Correct attenuation length model behavior at large depths

[marcomuzio]

The attenuation length models implemented in AraSim give unphysical behavior at large depths. The following changes should be implemented to produce more accurate behavior and an improved design going forward:

• For each of the three attenuation functions (IceModel::EffectiveAttenuationLength, IceModel::GetARAIceAttenuLength, and IceModel::GetFreqDepIceAttenuLength) two if statements should be added:
  • if the depth is < 2.5 km (depth where Dave Besson says the model becomes unreliable/unphysical), an attenuation length of 0 should be returned
  • if the depth is < bedrock, an attenuation length of 0 should be returned (this case is added as a fail-safe in case the model behavior between the bedrock and 2.5 km depth is corrected down the line)
  • NB this behavior may need to be handled with some care to avoid introducing nan's when the attenuation factor is calculated ~1/attenuation length in the Report class
• The ice temperature only enters through the attenuation length, but for safety/consistency, some safeguards should also be added to IceModel::temperature.
  • The function should do something like return min(temp, 0), to ensure it does not return positive temperatures at large depths.
  • One could consider causing the function to crash if a depth below/above the ice was passed (but this depends a bit on how the attenuation length function is corrected)

[clark2668]

Just claiming that I'm working on this.