Local Geomagnetic Indices and the Prediction of Auroral Power

The aurora has been related to magnetometer observations for centuries, and to geomagnetic indices for decades. As the number of stations and data processing power increases, just how auroral power (AP) relates to geomagnetic observations becomes a more tractable question. This paper compares Polar UVI AP observations during 1997 with a variety of indices. Local time (LT) versions of the SuperMAG auroral electrojet (SME) are introduced and examined, along with the corresponding upper and lower envelopes (SMU and SML). Also, the East\textendashwest component, BE, is investigated. We also consider whether using any of the local indices is actually better at predicting local AP than a single global index. Each index is separated into 24 LT indices with a sliding 3-h MLT window. The ability to predict AP varies greatly with LT, peaking at 1900 MLT, where about 75\% of the variance (r2) is predicted at 1-min cadence. The aurora is fairly predictable from 1700 MLT \textendash 0400 MLT, roughly the region in which substorms occur. AP is poorly predicted from auroral electrojet indices from 0500 MLT \textendash 1500 MLT, with the minimum at 1000\textendash1300 MLT. In the region of high predictability, the local index which works best is BE (East\textendashwest), in contrast to long-standing expectations. However using global SME is better than any local index. AP is best predicted by combining global SME with a local index: BE from 1500\textendash0300 MLT, and either SMU or SML from 0300\textendash1500 MLT. In the region of the diffuse aurora, it is better to use a 30 min average than the cotemporaneous 1-min SME value, while from 1500\textendash0200 MLT the cotemporaneous 1-min SME works best, suggesting a more direct physical relationship with the auroral circuit. These results suggest a significant role for discrete auroral currents closing locally with Pedersen currents.