Intro to the CMAQ model (an Eulerian model).  Overview of smog chemistry. Discussion of chemical-reaction eqs. used in CMAQ.  Assignment of HW7 on smog, but using simplified chemistry.  

If time, discussion of the PPM method for advection.  Assignment of CMAQ Tech Manual discussions (to be presented on Wed). 

Hopefully every student has used the poll in Piazza to vote for the CMAQ topics that they would like to present.

If time, start a few of the student presentations on the CMAQ Tech Manual. (see info below for 27 Mar., and also see more details in Piazza.

1. Numerics, Grids, Chemical Solvers (1 person)
2. WRF-CMAQ and MPAS-CMAQ (1 person)
3. Boundary Layer Processes & Turbulent Diffusion (1 person)
4. Emission Processes (anthropogenic, biogenic, wildfire, windblown dust, sea spray) (2 people)
5. Plume-in-Grid (1 person)
6. Dry Deposition (2 people)
7. Clouds, Wet Deposition, Bi-directional exchange (1 person)
8. Radiation and photolysis (1 person)
9. Chemical mechanisms (gas, aerosols, multiphase) (3 people)
   

HOLIDAYS

# ATSC 595D Atmospheric Dispersion Modeling -
# Homework exercise on 1-D advection, demonstrating PPM method from CMAQ,
# and comparing it to two other advections schemes: forward in time; and Runge-Kutta.
# PPM is Piecewise Parabolic Method.  For details, see:
# Chapter 7 of the CMAQ Science Document:
#      https://www.cmascenter.org/cmaq/science_documentation/pdf/ch07.pdf
# and the journal paper by Colella and Woodward, 1984, J. Computational Physics,
#      vol 54, pages174-201.
# But the best documentation is the actual fortran code for the
#      horizontal advection subroutine "hadv" from
#      https://github.com/USEPA/CMAQ/archive/5.2.1.zip and look in this path:
#      CCTM/src/hadv/yamo/hppm.F .  This version was updated on March 2018.
#      You can find a link to this fortran code at the end of this assignment.
# R. Stull - 18 Aug 2018, 24 Mar 2024


# In this HW, a 1-D pollutant puff being advected in the x-direction by a constant
# wind u.  The linear advection eq. is ∂c/∂t = - u ∂c/∂c .
# The three schemes you will compare are :
#   (a) FTBS - Forward in time, Backward in space;
#   (b) RK3 - Runga-Kutte 3rd order (WRF-ARW version) in time, Centered in space; and
#   (c) PPM - Piecewise Parabolic Method (from Stull's R code, adapted from the original CMAQ fortran code).
#          Hint: for FTBS & RK3, see Stull's brief review of numerical methods for advection .
#          Hint: for PPM, see Stull's R code fragment, near the end of this asssignment


# Create the grid and initial conditions
imax = 1000             # number of grid points in x-direction
delx = 100.             # horizontal grid spacing (m)
delt = 10.              # time increment (s)
u = 5.                  # horizontal wind speed (m/s) is constant for this exercise

# 1) Calculate and display the Courant number (Cr = u * delt / delx) .  You can display it in the
# graph in the question (3) if you would like. Cr is one measure of numerical stability.

# 2) (a) Create initial concentration distribution in the x-direction.
# Assume that these "conc" variables are actually concentration anomalies relative to a large
# constant background state. Thus, it is OK to have negative values of "conc".
conc <- rep(0.0, imax)   # initialize "conc" anomaly to zero relative to the background concentration.
cmax = 10.0             # max initial "conc" (in arbitrary concentration units)
conc[100:150] <- seq(0., cmax, len = 51)    # insert left side of triangle
conc[150:200] <- seq(cmax, 0., len = 51)    # insert right side of triangle
conc[20:40] <- seq(0., -0.5*cmax, len = 21)    # insert left side of triangle
conc[40:60] <- seq(-0.5*cmax, 0., len = 21)    # insert right side of triangle

#   (b) Plot the initial conc distribution on a graph.

# 3) Also, on the same plot, show the ideal exact final solution, as given by
cideal <- rep(0.0, imax)   # initial conc of ideal background is zero
cideal[800:850] <- seq(0., cmax, len = 51)    # insert left side of triangle
cideal[850:900] <- seq(cmax, 0., len = 51)    # insert right side of triangle
cideal[720:740] <- seq(0., -0.5*cmax, len = 21)    # insert left side of triangle
cideal[740:760] <- seq(-0.5*cmax, 0., len = 21)    # insert right side of triangle

# 4) Use "forward in time, backward in space" (FTBS) to advect conc
# for the following number of time steps
nsteps = (imax - 300) / (u * delt / delx)
# and plot the resulting conc on the same graph.

# 5) Repeat steps (2-4) to re-initialize, but plotting on a new graph, and using the WRF-ARW version of
# RK3-in-time, Centered-in-space for the advection.  Use same number of time steps.

# 6) Repeat steps (2-4) to re-initialize, but plotting on a new graph, and using
# CMAQ PPM for the advection.  Use same number of time steps.
# Here is a code fragment in the R language that shows the PPM algorithm for our simplified
# homework problem.  You can use or modify it to do this PPM exercise.
# Thanks to previous students taking this class, they provide a python version of the PPM code. (as a pdf , or as a zipped .py file)
# [In case you are interested, here is the original CMAQ fortran code for PPM that
# Stull modified to create his version in R.]

# 7) Discuss and compare the results of these three advection schemes.

CMAQ installation, Part 1. Lead by our TA, Tim Chui.  (See instructions on our CMAQ web page, not the HW page.)

CMAQ installation, Part 2. Lead by our TA, Tim Chui. (Save the CMAQ run output to your own laptop.)

HOMEWORK: Use Panoply to display the Ozone output from CMAQ that you produced during class and saved to your own computer. (Here are the Panoply intro slides presented by Tim Chui.)
A. Produce a plot of ozone with the following characteristics:

• Map Projection: use stereographic map projection
• Use mouse to zoom to the region having data (command-click on Mac, control-click on Windows)
• Plot tab (from top of Panoply window) Size Ratio is 100% in both Width and Height, or smaller to fit on your screen.
• Grid: display lat and lon every 5° (I used dark green color for the grid)
• Overlays: add coastal and state boundaries
• Scale: use range of 0 to 0.08, tic label format %0G, tics 8 major 2 minor
• Scale: color table: Panoply_16.act
• Labels: Title = "<yourname> CMAQ Output"

This produces a plot similar to mine, but this is NOT what you will turn in. Instead, pick your own favorite color table, and make any other adjustments needed so that the legend color bar works with your color table. Feel free to make other enhancements. You will submit as HW your resulting still image for this assignment.

B. Using the same format that you created from A (easy to do if you kept your Panoply plot open), create an .mp4 movie animation output (see the Panoply file menu), with a frame rate of 1 fps and Content of TSTEP (7 frames). Also submit this animation as your HW. (As an example, here is my ozone animation output.)

10 Apr

Last Class day 

Catch up and Summary of the Course.

Meanwhile, finish working on your Homework on Advection (FTBS, RK3, PPM).  Write your own code for this in your favorite computer language.