Prediction technique

Introduction

Geomagnetic storms and substorms are a natural hazard similar to hurricanes, floods etc. Severe storms cause communications problems, abruptly increase drag on spacecraft, and can even cause electric utility blackouts over a wide area. Severe substorms strongly affect communications and electrical systems in high-latitude and polar regions. Solar and geomagnetic activity also probably influences the Earth's climate and human health.

North-South (B_z) component of the Interplanetary Magnetic Field (IMF) in the GSM frame of reference is the main factor controlling geomagnetic activity. Under southward directed IMF the Earth's magnetosphere is open to input from the solar wind flow. This energy is accumulated in the magnetosphere and is subsequently released and dissipated during geomagnetic storms and substorms. Other characteristics of the solar wind are less important.

Since practical importance of solar wind measurements was recognized, a number of approaches was developed for prediction of the geomagnetic activity basing on the solar wind data. Usually geomagnetic indices K_p, D_st, AE are estimated with the use of linear filters or neural networks.

In our technique simple and a bit different approach is used: As geomagnetic activity is clearly controlled by the solar wind, we assume that an appropriate function of the solar wind parameters describing input to the magnetosphere, can characterize also the geomagnetic activity. Predicted level of the geomagnetic activity is formulated qualitatively (e.g., quiet, weak, strong ...)

Such qualitative characteristics appear to be less accurate compared with results of other methods in which concrete values of D_st and K_p indices are obtained. However, at present time interpretation of these indices in terms of their practical consequencies is anyway qualitative, especially during the real-time forecast. Difference between actual and predicted indices (that is error of prediction) is also often high. Therefore, we consider that qualitative prediction of the level of geomagnetic activity is adequate, at least at the current level of our knowledge.

Calculation algorithm

We have chosen epsilon parameter (ε) as such integral characteristic of the solar wind influence on the magnetosphere. It was first introduced by Perreault and Akasofu [Geophys. J. R. Astr. Soc., 54, 547, 1978] as the measure of the solar wind energy input due to the process of the magnetic field merging at the front-side magnetopause. Here we use it in the form ε = 2•10⁷•V_i•B²•sin(θ/2)⁴ Watt, where B is in nT, V_i is in km/s, θ is IMF angle in YZ plane. We integrate ε to obtain total energy input during some specific interval of time.

Typical geomagnetic storm starts after 3 hours of B_z=-10 nT [Gonsalez et al, JGR, 99, 5771-5792, 1994]. The corresponding energy input is about 10¹⁶ Joules. Typical energy accumulated during substorm growth phase is about 10¹⁵ J. [Pulkkinen et al., JGR, 103, 15-28, 1998]. Contracted-oval (small) substorms are often associated with only 10¹⁴ J input [Petrukovich et al, subm. to JGR].

Three parameters are calculated: integral of energy input during preceeding 90 min (E₉₀), integral of energy input during preceeding 180 min (E₁₈₀) and integral of energy input during interval with ε > 10¹⁰ W (E_ac). The reset to E_ac = 0 occurs after 30 minutes of input smaller than this threshold. E₉₀ parameter better describes substorms-level activity, while E₁₈₀ and E_ac are more adequate for storm-level activity. More information can be found in Petrukovich et al., [subm. to GRL].

Prediction algorithm

Geomagnetic activity predictions calculated by our method are presented in the plots of ACE data as color-coded bars.

Substorm activity is predicted basing on E₉₀:

Quiet: E₉₀ < 10¹⁴ J. No activity or very small activations observed mainly by optical tools. IMF is probably northward (at least on average).
Weak substorms: E₉₀ = 10¹⁴ - 10¹⁵ J. Contracted oval substorms and/or normal substorms of low amplitude are likely. Geomagnetic activity in the auroral zone is probably less than 500 nT.
Strong substorms: E₉₀ = 10¹⁵ - 5•10¹⁵ J. Strong global substorms, all night-side magnetosphere is involved. Auroral geomagnetic disturbances are probably less than 1000 nT.
Storm-time substorms: E₉₀ > 5•10¹⁵ J. Ground activity more than 1000 nT is anticipated. Most probable during geomagnetic storm intervals.

Storm activity is predicted using E₁₈₀, E_ac and E₉₀:

Quiet: No storm. All what doesn't fit into other categories below.
Weak activity: E₁₈₀ < 10¹⁶ J, but E_ac > 10¹⁶ and E₉₀ > 2•10¹⁵ J. D_st is probably within -50 nT. Typical during large recurrent substorms and HILDCAA type events (prolonged intervals with moderate southward IMF > -10nT ).
Intense Storm: E₁₈₀ = 10¹⁶ - 10¹⁷ J. D_st is probably between -50 nT and -200 nT.
Major storm: E₁₈₀ > 10¹⁷ J. D_st is probably < -200 nT.

D_st prediction

We also supply D_st geomagnetic index prediction calculated with the use of Burton et al. [JGR, 80, 4204-4214, 1975] approach with modifications of Murayama [Rev. Geophys. Space Phys., 20, 623-629, 1982] and Feldstein et al. [Space Sci. Rev., 59, 83, 1992, Planet. Space Sci., 32, 975-984, 1984].

Disclaimer. We cannot guarantee the quality and reliability of these predictions and do not encourage their use for other than illustrative purposes.

Author: Anatoli Petrukovich
Questions and comments are welcomed at apetruko@iki.rssi.ru