Seasonal biases of the GEOS5-generated O3 LBCs

[JianpingHuang-NOAA]

@bbakernoaa @ytangnoaa
Currently we face a pressing issue on overprediction of O3 in winter and underprediction in summer. There are two versions of GEOS5-simulations-generated monthly mean LBCs available for supporting the AQM v7.0 implementation.

1. V1: with the 100-ppbv maximum constraint on O3 LBCs. This may cause under-prediction of O3 near surface over CONUS domain in summer time, which is mainly because of too low background O3 (i.e., less than 30 ppbv). Below is an example from the retro run with C15 configuration on August 1 at 12z at the 7th forecast hour.

Meanwhile, it shows high O3 background (>50.0 ppbv) over the East Pacific which may cause over-prediction of O3 in winter months. Please see an example of O3 prediction on 20221127 at 12z cycle at the 7th forecast hour (from NRT run C15).

2. V2: with the 1000-ppbv maximum constraint on O3 LBCs, which causes overprediction of O3 near surface over CONUS domain.
   This may resolve the underprediction of O3 in summer but makes over-prediction even worse.

Based on our tests, the O3 LBC issue is not only caused by the maximum constraints but also by the background values of O3 in different seasons. The current LBCs have too high background values of O3 (>50 ppbv) in the lower troposphere in winter and too lower values in summer (<30 ppbv). Since the current operational LBCS are based on GEOS/Chem simulations (Harvard) in 2006 whereas the UFS-AQM uses the GEOS5-generated LBCs in 2013. Should we have a try to reuse GEOS/Chem simulations-generated LBCs? If this is not feasible, we need to have a pretest with one-week for each season with the current GEOS5-genreated LBCs to evaluate the model performance on O3-predictions. Based on that, we should have a better idea on how to address this LBC issue for O3. So, we need to address the seasonal biases of background values of O3 carefully before we start to do retro runs. Is it possible to get this done before Dec. 12, 2022?

Thanks,

Jianping

[bbakernoaa]

So, one thing to just clarify here. The AQMv6 operational NAQFC uses lateral boundary conditions from a 2006 GEOS-Chem simulation. UFS-AQM lateral boundary conditions are generated from a multi-year GEOS-CHEM simulation. @ytangnoaa can you clear this up with more details?

[bbakernoaa]

This is a plot of the o3_left lateral boundary condition (going from the top of the atmosphere to the bottom) and the o3mr_left (which is the gfs o3 concentrations)

Zooming into the troposphere

The new lateral boundary conditions have a maximum value of 1 ppm

[ytangnoaa]

UFS-AQM LBC uses GEOS-5 lateral boundary conditions. This LBC works OK for summertime, but has high ozone biases during winter. It is the issue of the global model's seasonal bias. I am looking for alternative global models, such as GFDL's AM4.

So, one thing to just clarify here. The AQMv6 operational NAQFC uses lateral boundary conditions from a 2006 GEOS-Chem simulation. UFS-AQM lateral boundary conditions are generated from a multi-year GEOS-CHEM simulation. @ytangnoaa can you clear this up with more details?

[ytangnoaa]

The capped ozone value was reduced to 0.1 ppm (version1)

/scratch2/NCEPDEV/naqfc/RRFS_CMAQ_NA13km/LBCS/RRFS_NA13km_GEOS5_v1

However, it still has the low-altitude high ozone, ~ 60 ppb, over Northern Pacific and Arctic

[ytangnoaa]

I processed the AM4 data, and it o3 north LBC over Arctic region is lower for the altitude below 2km.
https://docs.google.com/presentation/d/1dvD6U0sIXmvwY-vxl6URaqURic5mHL_t/edit?usp=share_link&ouid=115981223134185527389&rtpof=true&sd=true

However, due to the shallow vertical structures, the higher O3 could be brought down.

Jianping, could you help to quick test in your setting?

The AM4 LBC files can be found in
Hera:/scratch2/NCEPDEV/naqfc/RRFS_CMAQ_NA13km/LBCS/RRFS_NA13km_AM4_v1/

[ytangnoaa]

@JianpingHuang-NOAA If the September ozone overprediction issue is persistent, you can try the manually adjusted O3 LBC based on AM4
Hera: /RRFS_CMAQ_NA13km/LBCS/RRFS_NA13km_AM4_v1/am4_bndy.c793.201909-o3-adjust1.v1.nc

which uniformly minus 10 ppbv ozone for west and north boundaries.

[BrianCurtis-NOAA]

What is the latest on this issue? I will attach it to the AQM v7 project, but let me know if it's addressed or can wait until AQM v8.

[bbakernoaa]

What's going on with this?

[JianpingHuang-NOAA]

The issue remains. We need to revisit this issue for AQMv8.0