Development of an advanced γ [gamma], hadron separation technique and application to particular γ-ray [gamma-ray] sources with H.E.S.S. [Elektronische Ressource] / put forward by Stefan Ohm

Sujets

• Astronomie

Informations

• Publié par: ruprecht-karls-universitat_heidelberg
• Publié le: 01 janvier 2010
• Nombre de lectures: 15
• Langue: English
• Poids de l'ouvrage: 13 Mo

Extrait

Dissertation
submitted to the
Combined Faculties for the Natural Sciences and for Mathematics
of the
Ruperto-Carola University of Heidelberg, Germany
for the degree of
Doctor of Natural Sciences
Put forward by
Dipl.-Phys.: Stefan Ohm
Born in: Sondershausen, Germany
Oral examination: 21 May 2010Development of an
Advanced γ/hadron separation
technique
and application to particular
γ-ray sources with H.E.S.S.
Referees: Prof. Dr. Werner Hofmann
Prof. Dr. Heinz V¨olkAbstract
The High Energy Stereoscopic system, H.E.S.S. is an array of four imaging atmospheric
ˇCerenkov telescopes, designed for the study of non-thermal phenomena in the universe
at very high energies (VHE). The sensitivity of telescope systems such as H.E.S.S. can
considerablybeimprovedbyabetterdiscriminationofthevastnumberofhadroniccosmic-
raybackgroundeventsagainsttheveryrareγ-raysignalevents. Inthiswork,anelaborated
discriminationtechnique–theBoosted Decision Treemethod–hasbeendeveloped andits
capabilities in terms of γ/hadron separation and improved sensitivity are demonstrated.
Inthesecondpart, theBDT methodisappliedto dataobtainedinobservations ofmassive
starformingenvironments, namelythecollidingwindbinaryη Carinae, themassivestellar
cluster Westerlund 1 and the Starburst galaxy NGC 253. An upper limit on the γ-ray
flux of the famous colliding wind binary system η Carinae is derived and, for the first
time, an alternative model for the high-energy emission observed by the Fermi satellite
is presented. The detection of very extended VHE γ-ray emission from the vicinity of
Westerlund 1 is reported and thorough spectral and morphological tests are presented.
Large parts of the resolved emission can be explained in a hadronic scenario, however, a
decisive conclusion can not be drawn. Finally, the BDT method allowed to detect the first
Starburst galaxy, namely NGC253, in VHE γ rays. Spectral and morphological results
are presented and suggest that large parts of the CR energy content are convectively and
diffusively transported into the intergalactic medium.
Kurzfassung
ˇH.E.S.S. ist ein System aus vier abbildenden Cerenkov Teleskopen und untersucht das
nicht-thermische Universum bei Energien im 100 GeV−100 TeV Bereich. Die Sensitivitat¨
von H.E.S.S. ist haupts¨achlich durch die effektive Unterdru¨ckung der enormen Anzahl an
hadronischen Untergrundereignissen in der Analyse bestimmt. In dieser Arbeit wurdeeine
neue Analysemethode, die sogenannte Boosted Decision Trees (BDT) Methode, auf Daten
und Simulationen angewendet und zeigt ein enormes Potential in der γ/Hadron Separa-
tion. Ausfuhrliche Tests mit Simulationen und Beobachtungsdaten realer Gammaquellen¨
demonstrieren die Eignung der BDT Methode und zeigen eine deutlich h¨ohere Signifikanz
im Vergleich zur Standardanalyse. Im zweiten Teil der Arbeit wurde die BDT Methode
auf Beobachtungsdaten von Regionen massiver Sternformation angewendet. Die Analyse
von Daten aus Richtung des beruhmten Binarsystems η Carinae ergab eine Flussober-¨ ¨
grenze, die im Vergleich mit Messungen des Fermisatelliten interessante Implikationen auf
Beschleunigungsprozessezulassen.SehrausgedehnteGammastrahlungsemissionwurdeaus
der Umgebung des massiven SternhaufensWesterlund 1 detektiert. Detaillierte systemati-
sche Tests hinsichtlich der Morphologie und des Energiespektrums wurden durchgefuehrt.
Ein Teil der Emission konnte in einem hadronischen Szenario erklart werden. Schliesslich¨ ¨
konnte mit Hilfe der BDT Analyse das erste Gammasignal von einer “Starburst” Galaxie
– NGC 253 – nachgewiesen werden.There is no all-seeing, all-loving god who keeps us free from harm; but atheism
is not a recipe for despair – I think the opposite. By disclaiming the idea of a
next life, we can take more excitement in this one. The here and now is not
somethingto beendured beforeeternal blissordamnation; thehereandnow is
all we have – an inspiration to make the most of it. So atheism is life-affirming
in a way religion can never be.
Look around you: nature demands our attention, begs us to explore, to ques-
tion. Religioncanprovideonlyfacile, ultimatelyunsatisfyinganswers. Science,
in constantly seeking real explanations, reveals the true majesty of our world
in all it’s complexity. Sometimes people say, “There must be more than just
this world, than just this life...” But how much more do you want?
We are going to die, and that makes us the lucky ones. Most people are never
going to die because they are never going to be born. The number of people
who could be here in my place outnumber the sand grains of Sahara. If you
think of all the different ways in which our genes could be permuted, you and
I are quite grotesquely lucky to be here – the number of events that had to
happen in order for you to exist, in order for me to exist...
We are privileged to be alive, and we should make the most of our time on this
world.
Richard DawkinsContents
List of Figures V
List of Tables VII
Preface 1
1 Detection of VHE γ rays with H.E.S.S. 5
1.1 Air showers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.1 Electromagnetic showers . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.1.2 Hadronic showers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
ˇ1.1.3 Atmospheric Cerenkov light from air showers . . . . . . . . . . . . . 9
ˇ1.2 H.E.S.S. instrument and Imaging Atmospheric Cerenkov technique . . . . . 10
ˇ1.2.1 Imaging Atmospheric Cerenkov technique . . . . . . . . . . . . . . . 10
1.2.2 The H.E.S.S. instrument . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 H.E.S.S. data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.1 Data taking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.2 Data preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.3 Event reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.4 Shower shape parameters . . . . . . . . . . . . . . . . . . . . . . . . 18
1.3.5 Event selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.3.6 Limitations of the shower shape cuts . . . . . . . . . . . . . . . . . . 20
2 Improved γ/hadron separation using a multivariate analysis technique 21
2.1 Parameters with γ/hadron separation potential . . . . . . . . . . . . . . . . 22
2.1.1 Classifying variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.2 Properties of classifying variables . . . . . . . . . . . . . . . . . . . . 24
2.2 Classification using Boosted Decision Trees . . . . . . . . . . . . . . . . . . 28
2.2.1 Basics of the Decision Tree algorithm . . . . . . . . . . . . . . . . . 29
2.2.2 The training procedure for a single tree . . . . . . . . . . . . . . . . 29
2.2.3 Boosting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.2.4 BDT settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3 Training and Evaluation of the BDT method . . . . . . . . . . . . . . . . . 31
2.3.1 Training sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3.2 Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3.3 Importance of training variables . . . . . . . . . . . . . . . . . . . . 34
2.3.4 BDT response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.4 Systematic studies using H.E.S.S. data . . . . . . . . . . . . . . . . . . . . . 36
2.4.1 Background estimation with H.E.S.S. . . . . . . . . . . . . . . . . . 36
2.4.2 Comparison between simulations and data . . . . . . . . . . . . . . . 37
2.5 Performance of a BDT classification . . . . . . . . . . . . . . . . . . . . . . 40
IContents
2.5.1 Energy reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5.2 Angular resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5.3 Effective detection area . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.6 Sensitivity of the BDT classifier . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.6.1 Separation power of ζ cuts . . . . . . . . . . . . . . . . . . . . . . . 51
2.6.2 Sensitivity of ζ cuts . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.6.3 Conclusions and future prospects . . . . . . . . . . . . . . . . . . . . 55
3 Particle acceleration in massive star-forming environments 57
3.1 VHE γ-ray emission from Colliding Wind Binaries . . . . . . . . . . . . . . 58
3.1.1 Geometrical model of a stellar wind collision region . . . . . . . . . . 59
3.1.2 Particle acceleration and HE/VHE γ-ray spectra . . . . . . . . . . . 60
3.1.3 Detectability with IACTs . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2 VHE γ-ray emission from young massive stellar clusters . . . . . . . . . . . 63
3.2.1 Formation of a superbubble . . . . . . . . . . . . . . . . . . . . . . . 63
3.2.2 Characteristics of a superbubble . . . . . . . . . . . . . . . . . . . . 64
3.2.3 Particle acceleration and γ-ray production inside a superbubble . . . 65
3.3 VHE γ-ray emission from Starburst galaxies . . . . . . . . . . . . . . . . . . 66
3.3.1 Characteristic regions . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.3.2 Cosmic-ray acceleration in SB galaxies . . . . . . . . . . . . . . . . . 68
3.3.3 Cosmic-ray energy-loss processes in SB galaxies . . . . . . . . . . . . 68