cross-check. The abstract does not state how many storms were analyzed, what reanalysis product was used, or the uncertainty on those microsecond figures. Until the methods section is scrutinized in review, the 1.87 and 2.29 values are model-derived estimates, not observed clock errors.

What the framing misses is that weather already dominates short-term length-of-day variability. Since the early 1980s, geodesists have shown atmospheric angular momentum accounts for most LOD changes from days to seasons, with amplitudes of hundreds of microseconds, far above any single cyclone. Earthquakes such as Tohoku change the solid Earth's moment of inertia in a step that does not reverse. A cyclone's exchange is a temporary torque that should rise and fall as the vortex decays at landfall. Matching Tohoku's 1.8 microseconds compares peak size, not geophysical kind or practical consequence.

Navigation and timekeeping will not feel a 2-microsecond day. Leap seconds manage UT1–UTC gaps near 0.9 seconds, and Earth-orientation services already fold global atmospheric models into length-of-day predictions. The climate claim, that warmer-world cyclones become an increasingly important rotation source, is the weakest line in the abstract. It extrapolates two case estimates without a published share of total atmospheric excitation, while cyclone-frequency projections remain contested. A multi-decade storm census, stated uncertainties, and a residual test against IERS series after the background atmosphere is removed would strengthen the evidence.