CMR Header

CCMC Metadata Registry (CMR)

Menu Page

Go to:

CMR: View Simulation Model Info

ORIENT (1.0)

The Outer RadIation belt Electron Neural net model

Model Description

ORIENT (Outer RadIation belt Electron Neural neT model) is a set of machine learning model of Earth’s radiation belt electron flux. The ORIENT estimates the spin-average electron flux at different energy levels (50 keV ~ 2 MeV) at the L shell ranging from 2.6 Re to 6.6 Re(outer radiation belt) and at different MLT and MLAT. The model is trained using data from MagEIS and REPT instrument onboard the Van Allen Probes. The model driver is the time history of solarwind condition (velocity: Vsw, dynamic pressure: Psw), geomagnetic indices (AL or AE, SYM-H) and location (L, MLT,MLAT). The time history is ranging from 3 days to 20 days which depends on the target electron energy level. The model is trained using TensorFlow(https://www.tensorflow.org/) but can also be converted into different machine learning framework using ONNX(https://onnxruntime.ai/). The model has been extensively tested and validated showing very high accuracy performance for out-of-sample data (R^2 ∼ 0.78–0.92). Importantly, the ORIENT model successfully captures electron dynamics over long- and short timescales for a range of different energies.

Model Figure(s) :

Model Inputs Description

Time average of solar wind indices (Flow speed and Pressure), geomagnetic indices(AL or AE, SYM-H), location (MLT,MLAT,L)

Model Outputs Description

Electron flux at specific energy 

Model Caveats

The real-time of geomagnetic indices might be inaccurate and not accessible. 
e.g., Real time AE & AL might be not accessible.

Change Log


	
	 
	

Model Acknowledgement/Publication Policy (if any)

UCLA Atmospheric & Oceanic Science, Bortnik’s Group

Model Domains:

Magnetosphere.Inner_Magnetosphere.RingCurrent
Magnetosphere.Inner_Magnetosphere.RadiationBelt

Space Weather Impacts:

Near-earth radiation and plasma environment (aerospace assets functionality)

Phenomena :

Simulation Type(s):

Machine-Learning

Temporal Dependence Possible? (whether the code results depend on physical time?)

true

Model is available at?

CCMC

Source code of the model is publicly available?

true

CCMC Model Status (e.g. onboarding, use in production, retired, only hosting output, only source is available):

onboarding

Code Language:

python

Regions (this is automatically mapped based on model domain):

Earth.Magnetosphere

Contacts :

Yihua.Zheng, ModelHostContact
Jacob.Bortnik, ModelDeveloper
Donglai.Ma, ModelDeveloper

Acknowledgement/Institution :

UCLA Atmospheric & Oceanic Science, Bortnik’s Group

Relevant Links :

Publications :

  • Ma, D., Chu, X., Bortnik, J., Claudepierre, S. G., Tobiska, W. K., Cruz, A., et al. (2022). Modeling the dynamic variability of sub-relativistic outer radiation belt electron fluxes using machine learning. Space Weather, 20, e2022SW003079. https://doi.org/10.1029/2022SW003079
  • Chu, X., Ma, D., Bortnik, J., Tobiska, W. K., Cruz, A., Bouwer, S. D., et al. (2021). Relativistic electron model in the outer radiation belt using a neural network approach. Space Weather, 19, e2021SW002808. https://doi.org/10.1029/2021SW002808
  • Model Access Information :

    Access URL: https://github.com/donglai96/ORIENT
    Access URL Name: Public Repository
    Repository ID: spase://CCMC/Repository/NASA/GSFC/CCMC
    Availability: online
    AccessRights: OPEN
    Format: HTML
    Encoding: None

    Linked to Other Spase Resource(s) (example: another SimulationModel) :

    CMR Footer

    Curator: Chiu Wiegand | NASA Official: Dr. Masha Kuznetsova | Privacy and Security Notices | Accessibility | CCMC Data Collection Consent Agreement