Space Weather Applications - American Institute of … Space... · 2015-05-28 · Space Weather Applications David A. Vallado ... – Compare Orbital Mechanics content in •SMAD

AIAA

AIAA Rocky Mountain Section Seminar Series Whats New in the New SMAD

Space Weather Applications

David A. Vallado

MAY 9, 2015

Outline

Introduction Compare Orbital Mechanics content in

SMAD The New SMAD

Space Weather Application Long term solar cycle generation

Summary

Problem setup

Suppose you are tasked with determining how much fuel a Low Earth Orbiting satellite will require for a 15 year mission beginning in 2025? Where do you start?

SMAD applicable sections Introduction to Astrodynamics

Keplerian Motion Equations of Motion Classical Orbital Elements Ground Tracks Orbit Determination

Orbit Perturbations Third-body perturbations Non-spherical Earth Atmospheric Drag

Chang per rev

Solar Radiation Pressure

Orbit Maneuvering Coplanar maneuvers Plane Changes Rendezvous

Launch Windows Orbit Maintenance

Space Environment and Survivability Solar cycle Gravitational field and microgravity Upper Atmosphere

The New SMAD applicable sections Introduction to Astrodynamics

Keplerian Orbits Keplers Laws Orbital Elements and terminology

Orbits of the Moon and planets Satellite Orbit Terminology Orbit Perturbations

Non-spherical Earth Third-body perturbations Solar Radiation Pressure

Atmospheric Drag and Satellite decay Chang per rev

Cd discussion

Solar Flux cycles

Satellite decay

Specialized Orbits GEO Repeat GT Other

Orbit Maneuvering Coplanar maneuvers Plane Changes Rendezvous

Overview of Spacecraft Design Lifetime and Reliability Total Delta v

Some obvious Differences

The New SMAD More detail Better tie with Lagrange Planetary Equations

How perturbing forces affect the orbital elements

In the text Get More buttons

Web links for additional information Live calc button

Perform trade spaces in real-time

Some not so Obvious Differences Youll need to Consider Atmospheric Drag

Several aspects are important How atmospheric drag affects satellite orbits Solar cycles are important for long range planning

How much fuel is needed Look at the solar cycles

How do you estimate what the future cycles are supposed to be? Hathaway Schatten Polynomial

Can be quite far off Other?

Timing / magnitude issues

Remainder of the Presentation Discussion

Introduce atmospheric drag Factors influencing

Solution Approaches Atmospheric models

Input Data Solar Flux

Historical Prediction

Short, medium, long range

A new approach to estimating future solar cycles

Intro Atmospheric drag

Dominant perturbation force for Low Earth satellites Acts to retard motion

Over time, it eventually causes the satellite to re-enter atmosphere Related to the Suns solar cycle

About 11 year cycle Accurately estimating its effect

Requires numerous items! Next slide

Mass Maneuvers known? Area Model detail Attitude known? Materials cD

Accuracy

Satellite Parameters

Observations Quality Quantity Type, etc. Solution method Batch Least Squares Fit span correct? Kalman filter Process noise correct? Other Include Orbit Propagation

Orbit Propagation

Assumptions Indices F10.7, Kp, ap, etc. Data used in generation Satellite, ground, combination Specific to an orbit class Scope i.e. winds Fidelity?

Atmospheric Model Development

Indices Availability Accuracy Source Interpolate? Time lags Include Atmospheric Model Development

Atmospheric Model Use

Input State Accuracy? Solution method Integrator type? Propagator Force model fidelity Scope attitude, models, thrust, etc.? Include Satellite Parameters Atmospheric model choice Include Atmospheric Model Use Other

Orbit Determination

Atmospheric Model Development

Assumptions Indices

F10.7, Kp, ap, S10, Mg10, etc. Data used in generation

Satellite, ground, combination Availability

Specific to an orbital class Scope

i.e. winds Fidelity

10-15% in density Numerous models!

1960

1970

1980

1990

2000

ICAO/ARDC

USSA 62

2010

USSASupp

USSA 76

DENSELLOCKHEED

JACCHIA

JACCHIAWalker-BRUCE

GRAM(Justus)

GRAM-3

GRAM-90

J60(Jacchia)

J65

J70

J71(Jacchia-Roberts)

CIRA-65

J77

CIRA-72

CIRA-90

CIRA-61 Harris-Priester

MSIS-77

ESR04(von Zahn)

C(Kohnlein)

MSIS-83

HWM-93

GTM

CTIM

GOST-04

TGCM

TIGCM

TIEGCM

UCL(Fuller-Rowell,

Rees)

NCAR(Roble,

Dickinson)

MSIS UT(Long)

DTM-78(Barlier)

MSIS-86

MSIS-90

NRLMSIS-00

HWM-87

OG06(Hedin)

Total density from satellite drag

Temperature and composition from ground and in-situ

instruments

General Circulation models

1965

1975

1985

1995

MET-99

MET-88

M1, M2(Thullier)

AEROS(Kohnlein)

DCA*(Russian)

HASDM*

* Corrections to existing models

GRAM-99

DTM-94

DTM-03

GOST-84

JB-08*GRAM-07

GITM

Satellite Parameters

Size and shape affect how atmospheric drag affects a satellite Many sizes and shapes!

Higher Solar flux, greater impact on satellite orbits

GFZ

ERS-2

GPS

JasonGrace-a

ICESatCHAMP

Hubble Space Telescope

Using a single cD is analogous to using a two-body gravity field!

Various Satellite shapes

Wheres the errorOn this page?

Atmospheric Model Use

Indices Availability

Free? Span of data Classified?

Accuracy of the indices Source

Observed vs adjusted Prediction

Interpolate? Time lags

Includes all Atmospheric Model Development errors

Observed & Adjusted Solar Flux

Observed On the Earth at a particular time ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SOLAR_RADIO/FLUX/Penticton_Observed/

Adjusted Equivalent value at an average 1 AU distance ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SOLAR_RADIO/FLUX/Penticton_Adjusted/

Observations Algonquin Radio Observatory, Ottawa, Ontario

1947 to 1991 May 31 17:00 UTC observations

Dominion Radio Astrophysical Observatory, Penticton, BC (DRAO) 1991 to date 20:00 UTC observations

Lenhart Ottawa, at 17:00 UTC (adjusted)

ftp://ftp.ngdc.noaa.gov/STP/GEOMAGNETIC_DATA/INDICES/KP_AP/

210.7( )

10.7( ) 2adj

obs

F AUF

r -=

K L

Observed vs. Adjusted Solar Flux Data errors

Some inconsistencies from data bursts at measurement time 10-40 SFU

A Solution If DRAOadj Lenhart adj > 0.001

Lenhartadj = lenhartadj + DRAO-Lenhart

If DRAOobs - DRAOobs/corr > 0.5

DRAOobs = DRAOadj + DRAOobs - DRAOobs/corr

Corrected values in CelesTrak Should the bursts be included?

-40.0

-30.0

-20.0

-10.0

0.0

10.0

20.0

30.0

40.0

Jan-50 Jan-54 Jan-58 Jan-62 Jan-66 Jan-70 Jan-74 Jan-78 Jan-82 Jan-86 Jan-90 Jan-94 Jan-98 Jan-02 Jan-06 Jan-10 Jan-14

DRAO (obs) -Lenhart (adj)

data

DRAO (obs) -DRAO (adj)

data

DRAO (adj) -Lenhart (adj)

data

-20.0

-15.0

-10.0

-5.0

0.0

5.0

10.0

15.0

20.0


DRAO (obs) -Lenhart (adj)

data

DRAO (obs) -DRAO (adj)

data

DRAO (adj) -Lenhart (adj)

data

Space Weather Solar Flux and Geomagnetic Tracked for many years

Data to the 1930s CSSI Consolidated files

Produced since 2005 Data Integrity

Discrepancies Blank values, 0.0s translates as a 100-200 SFU error Corrected in CSSI files

Quality flag set to 4 as an indicator of CSSI correction Includes seasonal/solar cycle variations

Observed and adjusted to 1.0 AU values DRAO and Lenhart values

Atmospheric models use both

Space Weather Historical Data

0.0

50.0

100.0

150.0

200.0

250.0

300.0


Solar Cycle 23Solar Cycle 22Solar Cycle 21Solar Cycle 20Solar Cycle 19

F10.7center F10.7

avg apSolar Cycle 24

Space Weather: Predictions

Lots of Variability Short term

3, 27, and 45-day predictions Longer term

Current solar cycle Long term

Multiple solar cycles Schatten ESA PDFLAP Statistical sampling

Solar Flux Predictions: Short Term NOAA Predictions

27-day and 45-day (F10.7 and ap)

3-day 3-hourly Kp values off

significantly as well

Over time, the predictions average out

For a particular day, they can be very bad Mar 2015 for example

-10.0

-5.0

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

0 10 20 30 40 50

Diffe

renc

e (S

FU)

Days of Prediction

Average

StdDev

-40

-30

-20

-10

0

10

20

30

40

50

60

0 10 20 30 40 50

Diffe

renc

e

Days of Prediction

ap

F107

Solar Flux Predictions Shorter Term

Data differences

Min, Mid, and Max 30-50 SFU

Note timing of Cycle is off

Solar Flux Predictions Shorter Term

Early, Mid, and Late Also 30-50 SFU

differences

0.0

50.0

100.0

150.0

200.0

250.0

300.0

Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10 Jan-12

Solar Cycle 23, May 1996 - April 2000 - March 2008

Apr 95Early

Apr 95Late

Trend

Apr 95Mid

Solar Flux Predictions Long Term One or two Solar cycles in advance

Schatten predictions Trend Polynomial

Data differences One solar cycle

~150 SFU

Almost equal to the solar min-max difference!

50.0

100.0

150.0

200.0

250.0

300.0


Current Solar Cycle Solar Flux Prediction

0

50

100

150

200

250

300

07/24/98 01/14/04 07/06/09 12/27/14 06/18/20 12/09/25 06/01/31 11/21/36 05/14/42

Nov-2013Oct-2008

Dec-2002

Aug-2001

Mar-1997

Space Weather: Summary

So whats the impact on satellite positions over time? Study done a few years ago

Set various conditions to mimic the errors we saw in the predictions

Error Analysis Summary Topic Summary of Topic Effect Atmospheric Model Development Indices Which indices available when developed?

How were they used? Span of time indices available

Data used in generation Satellites were limited in atmospheric model development

Orbital Class Atmospheric model is only good for certain orbital classes

Fidelity Many studies, no conclusive winner 10-15% or more inaccuracy

Atmospheric Model Use Which Model Jacchia, MSIS, DTM, etc. 100 to 50,000 m in 4

days Implementation Code implements technical approach exactly? ?? Accuracy Inherent inaccuracy of the model 10-15% Data Indices Availability Publically available?

Time span available ??

Data Correction Observed vs adjusted 6,000 m in 4 days Daily vs Hourly values Frequency of updates to indices 3,000 m in 4 days Interpolation Interpolate or not

Which parameters to interpolate, or all? Method of interpolation

10-5,000 m in 4 days 3,000 in 4 days 10-20 m

Which Index to use Use ap or Kp preferentially? ?? Time lags 6.7 hours, other ?? Time of observation 1700 vs 2000 UTC for F10.7

1,000 m in 4 days

Averages Centered or trailing? 10,000 m in 4 Prediction Source, Schatten, ESA, Dan, Other?

Variability over time Short term accuracy

120 150 SFU (long) Varies 1-5 SFU (short)

Winds ?? All previous categories

Error Analysis Summary

Satellite Parameters Mass Accuracy 5 10%? Area Time varying? ?? cD Aerodynamics, gas dynamics, contium flow ?? Model detail Sphere, plate models, detailed CAD ?? Attitude known Consider a 10 m long 3 m diameter cylinder end vs

side 424% difference

Materials Interaction to atmosphere DSMC method? ?? Maneuvers Known or unknown

Magnitude, direction Can be very large 100s of km

Orbit Propagation Input state accuracy Example 1m initial error in the position 5,000 10,000 m in 7

days (extrapolated) Integrator type Small Force model fidelity Drag force is 10 100 km or more effect in 4 days Varies All previous categories Orbit Determination Obs Quality Not tested Obs Quantity Phillips Lab (1995) report 400 m at epoch Solution method BLS Not tested ?? Solution EKF Force Models Using a different atmospheric model 500-10000+ m in 7 days All previous categories

Space Weather: Predictions

But what data should I use for 2025-2040?? Current Schatten prediction goes to about 2045

Some distribution issues Some inaccuracies from prediction to prediction Lacks detail at daily level

Consider percentile approach Non physical technique Uses existing data

p. 29

How Does the Percentile Approach Work? Use entire data set

February 14, 1947 to date (ftp.ngdc.noaa.gov ) Determine average period (3954 days) and cycle representative start time

Entire data set (60 years) mapped into a single 3954-day time span Day 1 to 3954

Each day within 3954-day solar cycle period then has 3 or more representative groupings of (F10.7, F10.7 Bar, ap)

Trio of numbers (F10.7, F10.7 Bar and ap ) linked together Maintains linkage/correlation of values (F10.7, F10.7 Bar & ap) Allows historically-observed solar & geomagnetic activity variations and timing uncertainties

Select an interval to draw values for Single steps, orbital period, multiple periods Keep drawn values during that span

Percentile Approach

Historical data grouped by solar cycle

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Flu

x (S

FU)

Days into the Solar Cycle

19

20

21

22

23

24

Solar Cycle Percentile Avg StdDev Avg StdDev19 85 16.365 19.728 12.178 11.06120 30 11.798 13.659 8.452 7.81521 67 12.845 17.629 7.884 8.49722 68 15.431 20.123 11.038 11.11723 50 12.017 15.525 7.918 6.60924 4 3.190 7.489 1.934 3.44325 2526 45

Daily 81-day Avg

Previous Solar Cycles by Percentile Solar Cycle 19

Apr 1954 to Oct 1964

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)

Days into the Solar Cycle 19

19

Est %

Est 81 day Avg

Act 81 day Avg

End of cycle


Oct 1964 to Mar 1976

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


20

Est %

Est 81 day Avg

Act 81 day Avg

End of cycle


Mar 1976 to Jul 1986

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


21

Est %

Est 81 day Avg

Act 81 day Avg

End of cycle


Jul 1986 to Aug 1996

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


22

Est %

Est 81 day Avg

Act 81 day Avg

End of cycle


Aug 1996 to Nov 2008

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


23

Est %

Est 81 day Avg

Act 81 day Avg

End of cycle


Nov 2008 to xx

50

100

150

200

250

300

350

400

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


24

Est %

Est 81 day Avg

Act 81 day Avg

Current Time

Previous Solar Cycles by Percentile Solar Cycle 19 Geomagnetic

0

50

100

150

200

250

300

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Geo

mag

netic


19

Est avg ap


0

50

100

150

200

250

300

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


20

Est avg ap


0

50

100

150

200

250

300

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


21

Est avg ap


0

50

100

150

200

250

300

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


22

Est avg ap


0

50

100

150

200

250

300

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


23

Est avg ap


0

50

100

150

200

250

300

0 365 731 1096 1461 1826 2192 2557 2922 3287 3653 4018 4383

Sola

r Fl

ux (S

FU)


24

Est avg ap

Predicting Solar Cycles Not as easy as it would appear at first!

With only a few cycles, one could get the wrong idea Consequences could be dramatic if the fuel is not enough to

support the mission!

Solar Flux

Next 2 Solar Cycles using Percentiles

50

70

90

110

130

150

170

190

210

230

250

Nov-08 Nov-13 Nov-18 Nov-23 Nov-28 Nov-33 Nov-38 Nov-43

Sola

r Fl

ux (S

FU)


Cycle 2481 day Avg

Cycle 24NOAA 45 Day

Predict

Cycle 26Predictions

NOAAMonthly Predict

Cycle 25Predictions

Schatten 04/14 Schatten 11/08

Next 2 Solar Cycles using Percentiles Geomagnetic

0

20

40

60

80

100

120

Nov-08 Nov-13 Nov-18 Nov-23 Nov-28 Nov-33 Nov-38 Nov-43

Geo

mag

netic


Cycle 24ap Avg

NOAA 45 DayPredict

Cycle 26Predictions

NOAAMonthly Predict

Cycle 25Predictions

Schatten 04/14 Schatten 11/08

Summary

Modeling Atmospheric Drag is extremely difficult Texts go only so far Many factors affect accuracy

OD absorbs some of the errors but not all Largest contributors (roughly in decreasing order)

Predicted indices Biggest influence for long range studies as well!

Attitude Use of indices Observational data (qty, quality, geographic location, type, etc)

Short term effect Mathematical technique

Need to understand what the assumptions are!

So what about the original problem?

Questions?

Documents

Space Weather Applications - American Institute of … Space... · 2015-05-28 · Space Weather Applications David A. Vallado ... – Compare Orbital Mechanics content in •SMAD