Application of the INGV MUF nowcasting and forecasting methods for PECASUS to relevant cases for aviation

,

Such maps are created and delivered to PECASUS with the following specifications.
Spatial coverage: 40°W-100°E lon; 20°N-80°N lat Spatial resolution: 2.5° lat x 5.0° lon Update rate: 60 min Predicted time: 1-24 hours ahead Note that the spatial extension of such maps is larger than nowcasted maps, thanks to the lower dependence to real-time observations from the ionospheric stations.
The activity was triggered by high speed solar wind (~600 km/s) emanating from a solar coronal hole which, combined with a southward component of IMF, enhanced the energy input from the solar wind to the magnetosphere.
A moderate geomagnetic storm arised, with a maximum ap value of 56 (i.e.Kp=5+) during the three last UT hours of 27 Sept. 8 (blue square), while the Dst index reached a minimum of -57 nT at 04:00UT on 28 Sept. (red circle).Enhanced geomagnetic activity (Kp 4-5) continued still throughout the day on 28 Sept., but the activity ceased during the next days.
The first signs of Post-Storm Depression were observed around 05:30 UT on 28 Sept. by the Australia-Canada-France-Japan consortium (ACFJ), which served as the on-duty center at that time.During the event ACFJ sent altogether 13 advisories about subsequent depressions in different regions of the globe over a period of 3.5 days.
In particular, depression was observed in the Central European sector (lat 42°-55ºN; lon 14°-30°E) on the period 28-30 Sept., when moderate Post-Storm Depression took place during both daytime and nighttime.

MUF(3000) nowcasting
MUF(3000) hourly values have been used to test the method during the storm, comparing measurements and predictions at the test station of Fairford, UK (51 7 ºN, 1 5 ºW).Data from Chilton, UK (51 5 ºN 0 6 ºW) station have been excluded from assimilation, being too close to Fairford

Data analysis results
In this plot are shown observed and modelled MUF(3000) values at Fairford, along with their difference, and good agreement was observed, with RMSEs always lower than 1.25 MHz, and particularly low during the most disturbed day.
It should be noted that the anomalous value recorded on October 1st at 05.00 UT is a consequence of wrong autoscaling at the test station, which would affect the accuracy estimation if not correctly recognized and discarded.

Observed depression on 28 Sept. from maps
As can be seen from these maps of example, a depression from about 30% to 45% is observed on 28 Sept. in the European region, as a result of the upgrade of the background model with the data of 8-9 stations in the area.

Rejection and failures
Some cases are automatically discarded by the procedure: 1. number of assimilated stations < 4; 2. failure of variogram fitting (on the basis of the usual Q1 and Q2 statistical tests); 3. "bad" map (i.e.uniform ratio map with unitary ratio everywhere).
In particular, we can consider the two last cases as failures of the procedure, because something goes wrong within the Kriging procedure.
In this case, there are no a-priori rejections (reason 1) thanks to the good number of available data, which is not always the case for near real-time application of the method.
Conversely, a percentage of failures (reason 2 and 3) of 40% arised.Besides, automatic rejection for reason 3 is not fully efficient, and some "bad" maps are still created.
It is encouraging to notice that during this moderate negative MUF depression event, EUROMAP follows the observations much better than the IRI(STORM) model, although some signicative discrepancies are sometimes observed (see e.g.Juliusruh and Moscow on 28 Sept. during daytime).As can be seen, foF2 by EUROMAP, and MUF(3000) forecasting predictions are always better than IRI(STORM) ones.Sometimes (see e.g.Moscow), slightly worse results are obtained for MUF(3000) than foF2, due to the difference between observed MUF(3000) and those calculated using M(3000) from IRI.
Besides, better results for lead time from 1 to 3 hours, and equal to 24 hours are always obtained, probably due to the origin of the negative storm, which is related to neutral composition changes.The MUF(3000) predicted 1-24 hours ahead depends on f F2 and M(3000) ionospheric parameters: EUROMAP forecasting model and IRI model are used for the former and the latter, respectively.The method has been applied to Europe where there are ionospheric stations with long (for some solar cycles) historical data and current real-time f F2 observations.A mapping procedure applied to the European stations provides MUF(3000) short-term prediction over the whole area.
The application of these methods to storm events occurred after November 2019 is here presented, in order to study the ionospheric conditions they provide when HF COM advisories are expected to be issued.
As it is known, Space Weather (SWx) phenomena can have dramatic impact on satellite navigation and HF radio communication systems, being also responsible for increases on radiation levels at flight altitudes.For this reason, in recent years the International Civil Aviation Organization (ICAO) has been showing great interest in operational SWx services for aviation purposes in these three domains.Four global SWx centers have been then appointed since November 2019 by ICAO to provide real-time SWx advisories for aviation users.In particular, HF COM conditions are assessed by monitoring the F2-layer critical frequency f F2 or the MUF(3000) ionospheric characteristic (MUF = Maximum Usable Frequency), the latter representing the highest HF radio frequency that can be used for communications over a standard distance of 3000 km via F2-layer ionospheric reflection.As one of the designed SWx centers, several key operational 24/7 products for HF COM conditions assessment have been developed within PECASUS (Partnership for Excellence in Civil Aviation Space weather User Services).Nowcasting and forecasting (1-24hr) maps over Europe of MUF(3000) and its ratio with respect to a background level are then developed by INGV, as a PECASUS partner.The MUF(3000) nowcasting uses all the available real-time ionosonde measurements in different locations in order to upgrade IRI-CCIR-based background maps, and Ordinary Kriging method for spatial interpolation.The MUF(3000) modeling performance was assessed comparing predicted values to measured ones over two test stations during strong geomagnetic storm periods, obtaining an overall RMSE < 2 MHz at both stations.