Photons in a cloud

A Monte-Carlo-Simulation

Overview


Introduction

A photon injected into a cloud layer can travel a certain distance - the so called mean free path - before getting absorbed or scattered. If scattered the photon changes direction, travels again a small distance and has again the option to be scattered or absorbed. This sequence will continue until it finally gets absorbed (or ejected from the cloud layer). All these events -  absorption, scattering as well as the drawing of a new direction - are determined by key parameters of the cloud layer and by certain probability distributions. Hence, if the interaction of a photon is calculated over and over again - as done in a Monte-Carlo-Simulation - it is possible to statistically determine the amount of photons that are

In this way, the irradiance that is transmitted and reflected by a cloud can be calculated. 

Basics

As stated it is possible to describe the likelihood of what is happening to a photon inside a cloud by evaluating the uniform probability distributions of the possible events.  In their book Fundamentals of Atmospheric Radiation  (Bohren and Clothiaux 2006, Ch. 6.3) explains in great detail how to set up the necessary equations and density functions to perform this Monte-Carlo simulation. Here are only the basics summarized: 

Describing the set up

For the calculation a homogeneous cloud layer of a certain height is used. The cloud bottom is confined by the x-y-plane and the cloud top is parallel to this plane.  Photons are injected at the top at (x=0, y=0, z=cloudheight) under an incident angle α. This is the angle between the positive z-axis and the incident beam that lies in the x-z-plane.


Determine travelling distance of a photon

The distribution for the photon path length  x with a total mean free path mfp can be written as 

With the following equation (and by using a uniform random number generator rand) it is possible to calculate a photon path length that fit this distribution:

Determine if photon is absorbed or scattered

The so called single-scattering albedo controls how likely a photon is scattered within the cloud. Since the single-scattering albedo sca represents the fraction of the photons that are scattered, a sca=1 means that all photons are scattered and sca=0 means that all are absorbed.


Determine scattering direction of a photon

For a scattered photon the azimutal angle φ is in all direction equally likely, so that its calculation is simply : 

The scattering angle θ is a little bit more tricky, since not all direction are equally likely. By using the Henyey–Greenstein phase function, which has the so called asymmetry parameter g as single parameter, it is  possible to  incorporate a weighting difference between the forward and backward scattering direction. Thereby the asymmetry parameter g, which  is the mean cosine of the scattering angle, determines the weighting of the scattering directions (g=1 for just forward scattering, g=-1 for just backward scattering and g=0 for equal likely scattering directions). 

Through following equation it is possible to draw the  scattering angle θ

These angles just give the change of direction in the local coordinate system of each photon. To set it in a global scale a transformation as follows is necessary:

Determine reflection from ground 

Whether  a photon - that is ejected at the bottom of the cloud - is reflected or not, is determined by the albedo of the ground. The albedo parameter ag basically represents the fraction of the photons that are going to be reflected. Hence it is between ag=0 (no reflection) and ag=1 (always reflection).

Since there is no information about the surface incorporated, the azimuthal angle φ as well as the polar  angle θ for reflection at the ground are purely randomly distributed:

 

 


Implementation

The programming framework root 6 was used to implement the simulation and to display its results. As indicated above the simulation depends strongely on random number generation. In this case the TRandom3 random number generator for drawing uniform distributed numbers was used. The program code has to be executed in a root 6 session with .x photon.C. Due to a basic menu structure (fig.1) it is possible to easily set cloud and initial parameters to start the simulation and to explore the outcome in several ways. Note that after drawing the result diagrams, it is possible to go back to the main menu by calling the function photon().

Figure 1: Main menu structure 


The following parameters control the simulation and can be changed from their default value interactively.

Table 1: Simulation Parameters.

Parameter
Description
Default value
nphotons 
Photon numbers 
100 
cloudheight 
Cloud height 
meanfreepath 
Mean free path 
0.02 
asymmetryparameter 
Asymmetry parameter 
0.85 
singlescatteringalbedo 
Single scattering albedo 
0.0 
incidentbeamangle 
Incident beam angle 
180 
groundalbedo 
Ground albedo 
0.0 

root file name 
"mc_photons.root" 


Data Description

The raw data  is stored as a root tree and is available in a root file for further analysis. Note that the data is stored as calculated, that means, all iteration steps (over all photons) are depicted. Hence,  the variables after a scattering event x are shown.

In addition, the root file contains the run parameters, the calculated results as well as all plots that were viewed with this data set. The complete file structure is shown in table 2:

Table 2: Root file structure.

Tree
Branch
Description
tdata 
photon
Contains photon number
px,py,pz
Photon position in cartesian coordinates
nbottom
Number of photons ejected at cloud bottom
ntop
Number of photons ejected at cloud top
nreflected
Number of photons reflected at ground
nabsorbed
Number of photons absorbed within cloud
iscrossingtop
Photon crossed cloud top
iscrossingbottom
Photon crossed cloud bottom
dirx,diry,dirz
Photon direction in cartesian coordinates
travellingdistance
Photon traveling distance before scattering event
deviationangle
Scattering deviation angle at scattering event
scat_event
Number of scattering event

tresults
downirradiance
Calculated downward irradiance (normalized)
upirradiance
Calculated upward irradiance (normalized)
total_travellingdistance
Sum of all travelling distances over all photons
total_deviationangle
Sum of all deviation angles over all photons
mean_travellingdistance
Mean travelling distance of all photons
mean_deviationangle
Mean deviation angle of all photons

tparameter
nphotons
Calculation settings: photon numbers
cloudheight
Calculation settings: cloud height
meanfreepath
Calculation settings: mean free path
asymmetryparameter
Calculation settings: asymmetry parameter
bottomalbedo
Calculation settings: bottom albedo
singlescatteringalbedo
Calculation settings: single-scattering albedo
incidentbeamangle
Calculation settings: incident beam angle

Results

The main interest of the simulation is the irradiance that is transmitted and reflected by the cloud. Both irradiances are displayed right after each simulation run. However, beside these results, the simulation yields information on the photon behaviour inside the cloud. For instance the traveling distance or the deviation angle between two successive scattering events. Since several photons are calculated with each simulation run, it is feasible to display histograms on these variables. Such histograms on the traveling distance and the deviation angles are shown in fig. 2 (bottom). In addition, the figure displays the z-distribution of scattering events (top right) and provides information on how many photons are absorbed in the cloud, ejected on the top (=reflected back up), ejected on the bottom (=transmitted) and reflected back from the ground (top left; notice that the set 'reflected back from the ground' is a subset of 'ejected on the bottom'). 

Figure 2: Photon histograms.

The program simulates the path of a photon through the cloud and it is possible to display the trajectories of single photons as shown in fig. 3. 

Figure 3: Photon trajectories.


In most cases a photon will get ejected on the top or on the bottom of the cloud. An overview where this ejection takes place is provided by two polar plot representing the cloud top and bottom surface (fig. 4). The incident beam hits the cloud top in the center. 

Figure 4: Photons at cloud surfaces.


Basic information of the amount of scattering events per photon as well as an quick impression of the x-y- and z-distribution of the scattering events are provided in fig. 5.

Figure 5: Photon scattering.

Literature

Bohren, Craig F. & Clothiaux, Eugene E. (2006): Fundamentals of Atmospheric Radiation. An Introduction with 400 Problems. Weinheim: Wiley.