Code available for simulating tropospheric NO2 columns!

Commit by June 16, 2022

Code refactoring for transport error estimates and sounding selections for non-CO2 species.

:pushpin: Sounding selections:

• nf_dlat and nf_dlon: define the NEAR-FIELD regions around site

• dlat and dlon: define the ENTIRE target regions around site

• num_nf_lat * num_nf_lon: total # of soundings to be selected along either the latitudinal or longitudinal direction (e.g., ~7 * 14 total soundings over the NEAR-FIELD domain around site)

• num_bg_lat * num_bg_lon: total # of soundings to be selected along either the latitudinal or longitudinal direction (e.g., ~3 * 3 total soundings over the FAR-FIELD or BACKGROUND domain around site)

• get.recp.sensorv2(): function for the selection of satellite soundings to save computational cost; If selected near- or far-field soundings share the same coordinate, only one of them will be treated as model receptor.

Figure - An example of sounding selections (though not highly recommended for TROPOMI given its coarse resolution).

Commit by July 6, 2021

Changes to 'merge_stilt-hysplit' branch towards the release of X-STILT version 2 that can work with non-OCO sensors or without the dependence of satellite data:

:pushpin: X-STILT can now work with TROPOMI X-gas data, such as

• footprint weighting using TROPOMI species-specific averaging kernels;

• functions to grab vertical column density and convert to vertical mixing ratio in ppb;

• need to specify TROPOMI info via obs_sensor and obs_species that are currently limited to CO, CH4, and NO2;

:pushpin: IDEAL simulations for column footprints/enhancements from any locations (no requirement of satellite data).

• To use the ideal mode, obs_sensor needs to be set as NA;

• Column AK is simply treated as 1 from TOA to the surface (ak_wgt is forced as FALSE);

• Column PWF is calculated using modeled variables/profiles (pwf_wgt = TRUE).

• Must provide a .csv/.txt file containing the latitudes (lati) and longitudes (long) of your desired receptor locations. Receptor time (in form of YYYYMMDDHH or YYYYMMDDHHss) can either be included in the csv/txt file as the 3rd column (with name of time) OR assigned to timestr in run_xstilt.r. See example of receptor_demo.csv.

• [USE WITH CAUTIONS] NOT suitable for comparisons with real satellite sensors that have variable vertical sensitivities.

:pushpin: Built upon the latest HYSPLITv5: Replace discrete release levels with line source release -- i.e., particles are now evenly distributed between minagl and maxagl, e.g., default is 3000 particles from 0 - 3km;

:pushpin: PWF calculations in r/src/recp_trajec_foot

• Instead of interpolate PWF from OCO-2 to each discrete release level :arrow_forward: calculate PWF based on dp and dry air column density (mol m-2) between particles.

• To calculate the dry air X-density, we extracted specific humidity and temperature profiles via get.met.vars().

• If ak.wgt is turned off (ak.wgt == FALSE), the dependence on any satellite sensors will be removed. That being said, normalied AK is 1 and PWF is calculated based on MODELED surface pressures instead of RETRIEVED surface pressures.

:pushpin: Other minor changes:

• Allow the calculation of another set of footprint with a different temporal resolution from the default run at the same time, e.g., hourly footprint implied by time_integrate2 = F;

• Add interactive mode for choosing site, overpass time, etc. One can modify the code to turn it off;

• Insert the google API as the first line of insert_ggAPI.csv that will be sourced by the model for geolocation and google map plotting;

• Get rid of the additional .rds files end with X_wgttraj.rds to reduce file sizes :arrow_forward: trajectory-level footprints after the AK and PWF weighting are now stored as additional columns in the initial *X_traj.rds files. Here are the descriptions of columns in output$particle from rds files:

  • foot_before_weight: footprint BEFORE AK and PWF weighting;
  • foot: footprint AFTER AK and PWF weighting;
  • pwf, ak.norm, ap, ak.pwf: computed pressure weighting and interpolated apriori and averaging kernel profiles from OCO-2/3;
  • wgt: final weighting factors used for scaling trajectory-level footprint based on flags of ak.wgt and pwf.wgt.