SPARC CCMVal PhotoComp-2008

 

GOALS. Evaluate how models calculate photolysis (and indirectly heating) rates in the stratosphere and troposphere with the incentive of locating errors or biases and identifying improved and practical methods. There are three basic parts to PhotoComp2008:

(1) Basic test of all J- values for high sun (SZA=15º), w/ & w/o additional scattering layers (stratiform clouds & stratospheric volcanic aerosols).

            (2) Test of twilight, sphericity, and 24-hour averages (SZA = 84º - 96º).

(3) Test of wavelength integration w/o scattering (SZA = 15º).

There will be one standard atmosphere, whose primary definition will include air mass, ozone mass, and temperature in each layer.  This atmosphere is typical of the tropics, ozone column = 260 DU.  For efficiency, we will use this same atmosphere in all sections, even the low-sun, polar cases. 

 

PARTICIPATION.  This study is designed to aid development and testing of the photolysis and short-wave heating codes used in chemistry-transport models and coupled chemistry-climate models.  This project is open: any research group can participate by running the experiments and reporting the results as specified below.  We also encourage participation from groups (without CTMs or CCMs) who have participated in other model-measurement studies (e.g., IPMMI, POLARIS).  Many CTM/CCMs will be using “the same” photolysis scheme (e.g., fast-TUV, fast-J) and think their participation redundant – this is false.  The implementation of a standard scheme into any CTM/CCM will likely alter (intended or inadvertent) how the J-values are calculated:  thus it is very important when you perform these tests that the photolysis module that is as close a possible to that embedded within the CTM/CCM and not the original, standalone version that you used to derive your inline model.

 

EXPERIMENTS.

 

Part 1 is a basic test of all J-values for high sun (SZA = 15º) over the ocean (albedo = 0.10, Lambertian).

  • Part 1a: Clear sky (only Rayleigh scattering) and no aerosols.
  • Part 1b: Pinatubo aerosol in the stratosphere (layer 10).
  • Part 1c: Stratus cloud (layer 2).

The primary atmosphere (Table 1a) is specified in terms of pressure layers, mean temperature, and column O3 in each layer.  Please do not include absorption by NO2 or other species in calculating optical depths.  For 1b and 1c we recommend that you use the specified optical properties in Table 1c, interpolating across the 5 specified wavelengths.

 

Part 2 tests the simulation of a spherical atmosphere and twilight conditions that are critical to the polar regions.  Use the same atmosphere as Part 1 without clouds or aerosols.  Assume equinox (solar declination = 0º) and a latitude of 84ºN.  The surface SZA (not including refraction) varies from 84º (noon) to 96º (midnight).  Report all J-values at noon, midnight, and the 24-hour average (integrating as you would in your CTM/CCM).  With a spherical atmosphere, the local solar zenith angle changes with altitude and if refraction is included it will change the surface angle.  Please note how you treat the solar ray path in your model description.

 

Part 3 tests the accuracy of wavelength binning in the critical region 290-400 nm that dominates tropospheric photolysis.  Shut off all Rayleigh scattering and surface reflection (albedo = 0) giving effectively a simple Beer’s Law calculation.  Repeat the calculation in Part 1, but report only J-values for J-O3 (i.e., total), J-O3(1D) [O3 => O2 + O(1D)], and J-NO2 [NO2 => NO + O].  These are the two critical J-values for the troposphere, and they both have unusual structures in absorption cross section and quantum yields.  The organizers will make these calculations using very high resolution (0.05 nm) cross sections and solar fluxes and for different options (e.g., JPL-06 vs. IUPAC cross sections) to provide a benchmark.  NOTE that we will only use results below 20 km (L=1:11) for this comparison.

 

 

DIAGNOSTICS.

 

Model Documentation should include a brief outline of the methods and any references (limit: one page). Please include brief notes on:  how you treat sphericity and refraction, the Schumann-Runge bands (J-O2 and J-NO), Rayleigh scattering, multiple scattering, clouds and aerosols, seasonal changes in sun-earth distance, solar variability, and any specific parameterizations.   Default cross sections are JPL-2006, please note if you are using alternate.

 

Report all J-values and all standard model layers since this is a check on all modeled J-values, not just the radiative transfer solution.  See Examples or web PDF file for data formatting.  We are not specifying the day-of-the-year, so use solar fluxes for sun-earth distance = 1.0 au and average over the 11-yr solar cycle if possible.  UCI's high-resolution solar spectrum used in these experiments is the average of two high and low SUSIM spectra (29 Mar 1992 and 11 Nov 1994), this is not meant to be the 11-yr average.  It will be provided at 0.05 nm resolution, but we encourage you to use your own solar fluxes for the primary tests since changing solar fluxes will mostly likely require a complete re-averaging of all cross-sections (see Fast-J paper, Wild, Zhu, Prather, 2000).  Please report in model documentation what you are using for the solar spectrum and how the solar cycle is represented in your submissions, and if possible submit it as a separate file so that it may be used to address differences later. (With different wavelength binning, this will not be trivial.)   Reported photolysis rates should be calculated for the mass mid-point of each layer, this brings PhotoComp closer to current CTM usage rather than the original grid-point formulation used in M&M.  Results in the form of clearly labeled ascii text files should be uploaded to the BADC CCMVal web page or emailed to organisers (see web posting for specific details).

 

 

DISCUSSION.

 

Implementation into a particular model's code will up to the participant.  For example,  UCI has two models that they will use in PhotoComp:  a fast-JX model within the CTM that uses layers of uniform composition defined by mass (kg/m2); and a stand-alone photochemical box model that defines altitude (in cm) as the vertical grid and uses number densities for air and ozone.  For the latter, we have re-mapped the primary atmosphere (Table 1a) onto a grid-point structure (Table 1b) that has the same mid-layer properties as the layer mean value and the same columns of O2 and O3. 

One question will be:  What is the correct answer?  In some cases we may be able to define a "best" answer based on obvious physics or convergence of some of the more resolved models, but in others we may not.  Thus in all of our proposed experiments we will begin with a "standard model” result (not necessarily the best answer) from one of the models and then determine a best answer, if possible, after analysis of the results.

 

One approach to defining the correct answer would be to merge observed radiation fields or photolysis rates (e.g., IPMMI, POLARIS, see references below), but we feel this may be too difficult to match the exact observing conditions. One way to include the knowledge gained by these field studies is to ensure participation from some of the models (e.g., NCAR-TUV, APL).

 

We do not recommend reporting detailed actinic fluxes as a function of wavelength since everyone selects different ways of integrating over wavelength (e.g., bins) and trying to reconcile the different wavelength scales is not worthwhile.  If major problems show up, then a subgroup of models can consider how to resolve the differences.

 

Another major issue with photolysis and heating rates is the treatment of clouds and cloud fraction. This is very important, but probably beyond the current PhotoComp. It would require a special workshop. We do include an option for a plane-parallel volcanic aerosol layer (aka Pinatubo) and a stratiform cloud.