WYSH

The Wyoming Survey for Hα

Survey Overview

The Wyoming Survey for H&alpha is being conducted at the University of Wyoming with the Wyoming Infrared Observatory (WIRO) using the prime focus camera. The survey's purpose is to determine global star formation rates for the redshift, z=0.155, 0.24, 0.31, and 0.39 epochs of the universe (redshift is due to objects moving away from us, and objects moving faster away from us are more distant than the slower objects, and the more distant something is, the older that thing is, so as redshift increases, so does lookback time). It accomplishes this task by using filter pairs calibrated to 7598 and 7661, 8132 and 8199, 8615 and 8685, and 9155 and 9233 Angstroms. H&alpha is a strong emission line so we can distinguish between H&alpha and the continuum. Because redshift shifts the H&alpha emission line around we can determine the z from the wavelength we calibrate the filters to detect. Once we have an image it will be populated by galaxies and we can then measure the amount of H&alpha they are emitting and from that determine a global star formation rate (for example, redshift shifts H&alpha's wavelength from 6562.8 Angstroms at rest to between 7598 and 7661 Angstroms at z=0.155, or 13.7 Gyr since the Big Bang, so the data from those filters is indicative of the star formation at time=13.7 Gyr since the Big Bang).

Completeness Simulations

When we use a detector to find star formation rate we are limiting ourselves. Effects such as diffraction, aberrations due to optics, detector sensitivity, and noise limit our ability to detect every source in a field. This is significant when trying to determine star formation rates because galaxies which are not detected can contribute significantly to the star formation rate. To remedy this we have created simulated images with simulated galaxies. After running our detection routines on these images we compared the amount of detections to the amount of galaxies that we simulated. This fraction of detected galaxies to simulated galaxies was then used to make our luminosity curve more accurate.

Simulation Parameters

1. Luminosity Bins - [33.2, 33.6, 34.0, 34.4, 34.8, 35.2, 35.6] in log(H&alpha)

2. X and Y position on the chip - Between 50-1987 and 50-1998 respectively, found by a random number from a uniform distribution

3. Number of Galaxies simulated - 1500 for the results reported here

4. Equivalent Width - Either 0-2 or 1-2 in log(EW/), centered on 1.5 with a gaussian distribution

5. Position Angle - 0-360 from a uniform distribution

6. Axial Ratio - From 0.15-1.0 from a uniform distribution (The intrinsic fatness of the disk prevents an axial ratio < 0.15.)

7. Half Light Radius - R0.5 as defined by the equation -ln(0.5)*( &rho + randn*[&rho /3]) where the disk scale length &rho is 1.18 for z=0.81, the redshift our filter is calibrated to detect, and randn is a random number from a normal distribution with a standard deviation of 1 and a mean of 0

8. Extinction - For these results extinction was based on calculations from Hopkins et al. 2001.

9. Distance Threshold - How far a detected source can be from a simulated source to be considered a match (3 pixels or ~1.6'')

10. Signal to Noise Threshold - 3 for the results on this page

11. Auto-parameters in stack.sex

Results
log(EW/)=0-2, auto_params=3.5,4.5
log(H&alpha/W) z=0.155 z=0.24 z=0.31 z=0.39
332 14.60 4.000 0.4000 0.06670
336 70.30 23.40 2.267 0.7330
340 95.20 83.60 13.27 4.070
344 93.90 94.40 65.50 21.20
348 94.40 95.70 96.10 73.80
352 94.00 95.10 95.50 93.70
356 92.50 94.90 95.00 95.40
log(EW/)=1-2, auto_params=3.5,4.5
log(H&alpha/W) z=0.155 z=0.24 z=0.31 z=0.39
332 15.13 4.930 0.3330 0.06670
336 70.60 22.33 3.400 0.4000
340 94.80 86.90 14.87 4.000
344 95.20 95.70 69.10 20.80
348 94.90 95.10 96.30 79.20
352 95.30 94.80 95.70 96.20
356 93.90 94.30 95.20 96.30

log(EW/)=0-2, auto_params=2.5,3.5
log(H&alpha/W) z=0.155 z=0.24 z=0.31 z=0.39
332 29.00 8.730 0.8670 1.733
336 88.90 46.30 4.270 6.870
340 95.30 92.40 27.33 39.30
344 93.90 94.70 83.60 87.50
348 94.40 95.70 96.40 93.60
352 94.00 95.10 95.50 95.10
356 92.50 94.90 95.00 95.40
log(EW/)=1-2, auto_params=2.5,3.5
log(H&alpha/W) z=0.155 z=0.24 z=0.31 z=0.39
332 31.90 9.070 0.7330 1.200
336 89.20 46.20 4.930 9.070
340 94.90 95.00 29.20 43.10
344 95.20 95.70 88.80 93.00
348 94.90 95.10 96.30 96.80
352 95.30 94.80 95.70 96.30
356 93.90 94.30 95.20 96.10

log(EW/)=0-2, Extinction=1 Mag, auto_params=3.5,4.5
log(H&alpha/W) z=0.155 z=0.24 z=0.31 z=0.39
332 2.100 0.9000 0.06670 0.1000
336 23.10 6.800 0.6670 0.4670
340 91.50 55.00 6.430 1.800
344 93.50 94.10 53.60 14.57
348 93.70 95.40 96.00 77.40
352 93.30 95.10 95.40 94.00
356 92.70 94.70 94.80 95.50
log(EW/)=1-2, Extinction=1 Mag, auto_params=3.5,4.5
log(H&alpha/W) z=0.155 z=0.24 z=0.31 z=0.39
332 2.600 0.8000 0.06670 0.06670
336 24.70 6.700 0.7670 0.2333
340 94.30 60.00 7.400 1.600
344 95.40 95.50 56.90 15.70
348 94.90 94.80 96.30 83.00
352 95.10 95.00 95.70 96.20
356 94.20 94.60 95.30 96.20

 

Overlap Issues

When two sources are randomly assigned positions very close to each other sometimes SExtractor believes that they are a single source. This occurs when one of the galaxies is dimmer than the other. SExtractor then believes that the dimmer galaxy is part of the profile of the brighter one. If the pixel distance threshold is large enough then the two simulated galaxies will be matched to the same SExtracted galaxy. This is bad because if the simulated image was a real image then only one galaxy would have been detected, while in our completeness simulations it would have counted as two galaxies being detected and our luminosity curve would be incorrect. So to minimize this effect we made the pixel distance threshold only 3 pixels. Our completeness reaches a maximum (~95%) as can be seen in the above tables, and this overlap is the reason (~5% of our sources will overlap one another with 1500 sources per image). See the image below for an example.

The blue regions are detections from SExtractor while the red regions are where the galaxies were created by mkobjects. Note the galaxies in the upper left where SExtractor failed to keep the galaxies separated. Note that some galaxies appear boxy because of the stretch used in DS9, and not because they are saturated and really are boxy.

These simulations were done with help from Danny Dale and Micah Schuster

Last Update: 12:30 MST 4/14/2009