[P-EM09] Space Weather and Space Climate

[PEM09-P07] Forecast of Geomagnetic Field Disturbances Using the Empirical Model for Space Weather

*Naoko Takahashi¹, Aoi Nakamizo¹, Kaori Sakaguchi¹, Daikou Shiota¹ (1.National Institute of Information and Communications Technology)

The geomagnetic field disturbance is one of the essential parameters for the space weather forecast in terms of the indicator of disturbances of the Earth's magnetosphere. The magnetospheric condition strongly depends on the solar wind variation associated, for example, coronal mass ejection and/or co-rotating interaction region. Particularly, strong solar wind inputs cause a change of the ring current, resulting a magnetic storm that can be detected as a global change of magnetic field both in space and on the ground. The disturbance field (Dst) index, which is a parameter that measures the magnitude of the ring current, is referred to understand the magnetospheric condition. In this study, we adapt the Dst index forecasting model based on two empirical models proposed in O'Brien and McPherron (2000) and Keika et al. (2015). We also attempt to estimate K-index from estimated Dst index that commonly used as the criteria for geomagnetic disturbance alerts in Japan.
We perform a few hours forecast evaluation using the DSCOVR spacecraft data as the inputs. The estimated Dst index shows a good correlation with the observed Dst index during magnetic storms. However, the abrupt change such as sudden commencement at the beginning of magnetic storm cannot be reproduced well due to the limitation of time resolution. We also estimate K-index using the Dst index and compare with the K-index calculated from the geomagnetic field variation at Kakioka (called as Kakioka K-index). The estimated K-index overestimates comparing with Kakioka K-index because the Dst index is derived from geomagnetic field variation observed at defined four low latitude stations. We also perform a few days forecast evaluation using SUSANOO-CME data as the inputs. The decrease of Dst index estimated from SUSANOO-CME data is reproduced though it is smaller than that of the observed Dst index. In this presentation, we will discuss the forecast accuracy and future perspective for the forecasting using machine learning.