Periodic Reporting for period 2 - FARBES (Forecast of Actionable Radiation Belt Scenarios)
Periodo di rendicontazione: 2024-01-01 al 2025-12-31
Forecast in Space Weather modeling mostly ignore the fact everything is driven by the Sun, that is basically unpredictable. Propagating observed solar dynamics to Earth is questionable, it depends on models whos boundary conditions we are incapable of constraining. We are limited to data collected by satellites at L1 Lagrangian Point, 1.5 million km from the Earth, giving a one hour lead time and neural net type forecasts of controlling parameters ( e.g. Kp) that govern the physics of our best models.
Nowcasts are better: advanced data assimilation techniques with physics based models show great fidelity in reproducing the real radiation belt (RB) environment. Operational use of such Nowcasts is limited by lack of high quality real-time data beyond Geosynchronouos orbit.
The FARBES project is different: it limits its ambition to simple, achievable prediction goals that are of utility to satellite operators, while avoiding the pitfalls of past projects. We hold that while it may be impossible to accurately predict the break of a space weather event, once an event has started we have the tools to predict subsequent behavior and to update our predictions during the event.
While we may not be able to globally predict in detail the subsequent dynamic behavior, we can provide actionable forecasts for satellite operators on a few key event characteristics:
a. Time to most severe environment
b. Most severe Flux reached
c. Time to the end of event
These characteristics were deemed most useful by spacecraft operator representatives at ESWW16 [http://www.stce.be/esww13/contributions/public/S5-O1/S5-O1-03-PitchfordDave/FORECASTINGTHEPERFECTSTORM.ppt].
We overcome the data-assimilation nowcast limitations by using physics based models driven by simple, affordable and reliable ground-based real-time inputs only, we overcome our inability to accurately forecast magnetospheric drivers by using a scenario-driven forecast approach for RB dynamics starting with nowcast and is constantly refined during an event by the ongoing availability of real-time model inputs
Nowcasts are better: advanced data assimilation techniques with physics based models show great fidelity in reproducing the real radiation belt (RB) environment. Operational use of such Nowcasts is limited by lack of high quality real-time data beyond Geosynchronouos orbit.
The FARBES project is different: it limits its ambition to simple, achievable prediction goals that are of utility to satellite operators, while avoiding the pitfalls of past projects. We hold that while it may be impossible to accurately predict the break of a space weather event, once an event has started we have the tools to predict subsequent behavior and to update our predictions during the event.
While we may not be able to globally predict in detail the subsequent dynamic behavior, we can provide actionable forecasts for satellite operators on a few key event characteristics:
a. Time to most severe environment
b. Most severe Flux reached
c. Time to the end of event
These characteristics were deemed most useful by spacecraft operator representatives at ESWW16 [http://www.stce.be/esww13/contributions/public/S5-O1/S5-O1-03-PitchfordDave/FORECASTINGTHEPERFECTSTORM.ppt].
We overcome the data-assimilation nowcast limitations by using physics based models driven by simple, affordable and reliable ground-based real-time inputs only, we overcome our inability to accurately forecast magnetospheric drivers by using a scenario-driven forecast approach for RB dynamics starting with nowcast and is constantly refined during an event by the ongoing availability of real-time model inputs
1st year:
1. We have created a complex wave propagation model, consisting of three submodels:
a) magnetospheric propgation
b) transionospheric propgation
c) subionospheric propgation
The submodels have been separately verified. The full model was used to calculate the in-situ VLF power for selected events. The verification of the model is going on.
2. We have developed a code that can read data files from EMMA and ENIGMA magnetometers, perform FFT to obtain power spectral densities for hourly periods, transform those into electric field power spectral densities at the equator in space, and derive hourly values of the electric field radial diffusion coefficient. We have also added a functionality for including an assessed contribution of the magnetic field radial diffusion coefficient, using a statistical approximation that links the two, however imperfectly; and performed initial tests on historical dates of quiet and active geomagnetic conditions.
3. We have developed a method for the definition, automatic detection and analysis of the events In the Radiation Belts using Ca index.
4. We have implemented the Analog Ensemble method to provide scenario-based forecasts for FARBES.
2nd-3rd year:
1. We have validated the wave propagation model through simultaneous measurements from ground VLF data and Van Allen Probes burst mode data. We have developed a neural-network based (YOLO v11) chorus detector and used it to identify chorus event in ground and in-situ data. From the identified chorus events, the wave powers were extracted and then, using the wave propagation model developed in the 1st year, the estimated wave powers were calculated. These wave powers were used as proxies to scale the Salammbo VLF wave power statistics.
2.We have extended the method to derive the electric term of t the radial diffusion coefficient based on on ground ULF measurements. The method was validated by using in-situ data measured by the THEMIS spacecrafts. The diffusion coefficients were then integrated to Salammbô code.
3. We have developed a dynamic VLF wave power statistics based on solar wind event classes (CIR, CME and CIR+CME), magnetic activity (four AE index classes) and in-situ wave data from several satellite instruments, such as THEMIS A (FFT), CLUSTER 1 (STAFF SA), DE1 (PWI), POLAR (PWI), RBSP A (EMFISIS), creating a binned (L*/MLT/Activity/AE) chorus wave power statistics.
4. Salammbo simulations using Analog Ensemble method were run on two event scenarios (Ca75 (03/2015) and Ca90 (08/2015) events) using diffusion coefficients obtained from ground based VLF and ULF measurements. The simulation results are superior to the one based on diffusion coefficients obtained form Salammbo default wave power statistics and on certain energy ranges are better than the one based on the new dynamic wave power statistics.
1. We have created a complex wave propagation model, consisting of three submodels:
a) magnetospheric propgation
b) transionospheric propgation
c) subionospheric propgation
The submodels have been separately verified. The full model was used to calculate the in-situ VLF power for selected events. The verification of the model is going on.
2. We have developed a code that can read data files from EMMA and ENIGMA magnetometers, perform FFT to obtain power spectral densities for hourly periods, transform those into electric field power spectral densities at the equator in space, and derive hourly values of the electric field radial diffusion coefficient. We have also added a functionality for including an assessed contribution of the magnetic field radial diffusion coefficient, using a statistical approximation that links the two, however imperfectly; and performed initial tests on historical dates of quiet and active geomagnetic conditions.
3. We have developed a method for the definition, automatic detection and analysis of the events In the Radiation Belts using Ca index.
4. We have implemented the Analog Ensemble method to provide scenario-based forecasts for FARBES.
2nd-3rd year:
1. We have validated the wave propagation model through simultaneous measurements from ground VLF data and Van Allen Probes burst mode data. We have developed a neural-network based (YOLO v11) chorus detector and used it to identify chorus event in ground and in-situ data. From the identified chorus events, the wave powers were extracted and then, using the wave propagation model developed in the 1st year, the estimated wave powers were calculated. These wave powers were used as proxies to scale the Salammbo VLF wave power statistics.
2.We have extended the method to derive the electric term of t the radial diffusion coefficient based on on ground ULF measurements. The method was validated by using in-situ data measured by the THEMIS spacecrafts. The diffusion coefficients were then integrated to Salammbô code.
3. We have developed a dynamic VLF wave power statistics based on solar wind event classes (CIR, CME and CIR+CME), magnetic activity (four AE index classes) and in-situ wave data from several satellite instruments, such as THEMIS A (FFT), CLUSTER 1 (STAFF SA), DE1 (PWI), POLAR (PWI), RBSP A (EMFISIS), creating a binned (L*/MLT/Activity/AE) chorus wave power statistics.
4. Salammbo simulations using Analog Ensemble method were run on two event scenarios (Ca75 (03/2015) and Ca90 (08/2015) events) using diffusion coefficients obtained from ground based VLF and ULF measurements. The simulation results are superior to the one based on diffusion coefficients obtained form Salammbo default wave power statistics and on certain energy ranges are better than the one based on the new dynamic wave power statistics.
1. We have developed a method to identify scenarios and defined scenarios based on Ca index.
2. We have developed a VLF wave propagation model to estimate the equatorial wave power from ground based VLF measurements. The model has been validated by simultaneous ground based and in-situ chorus measurements. To detect chorus events and extract wave power , we developed a YOLO v11 based algorithm.
3. We have developed a method to calculate the electric term of DLL from Pc5 ULF waves . The method has been validated using ground (EMMA) and in-situ (THEMIS) data.
4. Both ground based ULF and VLF wave data were used to calculate diffusion coefficients to feed Salammbo simulations using Analog Ensemble method for selected scenarios. This method is proved to be superior to the one using statistics without ground based data.
5. The results show, that the existing/available ground based wave data are not enough to infer diffusion coefficients completely, this is why we have used statistics with scaling. The original idea of FARBES was to include extensive development of ground based wave recording facilities, however, the lack of fund for this task prevented this and thus we were restricted to use a rather limited ground dataset. Future extension of ground measuring networks/stations with extended capabilities, such as real time detection of wave phenomena and extraction of wave power for large range of magnetic latitudes would complete the original idea behind FARBES.
6. Further research is needed to resolve the apparent contradiction between the long time range (hours/days) based diffusion coefficients used in Radiation Belt simulation codes, like Salammbo and the short lived (milliseconds/seconds) wave phenomena that play crucial roles in wave particle interaction. This is very important as no real-time in-situ wave data are available to date and without them the forecasting/nowcasting models are limited to use statistics. Ground based data can substitute the missing real-time in-situ data.
2. We have developed a VLF wave propagation model to estimate the equatorial wave power from ground based VLF measurements. The model has been validated by simultaneous ground based and in-situ chorus measurements. To detect chorus events and extract wave power , we developed a YOLO v11 based algorithm.
3. We have developed a method to calculate the electric term of DLL from Pc5 ULF waves . The method has been validated using ground (EMMA) and in-situ (THEMIS) data.
4. Both ground based ULF and VLF wave data were used to calculate diffusion coefficients to feed Salammbo simulations using Analog Ensemble method for selected scenarios. This method is proved to be superior to the one using statistics without ground based data.
5. The results show, that the existing/available ground based wave data are not enough to infer diffusion coefficients completely, this is why we have used statistics with scaling. The original idea of FARBES was to include extensive development of ground based wave recording facilities, however, the lack of fund for this task prevented this and thus we were restricted to use a rather limited ground dataset. Future extension of ground measuring networks/stations with extended capabilities, such as real time detection of wave phenomena and extraction of wave power for large range of magnetic latitudes would complete the original idea behind FARBES.
6. Further research is needed to resolve the apparent contradiction between the long time range (hours/days) based diffusion coefficients used in Radiation Belt simulation codes, like Salammbo and the short lived (milliseconds/seconds) wave phenomena that play crucial roles in wave particle interaction. This is very important as no real-time in-situ wave data are available to date and without them the forecasting/nowcasting models are limited to use statistics. Ground based data can substitute the missing real-time in-situ data.