LANL* V1.0: a radiation belt drift shell model suitable for real-time and reanalysis applications

Journal Article · Tue Jan 01 00:00:00 EST 2008 · Geoscientific Model Development

OSTI ID:960830

Koller, Josep ^[1]; Reeves, Geoffrey D ^[1]; Friedel, Reiner H W ^[1]

---

1. Los Alamos National Laboratory

Space weather modeling, forecasts, and predictions, especially for the radiation belts in the inner magnetosphere, require detailed information about the Earth's magnetic field. Results depend on the magnetic field model and the L* (pron. L-star) values which are used to describe particle drift shells. Space wather models require integrating particle motions along trajectories that encircle the Earth. Numerical integration typically takes on the order of 10{sup 5} calls to a magnetic field model which makes the L* calculations very slow, in particular when using a dynamic and more accurate magnetic field model. Researchers currently tend to pick simplistic models over more accurate ones but also risking large inaccuracies and even wrong conclusions. For example, magnetic field models affect the calculation of electron phase space density by applying adiabatic invariants including the drift shell value L*. We present here a new method using a surrogate model based on a neural network technique to replace the time consuming L* calculations made with modern magnetic field models. The advantage of surrogate models (or meta-models) is that they can compute the same output in a fraction of the time while adding only a marginal error. Our drift shell model LANL* (Los Alamos National Lab L-star) is based on L* calculation using the TSK03 model. The surrogate model has currently been tested and validated only for geosynchronous regions but the method is generally applicable to any satellite orbit. Computations with the new model are several million times faster compared to the standard integration method while adding less than 1% error. Currently, real-time applications for forecasting and even nowcasting inner magnetospheric space weather is limited partly due to the long computing time of accurate L* values. Without them, real-time applications are limited in accuracy. Reanalysis application of past conditions in the inner magnetosphere are used to understand physical processes and their effect. Without sufficiently accurate L* values, the interpretation of reanalysis results becomes difficult and uncertain. However, with a method that can calculate accurate L* values orders of magnitude faster, analyzing whole solar cycles worth of data suddenly becomes feasible.

• Research Organization: Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC52-06NA25396
• OSTI ID: 960830
• Report Number(s): LA-UR-08-07430; LA-UR-08-7430; TRN: US1002644
• Journal Information: Geoscientific Model Development, Journal Name: Geoscientific Model Development; ISSN 1991-959X
• Country of Publication: United States
• Language: English

Related Subjects

70 PLASMA PHYSICS AND FUSION TECHNOLOGY
58 GEOSCIENCES
97 MATHEMATICS AND COMPUTING
ACCURACY
DATA
DENSITY
DYNAMICS
EARTH PLANET
EARTH MAGNETOSPHERE
ELECTRONS
FORECASTING
INFORMATION
LANL
MAGNETIC FIELDS
NEURAL NETWORKS
PARTICLES
PHASE SPACE
PRODUCTION
RADIATION BELTS
SATELLITES
SHELL MODELS
SHELLS
SIMULATION
SOLAR CYCLE
SPACE
STANDARDS
TRAJECTORIES
USES
WEATHER