Tilted column3

[magpowell]

This PR provides functionality to run the tilted column approximation with the ray tracer. It is NOT applicable for the 2-stream solver where path length attenuation is not applied. Results coming out of 2-stream are biased.

The TICA code works as follows:

• For gases, aerosols, temperature, and pressure, a single tilted path initiated from bottom pixel center is used.
• "To convert from the 3D cloud information to columns that can be used as input to a standard 1D radiative transfer code, the distribution of op-tical properties along individual rays is translated to a vertical column where the 3D grid boxes are translated to layers, the thickness of which reflects the pathlength in each 3D grid box."
• The number of layers that comes out of the tilted field increases with 1/mu. The layers are compressed starting from TOD to match the original number of layers. If SZA is too high and the field would have to be compressed by greater than a factor of 2 the code exits with an error message.
• For ice and cloud optical properties, the tilted path is initiated from a different start location for each tilted column. The tilted columns are compressed to match the original z layers.
• Within the ray tracer code, when TICA is passed, independent column is turned on too. Within the ray tracer, optical depths are scaled by 1/mu to reflect the increased path length. This scaling is not implemented in the 2-stream.

Surface Flux Fields from Cloud Botany for 3D and TICA:

With tilting and compression the Cloud Botany field still fits into memory and runs in minutes.

Differences are most prominent at cloud edges:

Due to random start location of tilted paths these differences are unbiased.

Differences in diffuse are due to physical mechanisms: side escape, entrapment of diffuse radiation. Differences in direct are due to computational error, the difference is less than 0.5% with randomized path start location implemented.

[magpowell]

@Chiil or @MennoVeerman , would you please provide me with a test file that contains aerosols and ice clouds? I've tested using warm clouds only.

[magpowell]

Working with two-stream now!

Difference between 3D ray tracer direct and TICA 1D ray tracer:

Difference between 3D ray tracer direct and TICA two-stream:

[MennoVeerman]

TICA implementation looks good and works well and efficiently!