000003602 001__ 3602
000003602 005__ 20181004113815.0
000003602 037__ $$aPOSTER-2018-0047
000003602 100__ $$aPodladchikova, O
000003602 245__ $$aStatistical approaches for the forecast of the F10.7 index 
000003602 260__ $$c2018
000003602 269__ $$c2018-11-09
000003602 520__ $$aSolar radio flux measurements at 10.7 cm provide a reliable monitoring dataset of the solar activity over the past six solar cycles. The radiation at 10.7 cm is coming from the upper chromospheric/low coronal layers of the Sun, and it is correlated with white light, sunspot number and UV radiation which impacts terrestrial atmospheric layers from the ionosphere till the stratosphere.   The statistically analysis of F10.7 index data set is very robust due to the nature of the ground based measurements which are practically unaffected by the weather conditions. Slow modulations of highly non-stationary F10.7 index data series have strong impact on the terrestrial climate, while fast changes - related to energetic solar events - have immediate impact on high frequency communications and on the satellite drag effect, which is significant for small size satellites. In this work, we discuss and update 3 different approaches for the forecast of F10.7 index: 1. SASFF (Self-Adjusted Solar Flux Forecasting) algorithm:  the radioflux is described by a non-stationary random walk model with variable drift and the forecast is performed using an adaptive Kalman Filter.  2. Random walk model with and unknown drift and also unknown variance.  The choice of the random walk model is justified by a very weak autocorrelation of the  radioflux increment. Uncertainty of the model parameters (drift, variance) is evaluated dynamically and applied for the next forecasting step. 3. Linear regression which considers dependences of the F10.7 index to other solar indices (e.g sunspot number). Considering the fact that constant  coefficients cannot reflect non-stationary radioflux behaviour, we correct the regression coefficients during observations using an adaptive Kalman Filter technique. A comparative analysis of the  forecasting errors is performed as the function of the solar cycle for every method. 
000003602 594__ $$aSTCE
000003602 6531_ $$aSun, space weather, models
000003602 700__ $$aMarque, C
000003602 773__ $$t15th, European Space Weather Week
000003602 8560_ $$felena.podladchikova@observatoire.be
000003602 980__ $$aCPOSTER