Atmospheric Monitoring in the TA experiment Thank you for give me a good opportunity. I am Takayuki Tomida from RIKEN in Japan. Today, I will talk about “Atmospheric monitoring system in Telescope Array experiment”. At first, I will make a introduction regarding our experiments. Takayuki Tomida and the TA collaboration RIKEN

Telescope Array(TA) Experiment Hybrid observation: SD (507 units) + FD (3 locations: 38 units) Fluorescence Detector (FD) TA is the inter national joint experiment with Japan, the United States, South Korea and Russia. The observation started in Apr. 2008 at Desert area of Utah. We have 2 type detector for ultra high energy cosmic ray. One is surface detector. Another one is Fluorescence Detector that are located in 3 station around the experimental site. Each station has about 12 telescopes. And One station was transferred from the HiRes. Surface Detectors SDs Plastic scintillator The joint experiment with Japan, the United States , South Korea, Belugium and Russia. The observation started in Apr. 2008 North American at Utah

Atmospheric monitor in TA LIDAR CLF LIDAR@CLF IR camera CCD camera weather monitor LR We will use the telescope, we are also owned atmospheric monitoring system. There are two purposes of atmospheric monitoring. One is the atmospheric transparency measurements for calibration of the observed data. We have 2 laser system for the atmospheric transparency measurements. These are LIDAR, CLF and LIDAR@CLF. Another is the cloud observation for the data cut and improvement of observation quality. We have IR camera, CCD camera and Eye-scan-code for cloud observation. Eye scan code records the appearance of the night sky using the observer's eye.

Contents LIDAR observation CLF Observation IR camera Observation The atmospheric transparency model of two kinds of altitude distribution was determined. Influence of using LIDAR’s atmospheric transparency for FD reconstruction. FD reconstruct fluctuation was estimated by using the atmospheric model. CLF Observation Correlated to the time variations was observed when compared to the CLF and LIDAR by Optical Depth. IR camera Observation Eye-scan Today’s contents is this. I will talk about the Atmospheric transparency monitoring in the first half. And In the after half, about the cloud monitoring.

Data condition for determination atmospheric model LIDAR System BRM-St. LIDAR 100m Telescope & dome of TA LIDAR BRM Station Measurement : Before and After FD observation Slope Horisontal shots - high power - 500 shots Klett’s Vertical shots - high/low power - 500 shots Incline shots - high power - 500 shots LIDAR system is our major atmospheric monitor, these data is used to calibrate the FD in currently. LIDAR is located at near the BR station that is FD station in South-East of our aria. The system was made to remodel the 30cm telescope of MEADE. Extending the elevation control axis, attached the table on the extending bar and mounting the laser head on the table. LIDAR is observed twice in FD observation night. In one observation, 4 kind measurement is performed. Observation of horizontal measures the transparency on the ground, another observation measures for determining the distribution of transparency in the sky. Data　condition for determination atmospheric model Data period ~2 year (Sep.2007 ~ Oct.2009) Using data Fine data Good LIDAR observation Transparent atmosphere Rayleigh Radiosonde atmosphere @ELKO

Models of Atmospheric transparency 1σ=+83%/-36%. single exponential double exponential We make model of Atmospheric transparency in TA site from Median of Extinction Coefficient. I make 2 models using exponential. One is Single exponential model, Another one is Double model. Green & Blue line are VAOD these are calculated from Models, These are in good agreement with the median of VAOD. VAOD model has fluctuations, that has red error bar. Extinction coefficient at each height VAOD at each height Double exponential Model Single exponential Model

Seasonally Aerosol scattering summer winter Median of VAOD for different seasons Distribution of VAOD at 5km above ground level for different seasons In TA site, the atmosphere transparency has a seasonal dependence. I shows the median of VAOD of winter and summer. Histograms shows distribution of VAOD at 5km above ground. This Value is median of VAOD at 5km above ground. The effect of the aerosol component in summer is 1.5 times greater than that in winter. Summer: 0.039 +0.020 - 0.012 The effect of the aerosol component in summer is 1.5 times greater than that in winter. Winter : 0.025 +0.010 - 0.007

=Method= Fluctuation of FD reconstruction using atmospheric transparency by the LIDAR measurement. =Method= MC simulation using daily atmospheric transparency to create a shower data. Simulated data are reconstructed using daily atmospheric transparency or model function. Estimating the impact of using a model function to compare the results with the reconstruction of each atmospheric transparency. ΔE is evaluated by the ratio, ΔXMax will be evaluated by difference. Reconstruction using Daily atmospheric data or two atmospheric models Next, I estimate fluctuation of FD reconstruction using model of atmospheric transparency. =Simulation conditions= Primary energy : logE= 18.5, 19.0 and 19.5 eV Direction: Zenith is between 0 ∼ 60 ◦ (the isotropic) Azimuth is between 0 ∼ 360 ◦ (the isotropic) Core position : within 25 km of the CLF (center of TA FDs). Number of event : 20 events at each energy for each of 136 good LIDAR runs. Quality Cuts : Reconstructed Xmax in field of view of FD.

Fluctuations by using the atmospheric model Daily vs model func. @logE=19.5 eV Energy XMax Comparison of reconstructed fluctuation in atmospheric model. The fluctuation not containing the reconstruction bias using atmospheric model at each energy These figures shows difference Energy and Xmax at 10^19.5 eV between daily reconstruction and model reconstruction. That is calculation equation. When 1019.5 eV air shower is reconstructed using the model function, the systematic uncertainty of energy is shown to be about 11%. And the systematic uncertainty of XMax to be about 9 g/cm2 by comparing MC simulation data. Rec. ΔE : 6%@18.5 9%@19.0 11%@19.5 Rec. ΔXmax : 9g@18.5 9g@19.0 9g@19.5

CLF container and power generation system and optics of CLF CLF System Starting CLF operation :2008.Dec〜 CLF container and power generation system and optics of CLF Next, Other hand laser system that is CLF. CLF is located in center of TA site. Distance form all of FD is 20.8 km. CLF was started form December 2008. CLF laser operation is 300 laser shooting in every 30 minutes. Laser frequency is 10 Hz. Laser diameter is expanded into 1cm, it is de-polarized light. Laser power is determined by direct measurement and indirect measurement. It is measured indirectly during operation. I will measure the relationship of indirect measurement and laser energy emitted into the atmosphere. Optical diagram of the CLF CLF laser is injected into FD’s FOV :300 shots :10Hz :vertical direction :every 30 minutes. Block diagram of devices for CLF

CLF‘s observation image VAOD eq. That is observation image of CLF. 図の説明 This formula indicates the number of photon received by the FD in a simple. NP0 is number of photon in the laser at CLF. T-ray & T-as is Optical depth in vertical laser path. S shows the scattering coefficient to the FD direction. T’ is the Optical Depth of up to FD from scattering point.

analysis method Uniform atmospheric No aerosols This formula is solved by the ratio with the ideal state that dose not exist aerosol. In addition, I will use the two assumptions. One is uniform atmosphere that means the same atmospheric conditions in the same height. Another one is no aerosol in the scattering point. Aerosol-scattering less data or simulation is used in the ideal state. I tried using Aerosol-scattering less data. No aerosols

This indicates the analysis procedure. By the laser power calibrated and camera calibration of the FD, we get a normalized waveform.

I exclude Cloudy data. After that, I look for [Aerosol-scattering less data] from all of the normalized data. And, I get VAOD by applying the formula.　次ページに図有り

VAOD (Example) & Comparison of BR &LR VAOD (LR) VAOD (BR) I shows a comparison of VAOD obtained by the BR and LR data. Horizontal axis is BR and Vertical is LR. Both are in good agreement.

This is a histogram of VAOD at 8km & 10 km above the ground. BR seems somewhat hazy than LR in the statistics. And, It is consistent with the results of the LIDAR in the statistics.

Comparison of time dependence between LIDAR and CLF 2009.Oct.16〜Oct.18 LIDAR can be measured VAOD to LIDAR from the cloud. CLF can measure VAOD until over the cloud, because CLF laser penetrate the cloud. It shows Comparison of time dependence between LIDAR and CLF

Conclusion of LIDAR The extinction　coefficient α is obtained from LIDAR observation, then the VAOD τAS(h) is defined as the integration of α from the ground to height h. A model of αAS with altitude was found by fitting two years of LIDAR observations. The range of variation of the daily data from the model is +83%/-36%. When 1019.5 eV air shower is reconstructed using the model function, the systematic uncertainty of energy is shown to be about 11%. And the systematic uncertainty of XMax to be about 9 g/cm2 by comparing MC simulation data.

Conclusion of CLF VAOD was analyzed by using the CLF event of high view camera's. BR and LR are consistent with a few %. There is a correlation VAOD measured in each of the CLF and LIDAR. Using the CLF, will be able to interpolate for the atmospheric transparency of the period where have not been observed by LIDAR.

LIDAR@CLF system Hardware (general drawing) Back-scatter detector is set up on top of the CLF. LIDAR@CLF use PMT of 20mm and 38mm in diameter. telescope & 20mm PMT for High altitude (1.5~7.0~ km) 38mm PMT for Low altitude (~2.5km) Because of data analysis, it is only introduction. Fig. Block diagram of LIDAR@CLF Fig. general drawing of LIDAR@CLF

Cloud monitor

TA IR camera Sensitive 8 ~ 14 us 320 x 236 pixels FOV: 25.8o x 19.5o Near the LIDAR dome Once every 50 min (~2009Jul) or 30min (2009Jul~) 320, 25.8o 236, 19.5o Our main cloud monitor is IR camera. IR camera is taken every 30 min now. 7 8 9 10 11 12 6 5 4 3 2 1

IR Sky Images Clear Cloudy The IR camera digitizes sky temperature. If there are clouds, the sky looks warmer. An IR image are split into 4 “sections” horizontally in data analysis, because lower elevation region looks like warmer. Deciding the probability of cloud in each section and each season. sec1 sec2 sec3 sec4 Cloudy The IR camera digitizes sky temperature. These histograms show distributions of pixel values of example IR images. If there are clouds, the sky looks warmer, and the distribution shifts right. In the data analysis, an IR image is split into 4 sections, because the average sky temperatures is different for different elevations. Deciding the probability of cloud in each section and each season. sec1 sec2 sec3 sec4 D: Pixel Data

Examples Total: 1.05/48.0 Clear night Total: 47.0/48.0 Cloudy night Score = 2.18/4.00 Total: 1.05/48.0 Clear night Total: 47.0/48.0 Cloudy night The IR camera digitizes sky temperature. These histograms show distributions of pixel values of example IR images. If there are clouds, the sky looks warmer, and the distribution shifts right. In the data analysis, an IR image is split into 4 sections, because the average sky temperatures is different for different elevations. Deciding the probability of cloud in each section and each season. Total: 13.0/48.0 0.034 0.035 0.029 0.174 1.991 3.790 Sparse night 0.068 0.653 1.314 1.532 2.046 3.834

Sum of Scores of All the Directions IR Camera Score Sections 3&4 of Bottom layer exclude from analysis. The ratio of clear-cloudy nights is about 7 to 3. This figure is the histogram of IR camera cloudy score. Sections 3&4 of Bottom layer exclude from analysis. Because, Pixels near the ground can not be distinguished in the Cloud and Non-Cloud under the influence of heat and aerosols from the ground. Left-hand side is the clear sky. The ratio of clear-cloudy nights is about 7 to 3. Clear Cloudy

Comparison between IR and Eye-scan Eye’s scan Code IR Camera Score Eye’s-Scan Code is index of the cloud to determine in the observer's eye to the FD observation night. The code is a total of 6 points. IR score and Eye-scan code is consistent. Eye’s-Scan Code is index of the cloud to determine in the observer's eye to the FD observation night. The code is a total of 6 points. All sky clear night is 0 point of eye-scan code. In BR case, the sky of the north and west is each 1 point, and the zenith direction has 4 points. In MD case, the sky of the South and East is each 1 point, and the zenith direction has 4 points. IR score and Eye-scan code is consistent. Clear Cloudy

Comparison between IR and CLF The data is extracted, when CLF and IR operate within 10 minutes Color-coded a histogram of the IR score by CLF’s weather condition. IR score and CLF data is consistent. The CLF weather conditions compared with IR camera cloudiness. The data is extracted, when CLF and IR operate within 10 minutes I color-coded a histogram of the IR score by CLF’s weather condition. Examples are determined to cloudy in CLF

Conclusions (Cloud monitor) About 70% of the TA observation night is Clear night IR score and Eye-scan code is consistent. IR score and CLF data is consistent.

Typicals of Extinction Coefficient less Aerosol scattering Aerosol distributed only low height α 10 Height above ground [km] Aerosol distributed high height Aerosol distributed both height

Height above ground [km] Typicals of VAOD Aerosol distributed only low height less Aerosol scattering VAOD 10 Height above ground [km] Aerosol distributed high height Aerosol distributed both height

Comparison between BR and LR (2009.08.26〜2010.02.14) VAOD of LR is larger than 6% more BR. The adjustment of de-polarization was shifted slightly in this observation term. The likely influence of de-polarization adjustment. For future, I will confirm in another observation term.

Comparison between LIDAR and CLF Conditions 2009.Sep〜2009.Dec No cloud |Timelidar-TimeCLF| <1hr

Effects on energy by atmospheric fluctuation single component 18.5 19.0 19.5 double component 18.5 19.0 19.5

VAOD (LR) VAOD (BR) I shows a comparison of VAOD obtained by the BR and LR data. Horizontal axis is BR and Vertical is LR. Both are in good agreement.

Effects on Xmax by atmospheric fluctuation single component 18.5 19.0 19.5 double component 18.5 19.0 19.5

Fluctuation of reconstruction by each atmospheric logE=19.5 eV Energy XMax The fluctuation Including the reconstruction bias using atmospheric model at each energy are result of reconstruction by each atmospheric conditions. Rec. ΔE : 10%@18.5 12%@19.0 16%@19.5 Rec. ΔXmax : 19g@18.5 18g@19.0 10g@19.5

Rayleigh scattering Jan Apr Jul Nov

Fluctuations by using the Monthly average

Date variation of VAOD @8km & 10km Winter atmosphere may be clear. There is correlation with LIDAR.

Normalized by VAOD of CLF. Analysis policy of LIDAR@CLF Analytical result only of LIDAR@CLF Analytical result only of CLF ? × ? × ＋ VAOD VAOD Height[km] Height[km] Normalized by VAOD of CLF. Shape of VAOD according to height is determined from LIDAR@CLF. VAOD at high altitude is determined from the analysis of CLF. × × VAOD Analytical result of LIDAR@CLF and CLF 42

Fluctuation of FD reconstruction using atmospheric transparency by the LIDAR measurement. Next, I estimate fluctuation of FD reconstruction using model of atmospheric transparency.

Typicals of Extinction Coefficient less Aerosol scattering Aerosol distributed only low height α α 0 5 10 Height above ground [km] 0 5 10 Height above ground [km] These are example of LIDAR observation data that is the vertical distribution of extinction coefficient from the ground. Data point shows LIDAR observational result. The color variations indicates the type of observation. Black solid line shows Rayleigh scattering by molecules. Mie scattering value by aerosols can be obtained after subtracting Rayleigh from LIDAR observational value. Distribution of atmospheric transparency are manifold. In Fine night, we get result like a left-top. Aerosol distributed high height Aerosol distributed both height α α 0 5 10 Height above ground [km] 0 5 10 Height above ground [km]

Typicals of VAOD VAOD VAOD 0 5 10 Height above ground [km] 0 5 10 Aerosol distributed only low height less Aerosol scattering VAOD VAOD 0 5 10 Height above ground [km] 0 5 10 Height above ground [km] And, We calculate Vertical Aerosols Optical Depth that is integral value of extinction coefficient. Aerosol was not observed most in the sky above 5km or more. So flat figure. Aerosol distributed high height Aerosol distributed both height VAOD VAOD 0 5 10 Height above ground [km] 0 5 10 Height above ground [km]

This indicates the analysis procedure. By the laser power calibrated and camera calibration of the FD, we get a normalized waveform.

I exclude Cloudy data. After that, I look for [Aerosol-scattering less data] from all of the normalized data. And, I get VAOD by applying the formula.　次ページに図有り

Simulation conditions Primary energy : logE= 18.5, 19.0 and 19.5 eV Direction: Zenith is between 0 ∼ 60 ◦ (the isotropic) Azimuth is between 0 ∼ 360 ◦ (the isotropic) Core position : within 25 km of the CLF (center of TA FDs). Number of event : 20 events at each energy for each of 136 good LIDAR runs. Quality Cuts : Reconstructed Xmax in field of view of FD. Reconstruction using Daily atmospheric data or two atmospheric models