Research Article

Assimilation of real-time riometer measurements into models of 30 MHz polar cap absorption

Space Plasma Environment and Radio Science Group (SPEARS), Department of Physics, Lancaster University, Lancaster LA1 4YB, UK

^* Corresponding author: ncrogers@physics.org

Received: 13 November 2014
Accepted: 9 March 2015

Abstract

Space weather events may adversely affect high frequency (HF) radio propagation, hence the ability to provide nowcasting and forecasting of HF radio absorption is key for industries that rely on HF communications. This paper presents methods of assimilating 30 MHz radio absorption measurements into two types of ionospheric polar cap absorption (PCA) model to improve their performance as nowcasting tools. Type 1 models calculate absorption as m times the square root of the flux of solar protons above an energy threshold, E_t. Measurements from 14 riometers during 94 solar proton events (1995–2010) are assimilated by optimising the day and night values of m by linear regression. Further non-linear optimisations are demonstrated in which parameters such as E_t are also optimised and additional terms characterise local time and seasonal variations. These optimisations reduce RMS errors by up to 36%. Type 2 models incorporate altitude profiles of electron and neutral densities and electron temperatures. Here the scale height of the effective recombination coefficient profile in the D-region is optimised by regression. Furthermore, two published models of the rigidity cut-off latitude (CL) are assessed by comparison with riometer measurements. A small improvement in performance is observed by introducing a 3-h lag in the geomagnetic index K_p in the CL models. Assimilating data from a single riometer in the polar cap reduces RMS errors below 1 dB with less than 0.2 dB bias. However, many high-latitude riometers now provide absorption measurements in near-real time and we demonstrate how these data may be assimilated by fitting a low-order spherical harmonic function to both the measurements and a PCA model with optimised parameters.