the aero-vista interactive spectrogram display: an

SSC21-V-01

The AERO-VISTA Interactive Spectrogram Display:An Original Software Solution for Scientific Operations of Twin 6U CubeSats

Andrew LangfordUniversity of Notre Dame

Department of Physics, Nieuwland Hall of Science; [email protected]

Mary Knapp, John Swoboda, Ryan Volz, Phil Erickson, Frank LindMIT Haystack Observatory

99 Millstone Rd, Westford, MA 01886

ABSTRACT

Led by MIT Haystack Observatory, the AERO and VISTA twin 6U CubeSats launching in 2022 willstudy the auroral regions of the Earth in the medium and high radio frequency bands. Novel vector sensors,developed by MIT Lincoln Laboratory, will produce hundreds of gigabytes each day while conducting inde-pendent and joint (interferometric) observations of auroral phenomena such as auroral kilometric radiation.With a limited number of downlink opportunities and volumes on the order of gigabytes, mission operationsrequire a tool to guide scientific processing and downlink decisions. The AERO-VISTA Interactive Spectro-gram (AVIS) Display, developed through open-source Dash Python and Digital RF libraries, fulfills this needthrough a dynamic user interface which allows mission scientists to perform preliminary data analysis andselection. Built-in data analysis tools such as channel selectors, data scaling, and axis sliders allow scientiststo critically assess preliminary data for further investigation. A variable resolution selection tool providesfor efficient and accurate selections of auroral phenomena viewed in the spectrogram. Standardized expor-tation procedures are integrated into uplink systems for in-flight-processing. The AVIS Display successfullydemonstrates an original UI for optimized data management between two spacecraft conducting scientificoperations.

Background

AERO-VISTA (A-V) is a twin 6U CubeSat mis-sion led by MIT Haystack Observatory with partnersat MIT Lincoln Laboratory, MIT AeroAstro, Moore-head State University (MSU), Dartmouth College,and Merrimack College. The mission, selectedby NASA’s CubeSat Launch Initiative (CSLI), isplanned to launch into a polar orbit no earlier than2022. The A-V science team will study the auro-ral regions of the Earth in the RF range from 400kHz - 5 MHz using the novel ‘vector sensor’ (VS)developed by MIT Lincoln Laboratory1. The A-Vmission is the first use of the VS in a space environ-ment2. The VS is unique in its capability to collecteach component of incoming electromagnetic radia-tion through orthogonal monopole, dipole, and loopantennas (Fig 1). This allows scientists to identifythe source location of radiation and characterize au-roral phenomena.

In addition to each spacecraft’s independent ob-servations of the aurora, the AERO and VISTAspacecraft will be conducting experimental interfer-

ometric observations of a test beacon as well as auro-ral radiation sources. In ground-based radio astron-omy, interferometric observations allow for greaterresolution and source characterization by collect-ing RF radiation simultaneously from telescopes lo-cated kilometers apart. Successfully demonstratinginterferometric capabilities in the space environmentopens opportunities for future space-based radio ob-servatories.

Figure 1: AERO and VISTA 6U Cube-Sats depicted with extended vector sensors.The vector sensors include orthogonal dipole,monopole and loop elements to collect eachcomponent of the electromagnetic radiation.

Langford 1 35th Annual Small Satellite Conference

AEROVISTA

Uplink commands for in-flight-processing

Data processing at MIT HaystackEvent identification and processing selection by global science team

Downlink Summary Data

Haystack, Morehead, and Near Earth Network dishes Ground operations at Morehead

Satellites in ~450-550 km, high inclination orbit

Interferometry 20-100 km apart

Graphic created by Drew Langford

Figure 2: Summary of AERO-VISTA mission ground operations involving the AVIS Display

Figure 3: RF Spectrogram illustrating auro-ral emmission phenomena similar to what thevector sensors will record and display on theAVIS Display dashboard3.

Auroral Emissions Radio Observer (AERO)Science Objectives

The AERO mission, led by PI Dr. Phil Erickson,aims to answer open questions surrounding Earth’saurora. RF electromagnetic radiation produced inthe auroral regions offers a probe into ionosphericconditions and processes1. Key emission types in-clude Auroral Kilometric Radiation (AKR), MediumFrequency Burst (MFB), and Auroral Roar and

Hiss4. A ground-based RF spectrogram of MF burstand Auroral Hiss is presented in Figure 3. Througha more advanced understanding of RF signatures ofplasma physics in planetary atmospheres, scientistswill be able to better predict extreme space weatherconditions and understand planetary environmentsin our solar system1.

In order to study these auroral phenomena, theAERO-VISTA spacecraft will record raw RF volt-age data during passes through the Earth’s northand south auroral zones. Summary spectrogram in-formation will be produced from these data collectsfor review on the ground to determine whether anyof the raw voltage data contains AKR, MFB, roar,or hiss and should therefore be post-processed anddownlinked.

Mission Operations Summary

The AERO-VISTA mission seeks to further un-derstanding of Earth’s ionosphere while pioneeringthe use of CubeSat technology and cooperative mis-sion architectures. Successful scientific operationsof two 6U CubeSats require software capabilitiesthat meet the demands of the mission. Software-based solutions for the A-V mission include theAERO VISTA Interactive Spectrogram (AVIS) Dis-play. Figure 2 illustrates a high-level overview of theA-V mission operations. Each component is summa-


rized as follows.

1. Observations of auroral radiation are simulta-neously obtained through vector sensors on-board the AERO and VISTA spacecraft.

2. Summary data files, including RF spectrogramand metadata, are downlinked through dishesat Moorehead State University and HaystackObservatory.

3. Summary data is processed through pipelineshosted at MIT Haystack facilities.

4. A-V team scientists use the AVIS Display forevent identification and select commands forin-flight-processing.

5. AVIS Display selections are integrated intogeneral up-link commands.

6. Dishes at MSU, Haystack, and NASA NENuplink commands for on-board data manage-ment and in-flight-processing.

Mission operations of two spacecraft producinghundreds of gigabytes of data each day with four tosix downlink opportunities each day presents a chal-lenge for data management and processing capabil-ities onboard the spacecraft. Further, mission oper-ations require the capability for responsive decision-making based on RF observations from the space-craft. The AVIS Display offers A-V scientists aninterface with observational data amid dynamic mis-sion operations. This interface is critical for efficientdecision-making and maximizing the science capa-bilities of the one-year mission.

AVIS Display Requirements

The AVIS Display is a project currently underdevelopment at MIT Haystack Observatory to meetdata management and scientific operation require-ments for the AERO-VISTA mission. The AVISDisplay project consists of two components, theAVIS software, and the AVIS dashboard. Generally,‘AVIS’ alone refers to the software portion, while the‘AVIS Display’ refers to the dashboard UI.

The project was initially guided under a set ofrequirements for an interactive data viewer and se-lector to accelerate the discovery of new auroral phe-nomena. The requirements set three main objec-tives.

1. Display dynamic and interactive RF spectro-gram plots from summary data files in DigitalRF format5.

2. Provide tools for user selection and identifi-cation of data by time and frequency rangeswithin the RF spectrograms.

3. Export user selections and commands in aspecified format for uplink to spacecraft. Theselected data is then post-processed onboardthe spacecraft for later downlink.

AVIS Display v1.0 meets each of these threeobjectives. This paper details the software imple-mented in order to achieve the original requirements.Further development of the AVIS software to im-prove efficiency and add capabilities is planned forSummer 2021.

AVIS Software

AVIS is an integrated piece of software writtenin Python 3 using Digital RF5, Dash6, Plotly7, Pan-das8, and Numpy9 open-source external libraries.The current capabilities of the software include a dy-namic and interactive RF spectrogram display, selec-tion and event identification, and command exporta-tion. A full block diagram of the software structurecan be found in Figure 4.

Open Source Implementations

Open-source libraries are used extensively in theAVIS software. Two packages, in particular, Dig-ital RF 5, and Plotly Dash7 set the foundation forthe software’s functionality. As the scope of smallsatellite missions increases, there is an opportunityfor open-source software to produce dynamic dash-boards to meet the specific needs of mission opera-tions.

Digital RF is an open-source disk storage andarchival format for digitally sampled voltage levelradio signals which uses the HDF5 standard andis maintained by MIT Haystack Observatory. ThePython library associated with Digital RF supportswriting and reading its formatted data-sets5. Thesummary data presented to the AVIS Display will bein Digital RF format. In the AVIS software, DigitalRF is used to extract RF spectrogram data, sam-pling rates, sampling times, and metadata from thedownlinked summary data files.

Plotly Dash is an open-source data science pack-age that allows for full-stack development of web ap-plications in Python, Julia, and R7. Given its abil-ity to scale for mission operations, Dash was chosenas the development platform for the AVIS Display.This capability is essential for the geographically


DigitalMetadata File

metareader.py

specmeta

metatime.pystartdate

datetimes

samplelist

reltimes

deltatimes

specpath

chandict

specmeta.py

specdata

times

assets

app.py

specinfo

AVIS DisplayAERO-VISTA Interactive Spectrogram Display

util.py

metaviewer

metaviewer

Drew LangfordMIT Haystack REU

Summer 2020

interactive functions data

AVISFile Selection

Data Selection Scaling

Selection Table

Spectrogram

Event Selection and Processing

AVIS Display

Figure 4: A block structure of the full-stack AVIS software

distributed A-V science team using the AVIS Dis-play around the the globe. Dash applications alsosupport the Plotly graphing library, which is capa-ble of producing dynamic spectrogram figures. TheDash Bootstrap Component library is also used ex-tensively for controlling dashboard layout10.

Back-end Modules

The AVIS software is built on four back-endmodules (metareader.py, metatime.py, specmeta.py,util.py) composing the metaviewer package whichaccesses the summary Digital Metadata file, formatdata, and builds Dash components implemented inthe dashboard. Each back-end module is summa-rized, and essential functions are described below.

metareader.py– metareader.py is the only mod-ule that reads straight from the Digital Metadatafile. This selection ensures that while various classinstances hold metadata information, they all callon the appropriate file. This is especially importantfor scaling operations to hundreds of summary files.The module is instructed to use its current direc-tory and redirect the path to the sample files folder

where the Digital Metadata files are stored. TheMetaReader class has several get functions that re-turn objects that are called on at higher points inthe software structure. Below are the objects whichcan be returned using get functions.

• get specpath – the file path of the spectro-gram data

• get specmeta – the digital metadata objectof the spectrogram data

• get specs all – a three dimensional arrayholding spectrogram time/frequency data foreach antenna channel

• get sample list – list of samples times-tamped in UNIX time

metatime.py – The metatime module completesthe necessary calculations and formatting of thesample timestamps using the AVIS metareader, Dig-ital RF util, and datetime modules. Each list re-turned in a get function of the metatime class iscomposed of datetime elements, except for reltime,


which is a list of relative time stamps since the be-ginning of the observation. Below are the objectswhich can be returned using get functions and aresuitable for use in the AVIS Display.

• get datetimes – list of datetime objects foreach sample in UTC

• get unixtime – list of unix time stamps foreach sample

• get startdate – starting date datetime objectof the samples

specmeta.py – The specmeta module holds theSpectra class, which takes many arguments to buildthe spectrogram as a Plotly graph object to be dis-played on the dashboard. The following are sampleinitialization arguments of the Spectra class.

• chan input – the name of the antenna chan-nel, ex. monopole

• time input – the time axis to display, ex.UTC

• cmap input – the color map of the spectro-gram, ex. Viridis

• scale – the scaling applied to the data, ex. log

• calibration – dictates if data values are rawor calibrated, ex. True

• t res – indicates the time resolution of the se-lection bounds

util.py –The util module serves as the bridge be-tween the back-end and front-end structure of theAVIS Display. The contents of util.py consist offunctions building Dash Core and Bootstrap Com-ponents, which utilize lists and objects importedthrough specmeta. The functions are categorizedby their supporting roles in the AVIS Display. Thecategories are the following.

• Basics – channels, time units, color maps

• Scaling – axis scaling and limits, calibration

• Spectrogram – graph object, loading spin-ner, file type

• Selection Table – table object, save anddelete buttons

• Data Selection – selection tool, resolutioncommands

• Event Selection and Processing Type -type selections and exportation

Frontend AVIS Display

While the AVIS back-end modules can be usedindependently of the total structure, the front-endscripts are dependent on the use of the metaviewerpackage. The app.py file imports each module fromthe metaviewer package to enable the util.py inter-active functions in Dash Callbacks.

assets folder – The assets folder is standard whenusing Plotly Dash. The directory holds the Cas-cading Style Sheet (CSS) script allowing for lay-out design of the dashboard. The predominant lay-out method in AVIS uses the Dash Bootstrap Lay-out Components. These consist of dbc.col, dbc.row,dbc.Form, and dbc.ListGroup components10. Thefolder also holds the image files displayed on thedashboard. The html.Img component directs the as-sets directory for the image file path, src.

app.py – The app.py file is commonly used inPlotly Dash as the script which initializes the dash-board on the local server. However, in dashboardexamples, the app.py file will play a varying role inthe dashboard layout and back-end structure. In thecase of the AVIS Display, the app.py module holdsboth the callback decorators/functions and the lay-out of the dashboard. In Plotly Dash, the callbackdecorators/functions create the interactive capabili-ties of the dashboard. Combining the callbacks andlayout into one script allows for the components anddashboard to be worked on simultaneously. How-ever, this does cause the script to be quite lengthyand must be well organized to be worked on effec-tively.

The organizational structure of app.py is as fol-lows:

• Imports

– Module Imports

– Util.py Component Imports

• Callbacks

– File Selection Callback

– Metadata Info Panel Callback

– Spectrogram Figure Callback

– Data Selection Callbacks

• App Layout

– Navbar row

∗ Dashboard title and logos

– Display bar row

∗ Data scaling, y-axis scale, color map,plot format, axis sliders


Figure 5: Screenshot of the AVIS Display’s dashboard in use. The user has selected a regionof RF spectrogram displaying the total RF signal. Note the selection bounds printed out andthe resolution grid has been increased.

– Dashboard Row

∗ Display controls

∗ Metadata info

∗ Spectrogram figure

∗ Selection tools

Dashboard Layout and Implementation

The AVIS Display dashboard layout is organizedinto four sections: Display Controls, Metadata Info,Display Bar, and Selection Tools. Figure 5 shows ascreenshot of the AVIS Display with a simulated vec-tor sensor RF spectrogram. The layout has been in-tentionally designed to follow a user’s natural work-flow. Frequently used features are located towardsthe middle of the screen, while entry and exit fea-tures are placed in the top left and bottom rightcorners respectively.

The user will begin interfacing with the dash-board through the display control section in the topleft corner. Here, the user can select the summaryfile directory and experiment, indicate time units,and select which channels to display on the spec-trogram. A recent feature incorporated the abilityfor the user to add multiple channels together in

the spectrogram. Each channel represents the volt-age measurement from an orthogonal dipole or loopantenna. The user can view the selected file’s meta-data info to identify characteristic information suchas the sample rate, sample length, and date.

In order to dynamically interface with the dis-played spectrogram, a user may use the display baralong the top of the dashboard. Features includedata scaling, color maps, and axis limits. The z-axis limit scale is useful for increasing the contrastof high or low amplitude signals.

Scientifically interesting RF signals are selectedusing a box-select tool built-in to the Plotly graphobject. The selection tool highlights points on ascatter plot grid which can be either shown or hid-den using the ‘Show Resolution’ slider. The selectionbox bounds are printed out and can be recorded us-ing the save ‘Save Selection’ button. The selectionsand their metadata are presented to an on-screen ta-ble and can be exported to a YAML file. The YAMLfile is easily integrated into uplink commands for in-flight processing on-board the spacecraft.

Science Operations Test

In November 2020, AVIS was used by the AERO-VISTA science team to conduct a Science Opera-


tions Test (SOT) exercise. The SOT simulated 48hours of AERO-VISTA science operations, includingdata downlink, review, selection, and forward obser-vation planning. Simulated vector sensor data wasgenerated for four auroral observation periods of 2-5minutes and processed into summary spectrogramsin Digital RF format11. The science team then usedthe AVIS Display dashboard to analyze the simu-lated data, discuss RF events to preserve for furtherprocessing and downlink, and save final selections for‘uplink’ to the spacecraft. This exercise generated alist of feature requests and suggestions for AVIS fromthe A-V science team, some of which have since beenimplemented into recent updates. More substantialupgrades for AVIS v2.0 are presented in the sectionbelow.

Future Additions

The AVIS Display has planned upgrades to im-prove efficiency and meet additional requirements.A current drawback of AVIS v1.0 is the high compu-tational burden of displaying/updating the spectro-gram plots. AVIS v2.0 will overcome this barrier bystoring RF spectrogram data in the server’s memoryrather than accessing the Digital RF file each timethe plot is refreshed.

AVIS v1.0 dashboard displays one RF spectro-gram and allows users to combine amplitudes ofchannels. However, future development will allowusers to display multiple spectrograms simultane-ously, including RF data from different summaryfiles. This addition will enhance the science team’scapability to make informed decisions based on aholistic picture of the summary data.

Additional features to the display include teleme-try and spacecraft position and orientation informa-tion. As the science team works to identify eventsfrom the summary data files, corresponding space-craft telemetry and visual orientation data will assistin the processes. Work on a Dash Plotly spacecrafttelemetry dashboard has already begun, and incor-porating features from the existing dashboard maybe possible with the shared use of Dash Plotly com-ponents.

Summary

The need for data management and analysis soft-ware during small satellite mission operations is notunique to the AERO-VISTA mission. With increas-ingly complex small satellite mission architectures,the AVIS Display offers a template for customized,

open-source software solutions. Further, the effec-tiveness of integrating a wide array of dashboard fea-tures creates opportunities for the addition of moreadvanced capabilities in the future.

Acknowledgements

The development of AVIS v1.0 was supportedthrough NSF Research Experiences for Undergrad-uates (REU) funding for MIT Haystack Observa-tory during Summer 2020. The VISTA project issupported by NASA opportunity NNH18ZDA001N-HTIDS in the Heliophysics Technology and In-strument Development for Science program. TheAERO project is supported by NASA opportunityNNH17ZDA001N-HTIDS in the Heliophysics Tech-nology and Instrument Development for Science pro-gram. Any opinions, findings, conclusions or recom-mendations expressed in this material are those ofthe author(s) and do not necessarily reflect the viewsof NASA.

References

[1] Philip Erickson et al. “AERO: Auroral Emis-sion Radio Observer”. In: Proceedings of theSmall Satellite Conference, Upcoming Mis-sions, SSC18-WKVI-04. Aug. 2018. url:https : / / digitalcommons . usu . edu /

smallsat/2018/all2018/453.

[2] Frank Lind et al. “AERO & VISTA: Demon-strating HF Radio Interferometry with VectorSensors”. In: Proceedings of the Small Satel-lite Conference, Upcoming Missions, SSC19-WKV-09. Aug. 2019. url: https : / /

digitalcommons.usu.edu/smallsat/2019/

all2019/96.

[3] J. LaBelle. “Spontaneous radio frequencyemissions from natural aurora”. In: NATOTechnical Report RTO-TR-IST-051 (2006).

[4] James LaBelle and Rudolf A. Treumann. “Au-roral Radio Emissions, 1. Hisses, Roars, andBursts”. In: 101.3 (Aug. 2002), pp. 295–440.doi: 10.1023/A:1020850022070.

[5] Ryan Volz, William C. Rideout, John Swo-boda, Juha P. Vierinen, and Frank D. Lind.Digital RF. Version 2.6.6. Feb. 12, 2021.url: https://github.com/MITHaystack/

digital_rf.


[6] Shammamah Hossain. “Visualization of Bioin-formatics Data with Dash Bio”. In: Proceed-ings of the 18th Python in Science Conference.Ed. by Chris Calloway, David Lippa, DillonNiederhut, and David Shupe. 2019, pp. 126–133. doi: 10.25080/Majora-7ddc1dd1-012.

[7] Plotly Technologies Inc. Collaborative datascience. Montreal, QC, 2015. url: https://plot.ly.

[8] The pandas development team. pandas-dev/pandas: Pandas. Version latest. Feb. 2020.doi: 10.5281/zenodo.3509134.

[9] Charles R. Harris et al. “Array programmingwith NumPy”. In: Nature 585.7825 (Sept.2020), pp. 357–362. doi: 10.1038/s41586-

020-2649-2.

[10] Boostrap. Components. 2020. url: https://dash-bootstrap-components.opensource.

faculty.ai/docs/components/alert/.

[11] A Marek, K Ammons, H Irshad, and J. Swo-boda. “Auroral Radio Emissions Simulationand Processing for AERO CubeSat Mission”.In: IEEE Aerospace Conference Proceedings.Big Sky, Montana, 2020.


the aero-vista interactive spectrogram display: an

Documents