Gamma-ray blazar spectra with H.E.S.S. II mono analysis : the case of PKS 2155−304 and PG 1553+113

A & A 600, A 89 (2017)

D O I: 10.1051/0004-6361/201629427

© E S O 2017

Astronomy

&

Astrophysics

Gamma-ray blazar spectra with H.E.S.S. II mono analysis:

The case of PKS 2155-304 and P G 1553+113

H.E.S.S. Collaboration*, H. A bdalla1, A. Abramowski2, F. A haronian3,4'5, F. A it Benkhali3, A. G. Akhperjanian6,5,t, T. A ndersson10, E. O. A nguner7, M. A rrieta15, P. Aubert23, M. Backes8, A. B alzer9, M. Barnard1, Y. Becherini10, J. Becker T jus11, D. B erge12, S. B ernhard13, K. Bernlohr3, R. Blackw ell14, M. Bottcher1, C. Boisson15, J. B olm ont16, P. Bordas3, F. Brun25, P. Brun17, M. Bryan9, T. Bulik18, M. Capasso27,

J. Carr19, S. Casanova20,3, M. Cerruti16, N. Chakraborty3, R. Chalm e-Calvet16, R. C. G. Chaves21, A. Chen22, J. Chevalier23, M. Chrćtien16, S. Colafrancesco22, G. Cologna24, B. Condon25, J. Conrad26, C. Couturier16, Y. Cui27, I. D. D avids1,8, B. Degrange28, C. D eil3, J. D evin17, P. deW ilt14, L. Dirson2, A. D jannati-Atai29, W. D om ainko3, A. Donath3, L. O ’C. D rury4, G. D ubus30, K. Dutson31, J. D yks32, T. Edwards3, K. Egberts33, P. Eger3, J.-P. Ernenwein20, S. Eschbach34, C. Farnier26,10, S. Fegan28, M. V. Fernandes2, A. Fiasson23, G. Fontaine28, A. Forster3, S. Funk34, M. FuBling35, S. G abici29, M. Gajdus7, Y. A. G allant17, T. G arrigoux1, G. Giavitto35, B. Giebels28, J. F. G licenstein18, D. Gottschall27,

A. G oyal36, M.-H. Grondin25, D. Hadasch13, J. Hahn3, M. H aupt35, J. H awkes14, G. H einzelmann2, G. H enri30, G. Herm ann3, O. H ervet15,42, A. H illert3, J. A. H inton3, W. H ofm ann3, C. Hoischen33, M. Holler28, D. Horns2, A. Ivascenko1, A. Jacholkowska16, M. Jam rozy36, M. Janiak32,

D. Jankowsky34, F. Jankowsky24, M. Jingo22, T. Jogler34, L. Jouvin29, I. Jung-Richardt34, M. A. Kastendieck2, K. K atarzynski37, U. Katz34, D. Kerszberg16, B. Khćlifi29, M. Kieffer16, J. King3, S. Klepser35, D. Klochkov27, W. K luzniak32, D. Kolitzus13, Nu. Komin22, K. Kosack18,

S. Krakau11, M. K raus34, F. Krayzel23, P. P. Kruger1, H. Laffon25, G. Lamanna23, J. Lau14, J.-P. Lees23, J. Lefaucheur15, V. L efranc18, A. Lemifere29, M. Lem oine-Goumard25, J.-P. Lenain16* , E. Leser33, T. Lohse7, M. Lorentz18, R. Liu3, R. López-Coto3, I. Lypova35, V. M arandon3,

A. M arcowith17, C. M ariaud28, R. M arx3, G. M aurin23, N. M axted14, M. M ayer7, P. J. M eintjes38, M. M eyer26, A. M. W. M itchell3, R. M oderski32, M. M oham ed24, L. M ohrm ann34, K. Mora26, E. M oulin18, T. Murach7, M. de Naurois28, F. Niederw anger13, J. N iemiec20, L. Oakes7, P. O ’Brien31, H. O daka3, S. O ttl13, S. O hm 35, M. O strowski36, I. Oya35, M. Padovani17, M. Panter3, R. D. Parsons3, M. Paz Arribas7,

N. W. Pekeur1, G. Pelletier30, C. Perennes16, P.-O. Petrucci30, B. Peyaud18, S. Pita29, H. Poon3, D. Prokhorov10, H. Prokoph10, G. Puhlhofer27, M. Punch29,10, A. Q uirrenbach24, S. Raab34, A. R eim er13, O. R eim er13, M. R enaud17, R. de los Reyes3, F. Rieger3,39, C. Romol©*, S. Rosier-Lees23, G. Rowell14, B. Rudak32, C. B. Rulten15, V. Sahakian6,5, D. Salek40, D. A. Sanchez23,*, A. Santangelo27, M. Sasaki27, R. Schlickeiser11, F. Schussler18, A. Schulz35, U. Schwanke7, S. Schwemmer24, M. Settim o16, A. S. Seyffert1, N. Shafi22, I. Shilon34, R. Sim oni9,

H. Sol15, F. Spanier1, G. Spengler26, F. Spies2, Ł. Stawarz36, R. Steenkam p8, C. Stegm ann33,35, F. Stinzing34© K. Stycz35, I. Sushch1, J.-P. Tavernet16, T. Tavernier29, A. M. Taylor4,*, R. Terrier29, L. Tibaldo3, D. Tiziani34, M. Tluczykont2, C. Trichard20, R. Tuffs3, Y. U chiyam a41,

D. J. van der W alt1, C. van Eldik34, B. van Soelen38, G. Vasileiadis17, J. Veh34, C. Venter1, A. Viana3, P. Vincent16, J. Vink9, F. Voisin14, H. J. Volk3, T. Vuillaume23, Z. W adiasingh1, S. J. W agner24, P. W agner7, R. M. W agner26, R. W hite3, A. W ierzcholska20, P. W illm ann34, A. W ornlein34, D. W outers18, R. Yang3, V. Zabalza31, D. Zaborov*,28, M. Zacharias24, A. A. Zdziarski32, A. Z ech15, F. Zefi28, A. Ziegler34,

N. Zyw ucka36, and

LAT Collaboration, M. Ackermann43, M. Ajello44, L. Baldini45,46, G. Barbiellini47,38, R. Bellazzini49, R. D. Blandford46, R. Bonino50,51, J. Bregeon52, P. Bruel28, R. Buehler43, G. A. Caliandro46,53, R. A. Cameron46, M. Caragiulo54,55, P. A. Caraveo56, E. Cavazzuti57, C. Cecchi58,59, J. Chiang46, G. Chiaro60, S. Ciprini57,58, J. Cohen-Tanugi52, F. Costanza55, S. Cutini57,58, F. D ’Am m ando61,62, F. de Palma45,63, R. Desiante64,40, N. Di Lalla49, M. Di M auro46, L. Di Venere54,55, B. D onaggio65, C. Favuzzi54,55, W. B. Focke46, P. Fusco54,55, F. G argano55, D. Gasparrini57,58,

N. G iglietto54,55, F. Giordano54,55, M. G iroletti61, L. G uillem ot66,67, S. G uiriec68,69, D. H oran28, G. Jóhannesson70, T. Kamae71, S. Kensei72, D. Kocevski69, S. Larsson73,74, J. Li75, F. Longo47,48, F. Loparco54,55, M. N. Lovellette76, P. Lubrano58, S. M aldera50, A. M anfreda49, M. N. M azziotta55, P. F. M ichelson46, T. Mizuno77, M. E. M onzani46, A. M orselli78, M. Negro50,51, E. N uss52, M. O rienti61, E. O rlando46,

D. Paneque79, J. S. Perkins69, M. Pesce-Rollins49,57, F. Piron52, G. Pivato49, T. A. Porter46, G. Principe80, S. Rainó54,55, M. Razzano49, D. Sim one55, E. J. Siskind81, F. Spada49, P. Spinelli54,55, J. B. Thayer46, D. F. Torres76,82, E. Torresi83, E. Troja69,84,

G. Vianello48, and K. S. Wood77 (Affiliations can be fo u n d after the references) Received 29 July 2016 / Accepted 1 D ecem ber 2016

ABSTRACT

Context. The addition of a 28 m Cherenkov telescope (CT5) to the H.E.S.S. array extended the experim ent’s sensitivity to lower energies. The low est energy threshold is obtained using m onoscopic analysis of data taken w ith CT5, providing access to gamm a-ray energies below 100 GeV for small zenith angle observations. Such an extension of the instrum ent’s energy range is particularly beneficial for studies o f active galactic nuclei with soft spectra, as expected for those at a redshift >0.5. The high-frequency peaked BL Lac objects PKS 2 1 5 5 -3 0 4 (z = 0.116) and PG 1553+113 (0.43 < z < 0.58) are among the brightest objects in the gamm a-ray sky, both showing clear signatures of gamm a-ray absorption at E > 100 GeV interpreted as being due to interactions w ith the extragalactic background light (EBL).

Aims. The aims o f this work are twofold: to dem onstrate the monoscopic analysis of CT5 data with a low energy threshold, and to obtain accurate m easurem ents of the spectral energy distributions (SED) o f PKS 2 1 5 5 -3 0 4 and PG 1553+113 near their SED peaks at energies »400 GeV.

Methods. M ultiple observational campaigns o f PKS 21 5 5 -3 0 4 and PG 1553+113 were conducted during 2013 and 2014 using the full H.E.S.S. II instrum ent (C T1-5). A m onoscopic analysis o f the data taken w ith the new CT5 telescope was developed along with an investigation into the systematic uncertainties on the spectral param eters w hich are derived from this analysis.

Results. Using the data from CT5, the energy spectra of PKS 21 5 5 -3 0 4 and PG 1553+113 were reconstructed down to conservative threshold energies of 80 GeV for PKS 2 155-304, which transits near zenith, and 110 GeV for the m ore northern PG 1553+113. The m easured spectra, well

* Corresponding author: H.E.S.S. and LAT Collaborations, e-mail: c o n t a c t .h e s s @ h e s s - e x p e r i m e n t .e u ł Deceased.

fitted in both cases by a log-parabola spectral model (with a 5.0ix statistical preference for non-zero curvature for PKS 2155-304 and 4.5ix for PG 1553+113), were found consistent with spectra derived from contemporaneous Fermi-LAT data, indicating a sharp break in the observed spectra of both sources at E « 100 GeV. When corrected for EBL absorption, the intrinsic H.E.S.S. II mono and Fermi-LAT spectrum of PKS 2155-304 was found to show significant curvature. For PG 1553+113, however, no significant detection of curvature in the intrinsic spectrum could be found within statistical and systematic uncertainties.

Key words galaxies: active - BL Lacertae objects: individual: PKS 2155-304 - BL Lacertae objects: individual: PG 1553+113 - gamma rays: galaxies

1. Introduction

T he very high energy (VH E, E > 100 G eV ) gam m a-ray exper­

im ent o f th e H igh E nergy S tereoscopic S ystem (H .E .S .S .) co n ­ sists o f five im aging atm ospheric C herenkov telescopes (IACTs) located in th e K hom as H ighland o f N am ib ia (23° 1 6 '1 8 " S, 16°3 0 '0 0 " E), 1835 m above sea level. F rom January 2004 to O ctober 2012, th e array w as operated as a four telescope in stru ­ m en t (H .E.S.S. p hase I). T he telescopes, C T 1 -4 , are arranged in a square form ation w ith a side length o f 120 m . E ach o f these telescopes has an effective m irror surface area o f 107 m 2, a field o f view o f 5° in diam eter, capable o f detecting cosm ic gam m a rays in the energy ran g e 0 .1 -1 0 0 TeV (A haronian e t al. 2006a) . In O ctober 2012 a fifth telescope, C T5, p laced at the centre o f the original square, started taking data. This set-up is referred to as H .E.S.S. p hase II, or H .E.S.S. II. W ith its effective m ir­

ror surface close to 600 m2 and a fast, finely p ixelated cam era (B o lm o n t e t al. 2014) , C T5 p otentially extends the energy range covered b y the array dow n to energies o f ~ 3 0 GeV.

In this study, w e focus on obtaining high statistic results w ith observations o f th e high-frequency p eaked B L L ac objects PKS 2 1 5 5 - 3 0 4 and P G 1553+113. T hese blazars are am ong the b rightest objects in the V H E gam m a-ray sky. F urtherm ore, the spectra o f b oth these blazars exhibit signatures o f gam m a-ray absorption a t energies E ~ 100 GeV, due to interactions w ith the extragalactic b ackground light (EB L).

PKS 2 1 5 5 - 3 0 4 is a high-frequency p eaked B L L ac (H B L) object at z = 0.116 (G anguly et al. 2 0 1 3 ; F arina e t a l. 2016) . This source is located in a galaxy p o o r cluster (F alom o et al.

1993) and the h o st galaxy is resolved (K otilainen e t al. 1998) . It w as first discovered as a high energy em itter by the H E A O 1 X -ray satellite (Griffiths et al. 1979; Schw artz et al.

1979) . G am m a-ray em ission in the energy ran g e 30 M eV to 10 G eV w as d etected from this blazar b y th e E G R E T instrum ent on b o ard th e C om pton G am m a R ay O bservatory (V estrand et al.

1995) . T he first detection in the V H E ran g e w as attained in 1996 by the U niversity o f D urham M ark 6 Telescope, w ith a statis­

tical significance o f 6.8^ (C hadw ick et al. 1999). Starting from 2002 th e source w as regularly observed w ith H .E .S.S., w ith the first detection b ased on the 2002 data subsequently p ublished w ith ju s t one telescope o f H .E .S.S. p hase I (A haronian et al.

2005) . A fter com pletion o f the array, this source w as detected in stereoscopic m ode in 2003 w ith high significance (> 1 0 0 ^ ) at energies greater than 160 G eV (A haronian e t al. 2005) . Strong flux variability w ith m ultiple episodes o f extrem e flaring a c ­ tivity in the V H E b an d w ere rep o rted (A haronian e ta l. 2 0 0 7 ; H .E.S.S. C ollaboration et al. 2 0 1 0 ; A leksic e t al. 2 012a) . A p h o ­ ton index ( r , describes th e spectral shape o f the p h oton en ­ ergy distribution, d N /d E <x E - r .) o f 3.53 ± 0.06stat ± 0.10syst w as obtained from analysis o f observations during a low flux state (2 0 0 5 -2 0 0 7 ) above 200 G eV (H .E.S.S. C ollaboration et al.

2010) . F or average and high flux states the p resence o f curvature or a cut-off w as favoured from the spectral fit analysis carried out (H .E.S.S. C ollaboration et al. 2010) .

T he H B L o bject P G 1553+113 w as first announced as a V H E gam m a-ray source b y H .E .S.S. (A haronian e t al. 2006b) and independently and alm ost sim ultaneously confirm ed by M A G IC using observations from 2005 (A lbert e t al. 2007) . The H .E.S.S. I m easurem ents (A haronian et al. 2008) yielded a p h o ­ ton index r = 4.5 ± 0.3stat ± 0.1syst above 225 GeV. A t high ener­

gies (HE, 100 M eV < E < 300 G eV ) the source was detected by Ferm i-LA T w ith a p h oton index o f 1.68 ± 0.03 (A bdo et al. 20 0 9 , 2010) , m aking P G 1553+113 an active galactic nucleus (AGN) w ith one o f the largest H E -V H E spectral breaks observed and a hint for long-term gam m a-ray flux oscillation (A ckerm ann e t al.

2015) . T he red sh ift o f P G 1553+113 is constrained by U V obser­

vations to the ran g e 0.43 < z < 0.58 (D anforth e t al. 2010) . The first upper-lim its o f z < 0.69 (pre-Ferm i-LA T) M azin & G oebel (2007) and m o re recently (post-Ferm i-LA T) z < 0.61 on the source red sh ift have been obtained A liu et al. (2015) using TeV d ata and o f z < 0.53 by B iteau & W illiam s (2015) using also G eV data. A ssum ing th at th e difference in spectral indices b e ­ tw een the H E and V H E regim es is im printed by the atten u a­

tion b y the extragalactic b ackground light, the red sh ift was co n ­ strained to th e ran g e z = 0.49 ± 0.04 (A bram ow ski et al. 2015) .

This p ap er reports on the first observations o f PKS 2 1 5 5 -3 0 4 and P G 1553+113 conducted in 2013 an d 2014 using the H .E.S.S. II instrum ent (CT5) in m onoscopic m ode. A descrip ­ tion o f th e analysis for both A G N s, using d ata from this in stru ­ m ent, is provided. S ystem atic errors associated w ith our results are also estim ated. P articular em phasis is p laced on th e spec­

tral m easurem ents at low energies and th eir connection w ith the Ferm i-LA T m easurem ents. U sing th e H .E.S.S. II m ono and Ferm i-LA T results, th e im plications on intrinsic source spectrum are considered.

2. The H.E.S.S. II experiment

T he H .E.S.S. II experim ent is the first hybrid C herenkov in stru ­ m e n t an d has the ability to take data in different m odes. The H .E.S.S. II system triggers on events detected either by C T5 only (m ono) or b y any com bination o f tw o or m o re telescopes (stereo, C T5 plus a t least one o f C T 1 -4 , o r at least tw o o f C T 1 -4 ). The field o f view o f C T5 is 3.2° in diam eter, sm aller than th at for C T 1 -4 . C onsequently, n o t all stereo triggers include C T5. The standard observation m o d e o f H .E .S.S. II is to co llect both m ono and stereo events during the sam e observation run.

T he analysis o f C T 1 -5 stereo data provides a low er energy threshold, b etter hadron rejection and b etter angular resolution than w ith C T 1 -4 only. T he analysis o f H .E.S.S. II m ono events p o tentially provides a factor o f approxim ately fo u r low er energy threshold than C T 1 -5 stereo. H owever, th e absence o f stereo­

scopic constraints m akes the rejection o f hadronic events m o re difficult, leading to a larger b ackground and reduced signal-to- b ackground ratio at th e analysis level. T he low energy threshold o f H .E .S.S. II m ono im plies high event rates, an d thus sm all sta­

tistical uncertainties on the background, w hich leads to tig h t re ­ quirem ents for the accuracy o f b ackground subtraction. T he an ­ gular reconstruction o f the m onoscopic analysis is significantly

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

less p recise than th at obtained in the stereoscopic m ode, leading to a reduction o f the sensitivity for p oint-like sources.

N evertheless, the H .E .S.S. II m ono analysis provides new opportunities to p robe astronom y a t energies < 100 G eV for southern sources, w hich are com plem entary to satellite experi­

m ents (e.g. F erm i L arge A rea Telescope, LAT) and to northern hem isphere facilities such as M A G IC and V ERITA S w hich can detect northern sources below 100 G eV (A leksic et al. 2 0 1 5 a;

A beysekara e t al. 2015) . T he low energy threshold pro v id ed by H .E.S.S. II m ono is, consequently, particularly beneficial for studies o f b rig h t variable objects such as gam m a-ray bursts and A G N s o u t to high redshifts (z > 0.5), along w ith associated spec­

tral features introduced into th e spectra through gam m a-ray in ­ teractions w ith th e extragalactic b ackground light (EBL).

T he full perform ance characterization o f the C T 1 -5 system w ill be p rovided in a forthcom ing publication.

3. H.E.S.S. II mono observations and analysis 3.1. H .E .S .S . II o b se rv a tio n s

PKS 2 1 5 5 - 3 0 4 was m o n ito red w ith H .E.S.S. II regularly for tw o consecutive years: in 2013 (from Apr. 21 to Nov.

5, 2013, M JD 5 6 4 0 3 - 5 6 6 0 1 ); and 2014 (M ay 2 8 -Ju n . 9, 2014, M JD 56 8 0 5 -5 6 817). P G 1553+113 was observed w ith H .E.S.S. II betw een M ay 29 and A ug. 9, 2013 (M JD 5 6 4 4 1 ­ 56 513). M o st o f th e observations w ere taken using the full H .E.S.S. II array. This p ap e r only reports on th e m onoscopic analysis o f this data, w hich provides th e low est achievable en ­ ergy threshold.

H .E.S.S. data taking is organised in 28 m in blocks, called runs. O bservations are usually taken in w o b b le m ode, w ith the cam era’s field o f view centred at a 0.5° o r 0.7° offset from the source position, in either direction along the right-ascension or declination axis. O nly runs for w hich the source position is lo ­ cated betw een 0.35° an d 1.2° off-axis from cam era centre are used in th e presen t analysis. R uns w ith n on-standard w obble offests w ere taken during the com m issioning p hase to assess the p erform ance o f the instrum ent. This is to ensure th at the source is w ell w ithin th e field o f view and allow b ackground subtrac­

tion using the reflected-region b ackground m eth o d (B erge et al.

2007) .

3.2. D ata quality se le c tio n

To ensure the quality o f the A G N data sets for the H .E.S.S. II m ono analysis the several ru n quality criteria w ere applied.

- S table clear sky conditions according to the telescope r a ­ diom eters. W e use the narrow field-of-view radiom eters in ­ stalled on the C T 1 -4 telescopes, requesting radiom eter te m ­ p erature to b e less than - 2 0 °C and stable during th e run w ithin ±3 °C;

- R elative hum idity <90% ;

- R un d uration > 5 m in and live tim e fraction >90% . A run m ay b e interrupted due to an autom ated target-of-opportunity o b ­ servation o f a transient source, deteriorating w eather co n d i­

tions, o r a technical issue;

- A t least 90% o f pixels in C T5 are active (pixels can b e te m ­ porarily sw itched off due to a star in the field o f view o r r e ­ m oved from the data due to b ad calibration);

- C T5 trigger in standard configuration p ixel/sector threshold

= 4/2.5, see A haronian e t al. (2006a) for a definition o f the trigger pattern;

- C T5 trigger rate betw een 1200 and 3000 H z (its nom inal value depends on the observed field o f view an d zenith angle) and stable w ithin ±10% during a run;

- T elescope tracking functioning norm ally;

3.3. D ata a n a ly sis

T he d ata sets w ere p rocessed w ith the standard H .E .S.S. analysis softw are using the M odel reconstruction (de N aurois & R olland 2009) w hich was recently adapted to w ork w ith m onoscopic events (H oller e t al. 2015) . T he M odel reconstruction perform s a likelihood fit o f the air show er im age to a sem i-analytical m odel o f an average gam m a-ray show er p aram eterised as a function o f energy, prim ary interaction depth, im p act distance and d i­

rection. G am m a-like candidate events are selected b ased on the value o f the goodness-of-fit variable and the reconstructed p ri­

m ary interaction depth. In addition, events w ith an estim ated er­

ror in direction reconstruction >0.3° are rejected. T he low energy threshold is controlled w ith a dedicated variable NSB G o o d n e ss, w hich characterises the likelihood o f accidentally triggering on fluctuations due to th e n ig h t sky background. Two cu t configu­

rations w ere defined for this analysis, loose and standard, w ith different settings for the NSB G o o d n e ss cut. L oose cuts provide the low est energy threshold, b u t m a y lead to a significant level o f system atic errors in the b ackground subtraction w hen applied to high statistics datasets. Standard cuts provide a b etter control over the back g ro u n d subtraction a t the co st o f increased thresh­

old. T he event selection cuts, except for the NSB G o o d n e ss cut, w ere optim ised to m ax im ise the discovery p otential for a poin t source w ith a photon index o f 3.0 observed at a zenith angle o f 18° for 5 h. T he optim ized analysis provides an angular reso lu ­ tion o f ~ 0.15° (68% co ntainm ent radius) at 100 G eV and energy resolution o f ~ 25% . F or p h oton indices harder than 3.0, standard cuts provide a b etter sensitivity than loose cuts.

T he b ackground subtraction is p erform ed using the stan­

dard algorithm s used in H .E .S .S .- the ring b ackground m ethod (for sky m aps) an d the reflected-region b ackground m eth o d (B erge e t al. 20 0 7 , w ith m u ltip le off-source regions, for spec­

tral m easurem ents). T he ring b ackground m e th o d uses a zenith- d ependent tw o-dim ensional acceptance m odel, an inner ring ra ­ dius o f 0.3° an d o uter radius o f 0.6°, an d top-hat sm oothing radius o f 0.1°. T he acceptance m odel, w hich describes the o b ­ served distribution o f b ackground events in th e cam e ra’s field o f view in absence o f gam m a-ray sources, is obtained from the data itself, using b ackground events outside o f a radius o f 0.3° from any k n o w n V H E gam m a-ray source (for this analysis, PKS 2 1 5 5 -3 0 4 and P G 1553+113). T he reflected-region b ac k ­ ground m ethod uses an on-source region radius o f 0.122°, w hich corresponds to an angular distance cu t 92 < 0.015 deg2. The nu m b er o f off-source regions was adjusted on a run-by-run b a ­ sis so as to alw ays use the m axim um possible n um ber o f them , given the w obble angle. F or instance, for a w obble angle o f 0.5°

nin e off-source regions w ere used. A sim ple acceptance m odel, w hich only corrects for linear gradients in the acceptance, is used w ith this m ethod. T he significance o f the excess after b ac k ­ ground subtraction is determ ined using the m eth o d described by L i & M a ( 1983) . S pectral m easurem ents are obtained u s­

ing the forw ard folding technique (Piron e t al. 2001), applied to the excess events observed w ith th e reflected-region background m ethod. T he energy threshold for th e spectral fit is defined as the energy at w hich the effective area reaches 15% o f its m axim um value, in line w ith the definition previously adopted in H .E.S.S.

analysis (H .E.S.S. C ollaboration et al. 2014a) . Such a definition ensures th at the system atic uncertainties in th e analysis are k ep t

Fig. 1. Top: (left) excess map of events observed in the direction of PKS 2 1 5 5 -3 0 4 using the H.E.S.S. II mono analysis (2013-2014 data).

The inset represents the point spread function o f the instrum ent obtained from simulations. The source position is indicated by a black dot. Right:

significance distribution that corresponds to the excess map (black his­

togram). The distribution obtained by excluding a circular region of 0.3°

radius around the source is shown in red; the results o f a G aussian fit to this distribution are also shown. Bottom: distribution of 92 (squared an­

gular distance to PKS 2155-3 0 4 ) for gamma-like events obtained with the H.E.S.S. II mono analysis (filled histogram ) in comparison w ith the norm alised 92 distribution for off-source regions (black points). The ver­

tical dashed line shows the lim it o f the on-source region. The energy threshold for this analysis is » 80 GeV.

under control. T he H .E .S.S. II m ono analysis was applied to all events th at include C T5 data (ignoring inform ation from C T 1 -4 ).

4. Results 4.1. P K S 2 1 5 5 - 3 0 4

T he PKS 2 1 5 5 - 3 0 4 d ata set, filtered as explained in Sect. 3.2, com prises 138 runs. T he total live tim e o f this d ata set is 56.0 h, 43.7 h taken in 2013 and 12.3 h taken in 2014. D uring these o b ­ servations, the source zenith angle ran g e d from 7° to 60°, w ith a m edian value o f 16°. This d ata set w as analysed using stan­

dard cuts as d escribed in Sect. 3 .3 . T he b ackground event counts obtained for the off-source regions in each ru n (in the reflected- region b ackground analysis) w ere used to perform an additional test o f the uniform ity o f the cam era acceptance. This was done using a likelihood ratio test (LLRT), w ith the b aseline hyp o th e­

sis that the event counts observed in all off-source regions com e from the sam e P oisson distribution, and a nested m odel allow ing for different m ean values in each region. T he results o f this test w ere consistent w ith an axially-sym m etric cam era acceptance.

T he sky m ap obtained for PKS 2 1 5 5 -3 0 4 using the H .E.S.S. II m ono analysis is show n in the top-left panel o f Fig. 1.

T he analysis found that the source is detected w ith a significance o f » 4 2t , w ith » 4 0 0 0 excess events. T he corresponding distribu­

tion o f the excess significance o f all skym ap bins is show n in the top-right panel o f Fig. 1. T he w idth o f the observed excess is approxim ately com patible w ith th e sim ulated p oin t spread fu n c­

tion (PSF; show n in the inset on Fig. 1). T he best-fit position o f the excess is found 3 2 " ± 10£(at from the target position.

Fig. 2. Top: PKS 21 5 5 -3 0 4 excess map (left) and significance distri­

bution (right) for events w ith reconstructed energy E < 100 GeV (H.E.S.S. II mono analysis, 2013-2014 data). Bottom: distribution of 92 (squared angular distance to PKS 2155-3 0 4 ) for gam m a-like events.

This offset can b e attributed to the system atic errors on the te le­

scope pointing. O utside th e exclusion radius o f 0.3° the signif­

icance distribution w as found to b e w ell fit by a G aussian w ith t = 1. 149 ± 0.004. This resu lt indicates th e presence o f a system ­ atic effect in b ackground subtraction, w hose T syst corresponds to ab o u t 57% o f the statistical errors (T stat equal one by co n ­ struction). W e here assum e th at the errors add in quadrature. A value o f t = ^ 1 + ^2yst > w ould then indicate th e do m i­

nance o f b ackground subtraction errors. This effectively reduces the observed excess significance from 4 2 t to » 3 6 t 1. This sys­

tem atic effect is currently u nder investigation as part o f a larger effort to understand the m ono analysis perform ance. R epeating the analysis using only events w ith reconstructed energy below 100 G eV leads to a 1 0 t ( 7 .3 t ) significance at th e position o f PKS 2 1 5 5 -3 0 4 in th e skym ap (Fig. 2 ) . T he significance distri­

bution outside th e exclusion region has t = 1.374 ± 0.005, in d i­

cating that the b ackground subtraction errors are slightly sm aller than th e statistical errors. Thus the source is confidently detected at E < 100 GeV.

T he distribution o f 92, the square o f the angular difference betw een the reconstructed show er position and the source p o si­

tion, is show n in the b ottom pan el o f F ig. 1 (filled histogram ).

A 4 3t excess over the b ackground (black crosses) is observed w ithin th e on-source region (92 < 0.015 deg2).

T he reconstructed spectrum o f PK S 2 1 5 5 - 3 0 4 obtained for 2013, and each o f the observation years (2013 and 2014), is show n in F igs. 3 and 4 , respectively. F or the full d ata set (2013+2014), a lo g -parabola m odel, d N /d E =

¢0 ( E / E 0) -r-/2l°8(E/E0), b etter fits the d ata w ith resp ect to a sim ­ ple pow er-law m odel w ith a log-likelihood ratio o f 25 (i.e. 5 t ) . T he flux norm alisation is found to b e ¢0 = (5.11 ± 0.15stat) x -1 a t a decorrelation 2 energy E 0 = 156 GeV, 10-10 c m -2 s-1 TeV

1 From this point forward, significance values are not corrected for this effect, with the corrected values being quoted within brackets im m edi­

ately proceeding these uncorrected values.

2 For the log-parabola model, the decorrelation energy is the energy where the error on the flux is the smallest, that is w here the confidence band butterfly is the narrowest in the graphical representation.

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

Fig. 3. Energy spectrum of PKS 2155-304 obtained from the H.E.S.S. II mono analysis (2013 data, shown by blue circles with confi­

dence band) in comparison with the contemporaneous Fermi-LAT data with an energy threshold of 0.1 GeV (red triangles and confidence band), 10 GeV (green band), and 50 GeV (purple band) and contem­

poraneous CT1-4 data (grey squares). In all cases the confidence bands represent the 1ix region. The right-hand y-axis shows the equivalent isotropic luminosity (not corrected for beaming or EBL absorption).

The inset compares the H.E.S.S. confidence band with the Fermi-LAT catalogue data (3FGL, 1FHL and 2FHL, see Sect. 4.4.2).

Fig. 4. SED of PKS 2155-304 separated into the 2013 and 2014 obser­

vation periods. Both the H.E.S.S. II mono and contemporaneous Fermi- LAT data are shown. The bands represent the 1ix confidence region.

w ith a p h oton index r = 2.63 ± 0.07stat and a curvature param eter P = 0.24 ± 0.06stat. T he spectral d ata points (blue filled circles) cover the energy ran g e from 80 G eV to 1.2 TeV (not including upper lim its). T he spectral param eters obtained for the 2013 and 2014 d ata sets are given in Table 1. T he isotropic lum inosity that corresponds to th e m easu red S E D is show n b y th e additional y-axis on the rig h t-h an d side o f the SED plots.

4.2. PG 1553+ 113

T he P G 1553+113 d ata set, filtered as explained in Sect. 3.2, com prises 39 runs (16.8 h live tim e), w hich w ere analysed u s­

ing loose cuts as described in Sect. 3 .3 . This analysis configura­

tion, providing low er energy threshold than standard cuts, is w ell suited fo r b right soft-spectrum sources, such as P G 1553+113.

Fig. 5. Top: (Left) excess map of events observed in the direction of PG 1553+113 using the H.E.S.S. II mono analysis (16.8 h live time).

The source position is indicated by a black dot. Right: significance dis­

tribution that corresponds to the excess map. The m eaning of the his­

tograms and statistics data is the sam e as in Fig. 1. Bottom: 92 distribu­

tion for PG 1553+113. The meaning of the data shown is the sam e as in Fig. 1. The vertical dashed line shows the lim it of the on-source region.

The energy threshold for this analysis is »400 GeV.

D uring the observations, the source zenith angle ranged betw een 33° and 4 0 ° , w ith a m ean value o f 35°. T he sky m ap obtained for P G 1553+113 using the H .E.S.S. II m ono analysis is show n in th e to p-left panel o f Fig. 5. This analysis found th at th e source is detected w ith a statistical significance o f 27<r (2 1 ^ ), w ith

» 2 5 0 0 excess events.

T he best-fit p osition o f the excess is found to b e 3 6 " ± 12'^at from the target position, this shift is attributed to the system ­ atic errors on the telescope pointing. T he w idth o f the observed excess is com patible w ith the sim ulated P SF w ithin a 10% sys­

tem atic uncertainty on th e P S F w idth.

T he significance distribution in the region outside o f the 0.3°

exclusion radius is co nsistent w ith a n orm al distribution (top- rig h t panel o f F ig. 5) . T he sam e holds true w hen the analysis is rep eated in only a low energy bin, w ith a reconstructed energy ran g e o f 1 0 0 -1 3 6 GeV. W ithin this energy bin, the source is d e­

tected w ith a 1 0 ^ (8 .2 ^ ) significance (Fig. 6 ) . T he significance distribution outside the exclusion region has <r = 1.219 ± 0.005 and 1.288 ± 0.005, for the full energy range and the first energy bin, respectively, indicating presence o f b ackground subtraction errors at a level sm aller than the statistical errors.

T he 92 distribution is show n in th e b ottom panel o f Fig. 5 . A 2 7 <r (2 1 ^ ) excess over the b ackground is observed w ithin the on-source region (92 < 0.015 deg2). T he reconstructed spectrum , w ith a threshold o f 110 GeV, is found to be w ell fit b y a log- parabola (w ith a L L R T o f 20 over the pow er-law m odel, Fig. 7 ), w ith a photon index r = 2.95 ± 0.23stat at decorrelation energy E0 = 141 GeV, curvature p aram eter P = 1.04 ± 0.31stat, and differential flux $ 0 = (1.48 ± 0.07stat) x 10-9 c m -2 s-1 TeV -1 at E0. T he spectral d ata points (blue filled circles) cover the energy ran g e from 110 G eV to 550 G eV (not including u pper lim its).

Table 1. Spectral analysis results of H.E.S.S. II mono observations.

Source Year M JD L ivetim e

[h]

Eq [GeV]

¢0

[10-9 c m -2 s-1 TeV -1]

r

P

PK S 2 1 5 5 -3 0 4 2013 5 6 4 0 3 -5 6 6 0 1 43.7 151 0 .530 ± 0.018stat 2.65 ± 0.09stat 0.22 ± 0.07stat 2014 56 8 0 5 -5 6 817 12.3 177 0.532 ± 0.029

stat

2.82 ± 0 . 13stat 0.16 ± 0.10stat 20 1 3 + 2014 5 6 4 0 3 - 5 6 817 56.0 156 0.511 ± 0.015stat 2.63 ± 0.07stat 0.24 ± 0.06stat P G 1553+113 2013 5 6 4 4 1 -5 6 513 16.8 141 1.48 ± 0.07stat 2.95 ± 0.23stat 1.04 ± 0.31stat Notes. For both blazars, the observational period is provided along with the spectral parameters: decorrelation energy E q; differential flux at the decorrelation energy ®0; photon index r ; and curvature param eterp . These three param eters describe the log-parabola fit to the spectra.

Fig. 6. Top: (Left) PG 1553+113 excess map and (right) significance distribution for events w ith reconstructed energy betw een 100 GeV and 136 GeV (H.E.S.S. II mono analysis). Bottom: distribution o f 92 (squared angular distance to PKS 2155-3 0 4 ) for gamma-like events.

4.3. C ro ss c h e c k a n a ly sis

T he robustness o f the new H .E.S.S. II m ono results p resented above has been tested through an independent analysis using the Im age P ixel-w ise fit for A tm ospheric C herenkov Telescopes (ImPACT) m eth o d d escribed in P arsons & H inton (2014) . This independent analysis provides a consistent cross-check w ith the above results, being successfully applied to th e reconstruction o f data com ing from C T 5-only triggers (P arsons e t a l. 2015) . The analysis was equally capable o f detecting PKS 2 1 5 5 - 3 0 4 below 100 G eV an d the derived spectra w ere found to b e in very good agreem ent w ith the M odel analysis fo r b oth PKS 2 1 5 5 -3 0 4 and P G 1553+113. F urtherm ore, the difference betw een th e spectral param eters derived using Im PA CT an d th e M odel analysis was adopted as an estim ate o f th e system atic uncertainties associated w ith the reconstruction and analysis techniques (see Sect. 5) .

A dditionally, th e robustness o f the analysis w as tested using an alternative cut configuration. W ithin th e statistical and sys­

tem atic uncertainties, th e results obtained w ith the different cut configurations w ere fo u n d to b e in good agreem ent w ith each other.

T he C T 1 -4 stereoscopic d ata collected sim ultaneously w ith th e H .E .S.S. II m ono data have been analysed u s­

ing the H .E .S.S. I version o f the M odel analysis m ethod

(de N aurois & R olland 2009) using the loose cuts (A haronian e t al. 2 006a) to ensure a low energy threshold.

In total, d ata sets o f 27.2 h o f live tim e for PKS 2 1 5 5 - 3 0 4 and 9.0 h for P G 1553+113 have been analysed, yielding a signif­

icance o f 4 6 ^ for PKS 2 1 5 5 - 3 0 4 and 9 .0 ^ for P G 1553+113.

W e no te that the live tim es differ from the corresponding m ono analysis live tim es due to different run qualities an d observation schedules for the different instrum ents. F or each d ata set the spectrum is w ell fitted b y a pow er-law m odel an d the resulting forw ard-folded d ata points for PKS 2 1 5 5 - 3 0 4 (2013 data) and P G 1553+113 are show n on F igs. 3 and 7 , respectively.

T he C T 1 -4 results for PKS 2 1 5 5 - 3 0 4 w ere found to b e in excellent agreem ent w ith the H .E .S.S. II m ono results. D ue to the lim ited statistics and relatively high energy threshold o f the C T 1 -4 analysis, th e C T 1 -4 results for pG 1553+113 are represented on Fig. 7 by 3 d ata points only. Taking into consideration the system atic uncertainties on th e energy scale and flux norm alization (see Sect. 5) , th e C T 1 -4 d ata w ere found to b e in satisfactory agreem ent w ith th e C T5 results.

4.4. H E g a m m a -ra y s o b s e r v e d b y Ferm i-LAT

4.4.1. Contemporaneous data

T he F erm i-L A T detects gam m a-ray photons above an en ­ ergy o f 100 MeV. D ata taken contem poraneously w ith the H .E.S.S. II observations w ere analysed w ith the publicly avail­

able ScienceTools v1® r® p53. P hoton events in a circular re ­ gion o f 15° radius centred on th e position o f sources o f in ­ terest w ere considered and the PASS 8 instrum ent response functions (event class 128 and event type 3) correspond­

ing to the P8R2_SOURCE_V6 response w ere used together w ith a zenith angle cut o f 90° . T he analysis was p er­

form ed using the E n r i c o P ython package (Sanchez & D eil 2013) adapted for PASS 8 analysis. T he sky m odel was co n ­ structed b ased on the 3FG L catalogue (A cero e t al. 2015) . The G alactic diffuse em ission has been m od elled using the file g l l _ i e m _ v ® 6 . f i t s (A cero et al. 2016) and the isotropic b ac k ­ ground using iso_ P 8 R 2 _ S O U R C E _ V 6 _ v 0 6 .tx t.

T hree energy ranges w ere considered w ith the corresponding d ata cuts in this analysis: 0.1 G eV -5 0 0 GeV, 10 G eV -5 0 0 G eV and 50 G eV -5 0 0 GeV, w ith tim e w indow s chosen to coincide w ith the H .E.S.S. II observation periods (as defined in Sect. 3.1) . T he spectral fit p aram eter results are given in Table 2 . F or both A G N s a log-parabola fit to the contem poraneous Ferm i- LAT d ata did n o t provide a sufficient im provem ent to the spectral fit, w ith respect to the pow er-law m odel. S om e evidence for a softening o f th e spectrum w ith energy in the Ferm i-LA T energy

3 See http://fermi.gsfc.nasa.gov/ssc/data/a nalysis/

documentation/

H.E.S.S. and F erm i-LAT Collaborations: G am m a-ray blazar spectra w ith H.E.S.S. II mono analysis Table 2. Fermi-LAT spectral analysis results for the time intervals contemporaneous with the H.E.S.S. II observations.

S ource Year M JD _{Eth ¢ 0}

r

E0 TS

(GeV) 10-11 ( p h c m -2 s-1 G eV -1) (GeV )

PKS 2 1 5 5 - 304 2013 5 6 4 0 3 -5 6 6 0 1 0.1 557 ± 26 ¹₁^{q o +}_{2 -}^0.03 _0.03 1.48 2162.6

10 2.52 ± 0.43

0 0+ I00<N 25.5 379.7

50 0.12 ± 0.05 ^{1q o +0-66}

1 2 -0.72 112 52.4

P K S 2 1 5 5 - 304 2014 5 6 8 0 5 - 5 6 8 1 7 0.1 996 ± 168 ^{1 7 9 +0}_{- 0}^._.¹³ ₁₃ 1.54 193.5

10 2.36 ± 1.18 1 ^{2 0 +}0 .45

2 ^-0.45 53.3 52.4

50 1.00 ± 0.71 1 5 3 + 1.03

- 1 .20 115 23.7

P G 1553+113 2013 5 6 4 0 3 -5 6 8 1 7 0.1 118 ± 13 ^{1 . 5 9 +0.07}

-0.07 2.95 455.6

10 2.04 ± 0.53 ^SO

O O+ 1 OOkq 33.5 169.9

50 0.64 ± 0.27

2.97

^+0.91

-1.13 80.8

66.8

Notes. For each data set and energy threshold, E th, the differential flux ^ at decorrelation energy E 0, photon index r , and value o f the test statistic (TS), for the power-law fit, are provided.

range, however, was suggested b y the analysis o f F erm i-L A T data for the scan o f energy thresholds show n in F igs. 3 and 7 w hose fit indices are given in Table 2 . T he d ata points have been obtained b y redoing the Ferm i-LA T analysis in a restrained en ­ ergy range freezing the spectral index o f the pow er-law m odel to the value fo u n d for th e global fit above 100 MeV. A n upper- lim it a t 95% confidence level is com puted if the TS is found to be below 9.

T hese Ferm i-LA T analysis results are used to p ro ­ vide gam m a-ray H E -V H E SED s o f PKS 2 1 5 5 -3 0 4 and P G 1553+113. In F ig. 3, the 2013 H .E .S.S. II data set o f PKS 2 1 5 5 - 3 0 4 is p resen ted along w ith th e contem poraneous Ferm i-LA T data analysed above 100 M eV (shaded red), 10 G eV (shaded green) and 50 G eV (sh ad ed m agenta) respectively.

T hese results show very good agreem ent betw een the Ferm i- LAT and H .E.S.S. II m ono d ata w ithin th e com m on o verlap­

ping region 4, presenting a com prehensively sam pled SED over m ore than fo u r orders o f m agnitude in energy. E v idence for a strong dow n-turn spectral feature w ithin this broad b an d SED, occurring n ea r the transition zone betw een the tw o instrum ents, is apparent.

F igure 7 presents the SED o f P G 1553+113 obtained from the contem poraneous Ferm i-LA T and H .E.S.S. II data. In this case, again, go o d agreem ent betw een the Ferm i-LA T and H .E.S.S. II m ono d ata is found w ithin the com m on energy range o f the tw o instrum ents. F urtherm ore, evidence o f a strong dow n­

turn feature w ithin this SED , occurring w ithin the overlapping energy ran g e o f the tw o instrum ents, is once again apparent.

4.4.2. Catalogue data

T he H .E .S.S. II m ono and contem poraneous Ferm i-LA T spec­

tra o f PKS 2 1 5 5 - 3 0 4 and P G 1553+113 obtained in the p re ­ vious sections are com pared here to the Ferm i-LA T catalogue results. D ifferent catalogues probing different p h oton statis­

tics and energy ranges are considered here, nam ely the 3FG L (A cero e t al. 2015) , the 1FHL (A ckerm ann e t al. 2013) an d the 2FH L (A ckerm ann et al. 2016) . T he 3FG L catalogue gives an average state o f the sources w ith 4 years o f data integrated in 4 80-500 GeV for PKS 2 1 5 5 -3 0 4 and 110-500 GeV for PG 1553+113.

Fig. 7. Energy spectrum of P G 1553+113 obtained from the H.E.S.S. II mono analysis (blue) in comparison with the contemporaneous Fermi- LAT data with an energy threshold of 0.1 GeV (red triangles and confi­

dence band), 10 GeV (green band), and 50 GeV (purple band) and con­

temporaneous CT1-4 data (grey squares). In all cases the bands shown represent the 1ix confidence region. The right-hand y-axis shows the equivalent isotropic luminosity (not corrected for beaming or EBL ab­

sorption) assuming redshift z = 0.49. The inset compares the H.E.S.S.

confidence band with the Fermi-LAT catalogue data (3FGL, 1FHL and 2FHL, see Sect. 4.4.2).

the Ferm i-LA T analysis above 100 MeV, w hile the 1FHL relies on the first 3 years o f d ata w ith a h igher energy cu t a t 10 GeV.

M oreover, the 2F H L catalogue was b u ilt w ith the highest en ­ ergy available to Ferm i-LA T only, w ith E > 50 GeV, probing a som ew hat different energy range, and thus potentially different spectral properties w ith resp ect to th e F G L source catalogues.

T he insets in Figs. 3 and 7 provide a com parison o f the H .E.S.S. II m ono results (show n by the blue band) w ith the Ferm i-LA T catalogue d ata (red for 3FGL, green for 1FHL, and purple for 2FH L), fo r PKS 2 1 5 5 - 3 0 4 and P G 1553+113, resp e c­

tively.

It is w orth com paring the Ferm i-LA T contem poraneous data obtained in Sect. 4.4.1 and the Ferm i-LA T catalogue data d is­

cussed here. F o r PKS 2 1 5 5 -3 0 4 , it is noted th at the Ferm i-LAT

catalogue flux is slightly above th e Ferm i-LA T contem poraneous flux in the high energy band. F or P G 1553+113, however, the catalogue flux is in close agreem ent w ith the Ferm i-LA T co n ­ tem poraneous flux in th e high energy band. S ince the Ferm i- LAT catalogue d ata rep resen t the average flux state o f the source since data taking com m enced in 2008, th e com parable level o f the fluxes (though slightly below for the case o f PKS 2 1 5 5 -3 0 4 ) is suggestive th at b oth sources w ere in average states o f activity during the observational cam paign. It has to be n oted th at the catalogues are b ased on different tim e intervals and different en ­ ergy ranges. F urtherm ore, th e results o f the fits are dom inated by the low er energy events and, in particu lar for the 2FH L, the statistics are rath er p o o r at the highest energies.

4.5. Variability

T he A G Ns considered in this w o rk are know n to b e vari­

able a t V H E, both having previously been observed to ex ­ hibit m a jo r flares (A haronian e t al. 2 0 0 7 ; A bram ow ski et al.

2015) . In the case o f PKS 2 1 5 5 -3 0 4 , this variability has been show n to also introduce changes in the spectral shape (H .E .S.S. C ollaboration e t al. 2010).

In both cases, th e presen t observational cam paign found the A G Ns to b e in low states. F or PKS 2 1 5 5 -3 0 4 , a t E >

300 G eV th e spectrum level from ou r new H .E .S.S. II m ono resu lt agrees w ith th e level rep o rted for the quiescent state observed by H .E .S.S. from observations during 2 0 0 5 -2 0 0 7 (H .E .S.S. C ollaboration e t al. 2010) . As seen in F ig. 3 , a t E <

300 G eV th e H .E.S.S. II m o n o spectrum level lies below the F erm i-LA T spectra rep o rted in the 3FG L and 1FHL catalogues.

T hese com parisons are all consistent w ith PKS 2 1 5 5 -3 0 4 being in a low flux state during th e observations analysed in this w ork, as is also indicated by the Ferm i-LA T contem poraneous analysis results.

A lthough observed in a low state, the H .E.S.S. II m ono lightcurve o f PKS 2 1 5 5 - 3 0 4 did exhibit nightly and m onthly variability w ith a fractional variability am plitude Fvar (V aughan et al. 2003) of, respectively - 4 7 % and - 5 9 % . Inter­

year variability at V H E w ith a fractional variability am plitude F var o f - 5 0 % has also been found. A nalysis o f this variability in the H .E .S.S. II m ono d ata set revealed th at an increase in the flux exists betw een the 2013 and th e 2014 dataset b y a factor 1.6 ± 0.1stat, though w ithout significant change in the spectral p aram eters. A sim ple pow er-law fit to the 2013 (resp. 2014) data yields a spectral index r2o13 = 2.92 ± 0.04stat (resp. r2014 = 2.91 ± 0.08stat). W e note, how ever, that the statistics o f the 2013 and 2014 PKS 2 1 5 5 - 3 0 4 H .E.S.S. II m ono data sets are signif­

icantly different in size. C onsequently, the 2014 PKS 2 1 5 5 -3 0 4 data set is n o t sufficient to discrim inate betw een a pow er-law o r a log-parabola shaped spectrum , w hereas the 2013 PKS 2 1 5 5 -3 0 4 data set is found to b e significantly b etter fit w ith a log-parabolic spectrum .

F or com parison, variability analysis o f the PKS 2 1 5 5 -3 0 4 contem poraneous Ferm i-LA T data, discussed in Sect. 4 .4 .1 , was carried out. F igure 4 show s the PK S 2 1 5 5 -3 0 4 2013 and 2014 m ulti-w avelength SED obtained. It is n otable that a b rig h te n ­ ing o f the source flux betw een these tw o epochs by ab o u t the sam e level as th at seen b y H .E.S.S. II m o n o is also observed in the Ferm i-LA T contem poraneous results, an d again w ithout any corresponding spectral variability. T hat is the Ferm i-LA T and H .E.S.S. II m ono p h oton indices are respectively consistent betw een th e tw o epochs, b u t the overall flux increased by about 60% .

T he variability in H E has also been p robed on a w eekly tim escale w hich gives a good b alance betw een the ability to p robe short tim escale variations and good statistics. F or the 2013 dataset (the 2014 dataset tim e ran g e being too short), PKS 2 1 5 5 -3 0 4 is found to b e variable w ith F var = 37% .

F or P G 1553+113, our new H .E .S.S. II m ono spectral re ­ sults are in reasonable agreem ent w ith the earlier m easurem ents b y H .E.S.S. (A haronian et al. 2 0 0 8 ; A bram ow ski et al. 2 0 1 5 ; at E > 200 G eV ), M A G IC (A lbert et al. 2 0 0 7 ; A leksic et al. 20 1 0 , 2012b) an d VERITAS (A liu e ta l . 2015) , as w ell as w ith the Ferm i-LA T catalogue spectra (at E < 200 G eV ). T hese com par­

isons w ith previous m easurem ents indicate th at P G 1553+113 w as indeed in a low state during th e H .E .S.S. II observation p e ­ rio d o f the results presented. N o significant n ig ht-by-night or w eekly variability is found in th e H .E.S.S. II m o n o lightcurve.

T he u pper lim it on Fvar is found to b e 21% at the 95% confi­

dence level. In the H E range, P G 1553+113 is n o t variable and Fvar < 110% a t 95% CL.

5. Systematic uncertainties

T he m ain sources o f system atic uncertainties in the H .E .S.S. II m o n o analysis p resented in this publication, and their estim ated contributions to the uncertainty on th e spectral param eters, are sum m arised in Table 3 . F o r each source o f u ncertainty the table gives th e flux norm alisation uncertainty, th e p hoton index uncer­

tainty and the uncertainty on the curvature p aram eter j3 (for the log-parabola m odel). In addition, the energy scale uncertainty is given in the second colum n. T he energy scale uncertainty im ­ plies an additional uncertainty on the flux norm alisation w hich depends on the steepness o f the spectrum . It is also relevant for the determ ination o f the position o f spectral features such as the SED m axim um or E B L cutoff. T he procedures used here for e s­

tim ating the system atic uncertainties generally rep e at the p ro ­ cedures used for H .E.S.S. I (A haronian et al. 2 006a). W e h ig h ­ light th at the d iscussion in this section focuses specifically on the sources and analysis presented. A m o re general discussion o f the system atic uncertainties o f the H .E .S.S. II m ono analysis w ill be p art o f a future publication.

E x cep t for b ackground subtraction, all sources o f uncertainty listed in Table 3 are rela ted to th e conversion o f the m easured event counts into flux. This conversion is done using the in ­ strum ent response functions (IR F) w hich are determ ined from M onte C arlo sim ulations. T he IR F uncertainties show how w ell the real instrum ent, after all calibrations, is d escribed b y th e sim ­ ulation.

T he first group o f uncertainties is related to the in terac­

tion o f particles and th eir production and to th e absorption o f C herenkov light in the atm osphere. T he estim ated uncertainty due to the show er interaction m o d e l does n o t exceed 1% (for ph o ton-induced show ers). T he atm ospheric uncertainties include the effects o f the atm ospheric density profile (w hich affects the height o f show er m axim um and C herenkov light production) and the atm ospheric transparency (light attenuation by M ie and R ayleigh scattering). T hese effects w ere studied extensively du r­

ing H .E .S.S. p hase I (B ernlohr 2 0 0 0 ; A haronian e t al. 2 0 0 6 a;

H ahn et al. 2014) . T he uncertainties w ere fo u n d to b e dom inated by th e atm ospheric transparency, w hich has d irect influence on the am ount o f C herenkov lig h t detected by th e telescopes, thus affecting the energy reconstruction. D ata from the telescope ra ­ diom eters and o ther atm ospheric m onitoring devices, as w ell as trigger rate data, are u sed to ensure good atm ospheric conditions during the observations used in th e analysis (see Sect. 3.2) . F or zenith angles relevant to this w ork, the rem aining uncertainty on

Table 3. Estimated contributions to the systematic uncertainties in the spectral measurements using H.E.S.S. II mono for the analyses presented in this work.

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

Source o f uncertainty E nergy scale Flux Index C urvature

M C show er interactions - 1% - -

M C atm osphere sim ulation 7% - -

Instrum ent sim ulation/calibration 10% 10% - -

B roken pixels - 5% - -

L ive tim e - <5% - -

R econstruction and selection cuts 15% 15% 0.1/0.46 0.01/0.8

B ackground subtraction - 6% /10% 0.14/0.46 0.12/0.6

Total 19% 20% /22% 0.17/0.65 0.12/1.0

Notes. Numbers separated by “/” correspond to PKS 2155-304 and PG 1553+113, respectively.

the absolute energy scale due to th e atm osphere is estim ated to be - 7 % (A haronian e t al. 2 006a, sim ilar to the uncertainty level reported in).

T he instrum ent sim ulation and calibration uncertainty in ­ cludes all rem aining instrum ental effects, such as m irro r reflec­

tivity and electronics response. T hese effects are co ntrolled u s­

ing various calibration devices (A haronian e t al. 2004), as w ell as C herenkov lig h t from atm ospheric m uons (L eroy e t al. 2003) . T he non-operational pixels in the C T 5 cam era (< 5% ) and the electronics dead tim e (<5% ) contribute only m arginally to the overall uncertainty.

T he event reconstruction an d selection uncertainties are d e­

rived from a com parison o f th e m easu red spectra w ith th e results obtained using an alternative analysis chain (see Sect. 4.3) .

Irregularities in the cam era acceptance (e.g. due to non- operational pixels) and the n ig h t sky b ackground (e.g. b right stars) can b o th have an effect on b ackground subtraction. The background subtraction errors are controlled in this study b y v i­

sually exam ining the raw and acceptance-corrected skym aps (to ensure th at there are n o artefacts, e.g. from b ad calibration o f individual data runs), as w ell as using additional d edicated tests and ru n quality selection. As show n already in Sect. 4 , the w idth o f the skym ap significance distributions is d o m inated b y statis­

tical errors. This is ensured fo r both objects, PKS 2 1 5 5 - 3 0 4 and P G 1553+113, and throughout the entire energy ran g e covered by this study (see F igs. 2 and 6) . H ence, arguably, the effect o f th e b ackground subtraction errors should n o t exceed th e sta­

tistical uncertainties. C onsequently, th e statistical uncertainties on the spectral param eters rep resen t a reasonably conservative estim ate o f the b ackground subtraction uncertainties. It should be noted, however, that th e reflected-region b ackground m ethod, w hich is used for the spectral m easurem ents, is potentially m o re sensitive to n o n-axially sym m etric effects in the cam era ac ce p ­ tance than the ring b ackground m aps (w hich use a 2D ac ce p ­ tance m odel). W e have investigated this further by splitting the full d ata set into tw o subsets, one o f w hich groups th e d ata from runs taken w ith a w obble offset in right ascension (in either p o si­

tive or negative direction) and another one fo r the rem aining runs (w ith w obble in declination). T he signal-to-background (S/B) r a ­ tios obtained w ith these subsets w ere com pared to the full dataset S/B ratio. It w as found th at the S/B ratio varied b y - 3 % , w hich is about tw ice th e b ackground subtraction accuracy observed w ith the ring b ackground m eth o d ( -1 .5 % o f the b ackground level).

T herefore in Table 3 the statistical uncertainties are doubled to obtain the values for the b ackground subtraction uncertainties.

T he n e t effect o f all uncertainties sum m ed in quadrature is given in the last row o f Table 3 . It can b e n oted th at the spectral index and curvature uncertainties are dom inated by the reconstruction, event selection and back g ro u n d subtraction

uncertainties, w hile the description o f th e atm osphere and instrum ent calibration contribute substantially to the energy scale and flux n orm alisation uncertainties.

It should lastly b e h ighlighted that the system atic u n certain­

ties are energy-dependent. In particular, th e b ackground sub­

traction uncertainties tend to becom e m o re im portant tow ards low energies, w here the signal-to-background ratio is usually sm aller. F or an analysis aim ing a t the low est energies this can lead to a large uncertainty in the m easu rem en t o f spectral index and curvature, especially fo r soft spectrum sources, as is the case for P G 1553+113.

In the context o f variability studies, the uncertainty val­

ues p resented in Table 3 can be considered as a conserva­

tive u pper bound. P relim inary studies o f steady sources w ith H .E.S.S. II suggest th at th e rm s variability induced by system atic effects is ab o u t 15-20% , a resu lt sim ilar to th at for H .E.S.S. I (A haronian e t al. 2006a) . This suggests th at a t least som e o f the spectral m easu rem en t uncertainties are constant in tim e and could therefore b e reduced b y m eans o f additional calibrations.

Variations related to changes in the atm osphere transparency can also b e red u ced b y m eans o f additional corrections (H ahn e t al.

2014) .

6. Discussion

T he successful H .E.S.S. II m o n o observations and analysis o f PKS 2 1 5 5 - 3 0 4 and P G 1553+113 convincingly dem onstrate th at the low energy p art o f the V H E spectrum is accessible to the H .E.S.S. experim ent, follow ing the addition o f th e C T 5 in ­ strum ent. This fact m akes E B L studies o f high red sh ift A G Ns by H .E.S.S. II m ono feasible, w ithout the n ee d for strong theo reti­

cal biases on the intrinsic spectra or th e need to rely on spectral extrapolations using results from other instrum ents.

H ere w e consider E B L deabsorbed fits to the H .E .S.S. II m ono and contem poraneous Ferm i-LA T spectra for b oth AGNs.

O ur aim here is tw ofold. T he first is to investigate evidence for curvature in th e tw o A G N intrinsic spectra, correcting fo r E B L absorption effects. Second, given the p rese n t system atic uncer­

tainties derived for these d ata sets, w e determ ine the correspond­

ing uncertainties on the com bined fit param eters. Such consid­

erations provide insight into the constraining pow er o f these re ­ sults, u nder the assum ption o f both a specific E B L m odel (in this w ork the one o f F ranceschini et al. 2008) and sim ple underlying spectral shape.

T he spectra in the H .E .S.S. II m ono energy range have been reconstructed w ith a spectral m o d el co rrected for E B L absorption. F urtherm ore, for P G 1553+113, w hose redshift is n o t w ell-constrained, w e adopt the w ell-m otivated value o f z = 0.49 (A bram ow ski et al. 2015) .

Table 4. Parameters obtained for the combined fit of the Ferm i-LAT and H.E.S.S. data.

Source 0o[10 11 cm 2 s 1]

r

P lo g1 0(Ep e a k[GeV]) Sig. X )

PKS 2 1 5 5 -3 0 4 P G 1553+113 P G 1553+113

2.35 ± 0.10s t a t ± 0.57Sy S

5^.97 ± 0^.25^{s t a t} ± 2^.19^{Sy S} 6.66 ± °.42s t a t ± b 4 3s y s

2.30 ± 0.04stat ± 0.09SyS 1.68 ± 0.05stat ± 0.13SyS L 83 ± ° . ° 8stat ± ° .2 9 sys

0.15 ± 0.02stat ± 0.02SyS 0.12 ± 0.05stat ± 0.13SyS

°. " ± °. 19stat ± °.19sys

2.76 ± 0.45sta t ± 0.93Sy S

5.1 2.2 Notes. The reference energy E 0 used here is 100 GeV. For both blazars, the log-parabola fits values are provided. For PG 1553+113, the values for the power-law model, which was marginally disfavoured, are also given. The last colum n gives the significance, obtained by comparing the X values for the log-parabola m odel against those for the power-law model, using only statistical errors in the analysis.

Fig. 8. Energy spectrum of PKS 2 1 55-304 obtained from the H.E.S.S. II mono analysis (blue) o f the 2013 data corrected for EBL absorption in com parison w ith the contem poraneous Ferm i-LAT data with a m inim al energy of 0.1 GeV (red). The black line is the best-fit log-parabola m odel to the points and the cyan butterfly indicates the 1ix region using only the statistical errors in the combined data set analy­

sis. The right-hand y-axis shows the equivalent isotropic lum inosity (not beaming corrected).

In order to look for a p ossible turnover in the intrinsic spec­

trum and, if present, to locate the peak em ission in the energy flux ( E 2d N /d E ) representation, th e E B L deabsorbed Ferm i-LA T and H .E .S.S. II m ono d ata points w ere fitted b oth separately and as a com bined d ata set w ith pow er-law , broken pow er-law and log-parabola m odels. In th e com bined fit procedure, a consider­

ation o f the system atic uncertainties for each o f th e d ata sets was taken into account in the analysis.

F or th e H .E.S.S. system atic uncertainties, the effect o f the energy system atic uncertainty on th e d eabsorbed spectrum fit results was found to be the d o m inant contributing system atic.

T he contribution o f this uncertainty on the results w as estim ated through the shifting o f the d ata points in th e E d N /d E re p re ­ sentation b y an energy scale factor o f 19% (see Table 3) b e ­ fore applying the E B L deabsorbtion. T he variation in the best-fit m odel, introduced via the application o f this procedure w ithin the full energy uncertainty range, w as then taken as th e system ­ atic contribution to the uncertainty on each m odel p aram eter (see Table 4 ) . A n estim ate o f the size o f the Ferm i-LA T system atic uncertainties w as also obtained, using th e effective area system ­ atic uncertainty, derived by the LAT collaboration 5. T hese u n ­ certainties w ere n oted to b e sm all in com parison to the statis­

tical errors such that th eir further consideration could be safely neglected.

In the case o f PKS 2 1 5 5 -3 0 4 , separate fits o f the Ferm i- LAT and H .E .S.S. II m ono E B L deabsorbed data, the pow er-law m odel w as found to provide a sufficient description in both cases.

5 See http://fermi.gsfc.nasa.gov/ssc/data/a nalysis/

scitools/Aeff_Systematics.html

Fig. 9. Energy spectrum of PG 1553+113 obtained from the H.E.S.S. II mono analysis (blue) corrected for EBL absorption in comparison w ith the contem poraneous Fermi-LAT data with a m inim al energy of 0.1 GeV (red). The assumed redshift is z = 0.49. The black line is the best-fit log-parabola m odel fit to the points and the cyan butterfly indi­

cates the 1ix (statistical error only) uncertainty region. The right-hand y- axis shows the equivalent isotropic lum inosity (not beaming corrected).

T he pow er-law fit o f th e H .E.S.S. II m ono 2013 data obtained an intrinsic spectral index o f r = 2.49 ± 0.05. Such an index ap ­ pears som ew hat softer than the pow er-law analysis o f th e Ferm i- LAT contem poraneous d ata ( r = 1.82 ± 0.03 see Table 2) . The spectral fits fo u n d for the com bined d ata sets, d om inated b y the low energy data points w here E B L effects can be neglected, a l­

low ed the continuity o f th e source spectrum to b e probed. The fit o f the com bined Ferm i-LA T and H .E.S.S. II m ono d ata w ith a log-parabola m odel w as preferred at th e 5 .1 ^ level w ith respect to the pow er-law m odel (See F ig. 8). T he broken pow er-law does n o t significantly im prove the fit in this case. T he results o f the fit are given in Table 4 . T he p ea k flux position w ithin th e SED was at a m o d erate energy (around 10 G eV ), in agreem ent w ith its 4-y r averaged p osition found in the 3FGL.

F or P G 1553+113, an E B L absorbed pow er-law fit to the H .E.S.S. II m ono spectra required an intrinsic spectral index o f r = 1.91 ± 0.13. F or com parison, Table 2 shows that the Ferm i-LA T spectral fits for pow er laws w ith thresholds o f 100 M eV an d 10 G eV give co nsistent spectral indices to this value. O n the other hand, the fit o f the com bined Ferm i-LAT and H .E .S.S. II m ono gam m a-ray data, however, found a log- parabola m odel preferred at the 2 .2 ^ level over the pow er-law m odel (See Fig. 9 ) . T he fit values fo r these tw o spectral m o d ­ els are p rovided in Table 4 . T he param eters th at results from fits w ith a broken pow er-law being close to one o f th e sin­

gle pow er-law m odel case. T he sizeable system atic errors, once also taken into account, however, w eaken this preference. Thus, this only m arginal im provem ent, b rought b y the log-parabola m odel, suggests th at the observed softening o f the P G 1553+113 spectrum is p redom inantly introduced by V H E interaction on