Bayesian filtering in incoherent scatter plasma parameter fits

Electron precipitation and ion frictional heating events cause rapid variations in electron temperature, ion temperature and F1 region ion composition of the high-latitude ionosphere. Four plasma parameters: electron density, electron temperature, ion temperature, and plasma bulk velocity, are typically fitted to incoherent scatter radar (ISR) data.Many ISR data analysis tools extract the plasma parameters using an ion composition profile from an empirical model. The modeled ion composition profile may cause bias in the estimated ion and electron temperature profiles in the F1 region, where both atomic and molecular ions exist with a temporally varying proportion.In addition, plasma parameter estimation from ISR measurements requires integrating the scattered signal typically for tens of seconds. As a result, the standard ISR observations have not been able to follow the rapid variations in plasma parameters caused by small scale auroral activity.In this project, we implemented Bayesian filtering technique to the EISCAT&#8217;s standard ISR data analysis package, GUISDAP. The technique allows us to control plasma parameter gradients in altitude and time.The Bayesian filtering implementation enabled us to fit electron density, ion and electron temperatures, ion velocity and ion composition to ISR data with high time resolution. The fitted ion composition removes observed artifacts in ion and electron temperature estimates and the plasma parameters are calculated with 5 s time resolution which was previously unattainable.Energy spectra of precipitating electrons can be calculated from electron density and electron temperature profiles observed with ISR. We used the unbiased high time-resolved electron density and temperature estimates to improve the accuracy of the estimated energy spectra. The result shows a significant difference compared to previously published results, which were based on the raw electron density (backscattered power) and electron temperature estimates calculated with coarser time resolution.&#160;

Download Full-text

Plasma parameter estimation from multistatic, multibeam incoherent scatter data

Journal of Geophysical Research Space Physics ◽

10.1002/2014ja020540 ◽

2014 ◽

Vol 119 (12) ◽

Cited By ~ 5

Author(s):

I. I. Virtanen ◽

D. McKay‐Bukowski ◽

J. Vierinen ◽

A. Aikio ◽

R. Fallows ◽

...

Keyword(s):

Parameter Estimation ◽

Plasma Parameter ◽

Incoherent Scatter ◽

Scatter Data

Download Full-text

Bayesian Filtering and Smoothing

10.1017/cbo9781139344203 ◽

2009 ◽

Cited By ~ 461

Author(s):

Simo Sarkka

Keyword(s):

Bayesian Filtering

Download Full-text

Imitation of incoherent scatter signals with consideration for altitudinal distribution of ionosphere parameters, plasma dynamics and noise components

Kosmìčna nauka ì tehnologìâ ◽

10.15407/knit2011.05.024 ◽

2011 ◽

Vol 17 (5) ◽

pp. 24-28

Author(s):

V.A. Puliaiev ◽

◽

O.V. Bohomaz ◽

Keyword(s):

Altitudinal Distribution ◽

Plasma Dynamics ◽

Incoherent Scatter

Download Full-text

Plasma parameter profile inference from limited data utilizing second-order derivative priors and physic-based constraints

Physics of Plasmas ◽

10.1063/5.0039011 ◽

2021 ◽

Vol 28 (3) ◽

pp. 032504

Author(s):

T. Nishizawa ◽

M. Cavedon ◽

R. Dux ◽

F. Reimold ◽

U. von Toussaint ◽

...

Keyword(s):

Plasma Parameter ◽

Second Order ◽

Order Derivative ◽

Limited Data ◽

Parameter Profile

Download Full-text

How to train your differentiable filter

Autonomous Robots ◽

10.1007/s10514-021-09990-9 ◽

2021 ◽

Author(s):

Alina Kloss ◽

Georg Martius ◽

Jeannette Bohg

Keyword(s):

Bayesian Filtering ◽

Learning Approaches ◽

Filtering Algorithms ◽

Recursive Filtering ◽

Process Dynamics ◽

Noise Characteristics ◽

Complex Models ◽

End To End ◽

Robotic Applications ◽

And Training

AbstractIn many robotic applications, it is crucial to maintain a belief about the state of a system, which serves as input for planning and decision making and provides feedback during task execution. Bayesian Filtering algorithms address this state estimation problem, but they require models of process dynamics and sensory observations and the respective noise characteristics of these models. Recently, multiple works have demonstrated that these models can be learned by end-to-end training through differentiable versions of recursive filtering algorithms. In this work, we investigate the advantages of differentiable filters (DFs) over both unstructured learning approaches and manually-tuned filtering algorithms, and provide practical guidance to researchers interested in applying such differentiable filters. For this, we implement DFs with four different underlying filtering algorithms and compare them in extensive experiments. Specifically, we (i) evaluate different implementation choices and training approaches, (ii) investigate how well complex models of uncertainty can be learned in DFs, (iii) evaluate the effect of end-to-end training through DFs and (iv) compare the DFs among each other and to unstructured LSTM models.

Download Full-text