Search

Abstract During geomagnetic storms, the capabilities of current climate models in predicting ionospheric behavior are notably limited. A data assimilation tool, Estimating Model Parameters Reverse Engineering (EMPIRE), implements a Kalman filter to ingest electric density rate correcting the background electric potential and neutral wind. For the baseline setup, or case (1), EMPIRE ingests electron density global map output from the Ionospheric Data Assimilation 4‐Dimensional (IDA4D) algorithm. In this work, a new augmentation method is evaluated in which ion drift measurements are also assimilated into EMPIRE. The ion drift measurements used in the new augmentation method are obtained from Super Dual Auroral Radar Network (SuperDARN) sites in the mid‐to‐high latitude region of the northern hemisphere. Cases (2) and (3) are set up for evaluating the impacts from ingesting different types of observations: SuperDARN fit and grid data, respectively. Six independent data sources are used as validation data sets to compare outcomes with or without ingesting ion drifts. One is the vector ion velocities derived from the Millstone Hill Incoherent Scatter Radar (MHISR) and a second is the vertical drift from Arecibo site. The other four are SuperDARN ion velocity grid data from Saskatoon, Kapuskasing, Christmas Valley West, and Hokkaido East. Results show improvements in performance at mid‐latitudes by augmenting electron density rates with 3D spatially distributed line‐of‐sight ion drift measurements, with negligible improvements to low and high latitude estimations. The lack of improvement at high‐latitudes is attributed to the increase in EMPIRE ion drift error poleward of 60° magnetic.

Abstract In this work we quantify the representation error of the algorithm Estimating Model Parameters from Ionospheric Reverse Engineering (EMPIRE), which estimates global neutral winds and ion drifts given time‐varying plasma densities. SAMI3 (Sami3 is A Model of the Ionosphere) serves as the background climate model and pseudo‐measurements for the EMPIRE observation system. This configuration allows the data assimilation inputs to be self‐consistent between each other and with the validation data. The estimated neutral winds and ion drifts are compared to the Horizontal Wind Model (HWM14) and SAMI3 “truth.” For both the quiet period on 25 August 2018 and subs7equent storm on 26 August, the EMPIRE estimation of ion drifts is better at low‐to‐mid geomagnetic latitudes with mean error up to 20 m/s. For the high latitudes (poleward of ±60° magnetic), the mean errors exceed 50 m/s with variances up to 200 m/s, and the relative errors are higher than the “truth.” At latitudes of ±87°, the large errors are attributed to a boundary effect. However, the neutral wind mean errors peak at 20 m/s at mid‐latitudes (40°–60° magnetic), with larger uncertainties, then converge to 0 approaching higher latitudes. By conducting this study, we define a method for obtaining the representation error covariance for future use of EMPIRE with SAMI3 as background.

This paper presents estimates of surface reflectivity (SR) over time of global navigation satellite system (GNSS) signals scattered from a partially frozen lake surface. A portable ground-based GNSS reflectometry sensor system that collects both scattered global positioning system L1 signals and independent validation data (lidar and camera) was deployed on the Lake Michigan waterfront in Chicago at a time when the lake surface was a mixture of ice and water. Lidar surface scans were merged with camera images and mapped, along with estimated reflection zones. For three satellites whose reflection points scan across ice and water over time, the relative SR and mean red intensity (differentiating ice from water) of camera pixels inside the first Fresnel zone were computed and shown to be correlated. This system concept will be used in the future for more complete mapping of phase changes of snow and ice in the cryosphere.

Abstract We present a binary hypothesis test for detecting clear sky in auroral all‐sky images based on single‐wavelength keograms. The coefficient of variation c, the ratio of the sample standard deviation to the mean over elevation angle along the meridian, is the test statistic. After image‐correcting keograms and excluding dark sky intervals, detection performance is compared to true conditions as determined by Advanced Very High Resolution Radiometer satellite imagery. The cloud mask, an index of cloud cover, is selected at the corresponding nearest time and location to the site of a meridian spectrograph at Poker Flat Research Range. With training data from 2014 to 2016, theoretical Rayleigh distributions fit to the histograms show a decision threshold of 0.40 could yield an accuracy of about 80%. Separately, we numerically compute the false alarm and missed detection statistics of the greenline 557.7 nm emission and of the redline 630.0 nm emission. We find a threshold of 0.25 for the greenline c maximizes the percent of events correctly identified at 76%. Applied to testing data from 2015 to 2017, the 0.25 threshold yields an accuracy of 68%. Diffuse aurora can have coefficient of variation around 0.2 to 0.5, which would be included by the numerical minimum, but partly excluded by the theoretical model obtained. Numerical results are a few percent worse for the redline emission.

Abstract In this work we enable the data assimilation algorithm Estimating Model Parameters from Ionospheric Reverse Engineering (EMPIRE) to estimate global neutral winds. EMPIRE corrects the ion drifts and neutral winds from background models using electron density in the F region derived mainly from total electron content measurements. The new EMPIRE basis functions for the neutral winds are vector spherical harmonics, enforcing the field to be smooth and continuous globally. Global basis functions allow us to estimate the horizontal wind vector from time‐varying plasma densities. The geomagnetic storm on 25 October 2011 is studied to investigate the global implementation of neutral wind estimation. During this storm, the estimates are also compared to those of the old method of estimation using power series and to Fabry‐Perot interferometer (FPI) measurements for validation at three different sites: Pisgah, Cariri, and Nasca. The new global estimation is 10%–20% closer to the FPI measurements than the model and than the old method for most of the viewing directions of the sites, in terms of root‐mean‐square residuals. In the northward direction at Pisgah the estimates using either the old or new implementation disagree with the model, and might be related to an insufficiently small temporal scale used in EMPIRE.

Electron density irregularities in the ionosphere modify the phase and amplitude of trans-ionospheric radio signals. We aim to characterize the spectral and morphological features of E- and F-region ionospheric irregularities likely to produce these fluctuations or “scintillations”. To characterize them, we use a three-dimensional radio wave propagation model—“Satellite-beacon Ionospheric scintillation Global Model of upper Atmosphere” (SIGMA), along with the scintillation measurements observed by a cluster of six Global Positioning System (GPS) receivers called Scintillation Auroral GPS Array (SAGA) at Poker Flat, AK. An inverse method is used to derive the parameters that describe the irregularities by estimating the best fit of model outputs to GPS observations. We analyze in detail one E-region and two F-region events during geomagnetically active times and determine the E- and F-region irregularity characteristics using two different spectral models as input to SIGMA. Our results from the spectral analysis show that the E-region irregularities are more elongated along the magnetic field lines with rod-shaped structures, while the F-region irregularities have wing-like structures with irregularities extending both along and across the magnetic field lines. We also found that the spectral index of the E-region event is less than the spectral index of the F-region events. Additionally, the spectral slope on the ground at higher frequencies is less than the spectral slope at irregularity height. This study describes distinctive morphological and spectral features of irregularities at E- and F-regions for a handful of cases performed using a full 3D propagation model coupled with GPS observations and inversion.

Abstract Geomagnetic storms produce significant electrodynamics at midlatitudes. Strong ion convection can affect thermospheric neutral wind motion. The converse is also true, such that both fields and winds can drive ionospheric plasma movement. This work adjusts a background modeled high‐latitude electrostatic potential to estimate the storm time electric field based on data‐derived plasma densities and measurements of neutral wind. Electron densities are derived from global navigation satellite system (GNSS) total electron content measurements using Ionospheric Data Assimilation Four‐Dimensional (IDA4D). These are input to Estimating Model Parameters from Ionospheric Reverse Engineering (EMPIRE) to estimate electric potential corrections to the background Weimer 2000 potential model. The EMPIRE basis functions for electric potential are spherical harmonics in dipole magnetic colatitude and longitude, enforced to be constant along a magnetic field line. For the 17 March 2015 storm, EMPIRE electric potential produces westward zonal ion drifts that more closely agree with incoherent scatter radar (ISR) measurements made at Millstone Hill than the background Weimer 2000 model alone, when electric potential and meridional neutral winds are both corrected. Additionally ingesting northward line‐of‐sight neutral wind measurements from a Fabry‐Perot interferometer at Millstone Hill makes little difference in the agreement between zonal ion drift predictions and measurements. Estimation of only electric potential reduces the agreement between the assimilated prediction of the field‐perpendicular zonal drifts and ISR measurements significantly.