The effect of background subtraction on the temperature of EIT coronal loops

We have selected a sample of 10 coronal loops that were clearly visible in the 171 Angstrom passband of the SOHO EIT, five on the limb and five on the disk. Our analysis was limited to 171/195/284 image "triplets"-observations taken when the instrument cycles through the different passbands during routine operations. This cycle takes only a few minutes, so each of these nonflaring structures did not change significantly during the cycle. We chose five pixels along each loop and five carefully selected background pixels. Temperature analysis was done four different ways: (1) standard EIT analysis on the five loop pixels with no background subtraction; (2) constant background subtraction for each EIT image; (3) pixel pair background subtraction; and (4) radial background array subtraction (this method works only for loops observed above the limb). Each method produced two estimates of temperature for each loop pixel, one from the 171:195 ratio and the second for the 195:284 ratio. Both ratios produced loops with a uniform temperature, but each ratio results in a statistically different temperature value, perhaps indicating that the plasma along the line of sight was not isothermal. Background subtraction did not affect the EIT temperature analysis, i.e., the results were the same with and without background subtraction. The results for loops on the limb were "cleaner;" i.e., had less scatter, than for loops on the disk. Finally, we did a similar temperature analysis with five randomly chosen pixels for each data set. The results were the same as for the loop pixels: two statistically different, uniform temperature "structures." These findings indicate that EIT ratio analysis does not generate a physically meaningful value for the electron temperature.