An Application for Improving Air Quality (a Houston Case Study)

In this work, we focus on the surface layer scheme, which provides input to the land and surface sub-models to calculate surface heat, momentum and moisture fluxes that drive the planetary boundary layer schemes that determine near surface wind speeds.

Pius Leea , Fantine Ngana,b, Hang Leia,c, Barry Bakera,d , Bright Dornblasere, Gary McGauheyf,and Daniel Tonga,b,c
aAir Resources Laboratory, NOAA Center for Weather and Climate Prediction, College Park, Maryland
bCooperative Institute for Climate and Satellite, University of Maryland, College Park, Maryland
cCenter for Spatial Information Science and Systems, George Mason University, Fairfax, Virginia
dDepartment of Physics, University of Maryland, Baltimore County, Maryland
eTexas Commission on Environmental Quality, Austin, Texas
fTexas Air Quality Research Program, University of Texas, Austin, Texas

Abstract

Near surface level winds present a challenge to the Weather Research and Forecasting (WRF) model especially in areas with complex terrain, heterogeneous land cover, and prevalent sea/lake and land breezes. Areas around the Gulf of Mexico, such as the Houston-Galveston-Brazoria (HGB) area, exhibit frequent positive biases in the modeled near surface wind speed at night, displacing air pollutant plumes from their actual locations. In this work, we focus on the surface layer scheme (SL), which provides input to the land and surface sub-models (LSM) to calculate surface heat, momentum and moisture fluxes that drive the planetary boundary layer (PBL) schemes that determine near surface wind speeds.  SLs have inherent uncertainties in their empirically derived parameters. A recently added SL option (Jiménez et al., 2012)[1] improves the prediction of friction velocity – an important input to PBL schemes. Further, we examine LSM options to use soil moisture observations to constrain heat and moisture fluxes in WRF. The SL option[1] together with the community NOAH Land-Surface Model (LSM) option-pair improves sensible heat flux prediction significantly and may lead to reductions in wind biases at night.

NOAH is derived from:  N for National Centers for Environmental Prediction and National Weather Service (NWS), O for Oregon State University (Department of Atmospheric Science), A for Air Force, and H for Hydrology Lab, (NWS).

Figure 1. WRF domain setup: 36-km (NA36), 12-km (SUS12) and 4-km (TX04).

Figure 1. WRF domain setup: 36-km (NA36), 12-km (SUS12) and 4-km (TX04).

1. Introduction

Air pollutants adversely impact human health and ecology. Air quality managers who regulate and design pollution abatement and adaptation policies depend on transport modeling of air pollutants to make informed decisions. A chemical transport model (CTM) is used to provide such information. A CTM depends on the mean wind and wind fluctuation predictions of a meteorological model, such as WRF[2], to determine advection and diffusion of air pollutants.

Wind prediction is a challenge for WRF where there is a considerable heterogeneity in topology and land use and is compounded by influences of lake/sea and land breezes. Air pollutant concentration prediction is even more difficult as it is affected by uncertainties in wind and turbulence as well as uncertainties in emissions and the chemistry and physics of chemical constituents in the atmosphere. In particular, air quality (AQ) modeling of Houston-Galveston-Brazoria (HGB) is difficult due to its frequent sea and land breezes, highly variable land use, and the clustering of pollutant emitters such as refineries and port facilities along the Houston Ship Channel.

HGB suffers from high mono-nitrogen oxides (NOx) and high volatile organic carbon (VOC) emission conditions and is subject to frequent strong solar flux. These conditions favor photolytic production of surface O3. Intensive efforts in the last decade have been made to improve the AQ in this area. In 2012, HGB achieved an upgrade from being a “non-attainment zone” to a “marginally non-attainment zone” under the 2008 National Ambient Air Quality Standards (NAAQS) for ozone. Sophisticated meteorological and AQ monitoring networks were established in HGB to survey AQ conditions continually. HGB was also a focus of the 2006 Air Quality Study (TexAQS II). We take advantage of the Continuous Air Monitoring Site (CAMS) observational network in HGB and the TexAQS II field campaign including data from a coastal site[3] to study the modeled wind speed biases in the area.

aWRF Single-Moment 5-class. bWRF Single-Moment 6-class. cKain and Fritsch scheme. dRapid Radiative Transfer Model scheme. eYonsei University scheme. fGrell et. al.[8], g5-layer soil thermal diffusion model[8].

aWRF Single-Moment 5-class. bWRF Single-Moment 6-class. cKain and Fritsch scheme. dRapid Radiative Transfer Model scheme. eYonsei University scheme. fGrell et. al.[8], g5-layer soil thermal diffusion model[8].

The WRF model frequently overestimates near surface wind speeds at night in HGB[4, 5]. This has important consequences on inaccurate predictions of transport and concentration of air pollutants[6, 7]. In the WRF model, parameterizations of the SL and PBL schemes are directly responsible for the determination of the near surface wind speeds. However, the characteristics of the wind speed biases in the region require investigation of the tunable parameters in the modeling of surface winds. The Monin-Obukhov SL (MO SL) similarity scheme is used in the WRF model to provide a linkage between the model’s surface and PBL parameterization schemes. MO SL governs the calculation of friction velocity and exchange coefficients for momentum, heat and moisture. These parameters enable the calculation of the surface heat and moisture fluxes in the LSM that feed the PBL schemes as bottom boundary conditions (BC) and thus contribute to the PBL schemes’ accuracies in predicting near surface winds. The SL schemes have a few empirically determined tunable parameters which lend themselves to further investigation.

2. Data and methods

WRF version 3.4.1, released in summer of 2012, was used in this study. The model configuration (Figure 1 and Table 1) and verification data (Figure 2) are as described by Ngan et al.[6].

Four simulations were performed for the study period between June 4 and June 13, 2006. Table 1 summarizes the WRF physics options used in the Slab1 Case (Table 2). Slab1 uses the Penn State-National Center for Atmospheric Research (NCAR) fifth-generation Mesoscale Model (MM5) SL scheme and the MM5 5-layer thermal diffusion LSM[8] – also abbreviated as the Slab LSM. Table 2 lists the four simulations described as follows: (Slab11) same as Slab1 but uses the modified MM5 SL scheme following Jiménez et al. (2012)[1]; (Noah1) same as Slab 1 but uses the NOAH LSM; and (Noah11) combines the modified MM5 SL and the NOAH LSM. The modified MM5 SL scheme following Jiménez et al. (2012)[1] expands the original scheme to account for atmospheric regimes for extremely unstable and extremely stable conditions according to Fairall et al. (1996)[9] and Cheng and Brutsaert (2005)[10], respectively. These two extensions and the lowering of friction velocity values[11] are improvements suggested by Jiménez et al. (2012)[1]. Cases Noah1 and Noah11 used the U.S. National Centers for Environmental Prediction; Oregon State University; Air Force; Hydrological Research Laboratory (NOAH)[12, 13] LSM illustrated potential improvements by studies[14, 15].

 

a Following Jiménez et al. (2012)[1] to accommodate extreme atmospheric regimes.

a Following Jiménez et al. (2012)[1] to accommodate extreme atmospheric regimes.

 

Figure 2. CAMS stations in TX04. Inset shows those in HGB.

Figure 2. CAMS stations in TX04. Inset shows those in HGB.

 

3. Diurnal distribution of low level wind biases and surface heat fluxes

All sensitivity cases captured the 2-meter temperature and the 10-meter wind directions well; i.e., within 15 percent and 20 percent biases, respectively.

3.1 Modeled 10-meter wind biases over HGB around sunset

Figure 3. Grid-cell value for 10-meter wind speed versus CAMs measurements in HGB for June 4-13, 2006, for (OBS) observation, and prediction when applying: (slab1), (slab11), (noah1), and (noah11) (see Table 2).

Figure 3. Grid-cell value for 10-meter wind speed versus CAMs measurements in HGB for June 4-13, 2006, for (OBS) observation, and prediction when applying: (slab1), (slab11), (noah1), and (noah11) (see Table 2).

Figure 3 shows significant positive biases at night for all cases. At 21:00 CST, positive biases of 0.5 m s-1, 0.7 m s-1, 0.9 m s-1 , and 1.2 m s-1 are noted against the observed wind speed of 2.4 m s-1, for the Slab1, Slab11, Noah1, and Noah11 Case, respectively.  Although the Noah1 and Noah11 cases aggravated the modeled wind biases by as much as 0.4 m s-1, to 0.5 m s-1 compared to the corresponding cases of Slab1 and Slab11. Nonetheless, the Noah1 and Noah11 cases significantly reduced the unobserved abrupt drop in wind speed between 18:00 and 19:00 CST.

Figure 4 shows 10- meter wind speed over the University of Houston Coastal Center (UHCC) site.  It demonstrates that the Noah1 and Noah11 cases performed slightly worse than that by the Slab cases. All cases did rather well in reconstruction the wind speed time revolution during the day. On the contrary, obvious distinction for night-time performance is noticed. Between 19:00 CST and 3:00 CST the next morning the Slab1 and Slab11 predictions are tens of percentage points better.

3.2   Modeled sensible heat flux

Figure  shows substantial (up to 50 percent) reduction in mid-day positive biases in sensible heat flux when replacing the Slab LSM with the NOAH LSM. This is attributable to the more physically based parameterizations in land use, hydrological processes and evapotranspiration in the NOAH LSM.  Soil moisture data assimilation available in NOAH LSM and utilizing the North American Land Data Assimilation System (NLDAS) can constrain the heat and moisture fluxes to improve prediction of vertical distribution of momentum and thermal gradients.

4. Summary and future work

Figure 4. Modeled 10-meter wind speed versus measurement at the UHCC station for June 4-13, 2006, for (OBS) observation, and prediction when applying: (slab1), (slab11), (noah1), and (noah11).

Figure 4. Modeled 10-meter wind speed versus measurement at the UHCC station for June 4-13, 2006, for (OBS) observation, and prediction when applying: (slab1), (slab11), (noah1), and (noah11).

In this work, we investigated and aimed to remedy the WRF modeled wind speed positive biases of nocturnal near surface winds in HGB. The SL and PBL modules in WRF are directly attributable to wind speed prediction. The modeled wind speed biases on clear sky days such as the studied June 2006 period in HGB suggested discrepancy in the surface layer decoupling around sunset. This inaccuracy was likely a cause for the wind speed biases. At sunset, the rapid collapse of the daytime PBL and formation of the shallow nocturnal boundary layer (NBL) results from cooling of the surface by long-wave radiation and alters the distribution of momentum and heat. The remnant turbulence from the convective daytime PBL forms the residual layer aloft and detaches itself from the NBL below. The LSM-SL-PBL triplet with each of its components contributes strongly to the forcing of the PBL processes.

In this study, the modified MM5 SL scheme following Jiménez et al. (2012)[1] with applicability to a wider range of stability regimes was used to replace the original MM5 SL scheme in a series of sensitivity simulations. The modified MM5 SL scheme did not improve the wind biases but significantly improved the friction velocity biases. Furthermore, the sensitivity runs also included replacement of the Slab LSM with the NOAH LSM. The NOAH LSM did not improve the wind biases but improved its diurnal evolution tendency —  the erroneous abrupt drops in wind speed prediction between 18:00 and 19:00 CST were largely removed. The Noah cases also significantly reduced mid-day sensible heat flux positive biases. The more physically base Noah11 Case seemed to perform better compared to the other cases. The effort of this study may translate into better spatial simulation of O3 and its precursors – thus helping air quality managers.

Figure 5. Modeled sensible heat flux versus measurement at the UHCC station for June 4-13,  2006, for (OBS) observation, and prediction when applying: (slab1), (slab11), (noah1), and (noah11).

Figure 5. Modeled sensible heat flux versus measurement at the UHCC station for June 4-13, 2006, for (OBS) observation, and prediction when applying: (slab1), (slab11), (noah1), and (noah11).

Gupta et al. (1999)[16] pioneered a multi-criteria optimization methodology to derive optimally tuned parameters for a LSM for a global model. Sen et al., (2001)[17] expanded the methodology with a preference hierarchy based multi-criteria calibration algorithm. These methodologies optimize tunable parameters in the NOAH LSM and SL schemes to resolve a significant part of the modeled wind biases. Secondly, more utilization of observed soil-moisture such as the hourly NLDAS data can improve modeling of the PBL diurnal evolution.

5. Acknowledgements

We thank the Texas Commission for Environmental Quality (TCEQ) for the CAMS data and advice on the customary selection of WRF physics options used by the agency. We thank Dr. Mark Estes of TCEQ for many insightful discussions. We thank the NASA Air Quality Applied Sciences Team program for providing impetus and insights for improving air quality modeling using continuous measurements.

6. References

[1]    Jimenez, P., J. Dudhia, J. Gonzalez-Rouco, J. Navarro, J. Moneavez, and E. Garcia-Bustamante, 2012: A revised scheme for the WRF surface layer formulation, Monthly Wea. Review, 140, 898-918.
[2]    Skamarock, W. C., and Coauthors, 2008: A description of the advanced research WRF version 3. NCAR Tech Note NCAR/TN-475+STR, 125 pp.
[3]    Rappengluck, B., R. Perna, S. Zhong, G. A. Morris, 2008: An analysis of the vertical structure of the atmosphere and the upper-level meteorology and their impact on surface ozone levels in Houston/TX, J. Geophys. Res., 113, D17315, doi:10.1029/2007JD009745.
[4]    Hu, X. M., J. W. Nielsen-Gammon, F. Zhang, 2010: Evaluation of three planetary boundary layer schemes in the WRF model, J. Appl. Meteor. and Climate, 49, 1831-1844, DOI: 10.1175/2010JAMC2432.1
[5]    Lee, S. H., S. W. Kim, W. M. Angevine, L. Bianco, S. A. McKeen, C. J. Senff, M. Trainer, S. C. Tucker, and R. J. Zamora, 2010: Evaluation of urban surface parameterizations in the WRF model using measurements during the Texas Air Quality Study 2006 field campaign, Atmos. Chem. Phys., 11, 2127-2143.
[6]    Ngan, F., D. W. Byun, H. C. Kim, D. G. Lee, B. Rappenglueck and A. Pour-Biazar, 2012: Performance assessment of retrospective meteorological inputs for use in air quality modeling during TexAQS 2006. Atmos. Environ., 54, 86-96.
[7]    Lee, P.; F. Ngan, 2011: Coupling of Important Physical Processes in the Planetary Boundary Layer between Meteorological and Chemistry Models for Regional to Continental Scale Air Quality Forecasting: An Overview. Atmosphere, 2, 464-483.
[8]    Grell, G.A., Dudhia, J., Stauffer, D., 1994. A description of the fifth-generation Penn State/NCAR mesoscale model (MM5), NCAR Technical Note: NCAR/TN-398bSTR.
[9]    Fairall, C. W., E. F. Bradley, D. P. Rogers, J. B. Edson, and G. S. Young, 1996: Bulk parameterization of air-sea fluxes for Tropical Ocean Global Atmosphere Coupled-Ocean Atmosphere Response Experiment. J. Geophys. Res. 101(C2), 3747-3764.
[10]    Cheng, Y., and W. Brutsaert, 2005: Flux-profile relationships for wind speed and temperature in the stable atmospheric boundary layer. Bound.-Layer Meteor., 114, 519-538.
[11]    Shin, H. H., S. Y. Hong, 2011: Inter-comparison of planetary boundary layer parameterizations in the WRF Model for a single day from CASES-99, Bound.-Layer Meteor., 139, 261-281
[12]    Ek, M. B., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, V. Koren, G. Gayno, J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model, J. Geophys. Research, 108 D22, DOI: 10.1029/2002JD003296.
[13]    Chen, F. and J. Dudhia, 2001: Coupling and advanced land surface-hydrology model with the Penn State-NCAR MM5 modeling system. Part I: model implementation and sensitivity. Mon. Wea. Rev., 129, 569-585.
[14]    Yerramilli, A., V. S. Challa, V. B. R. Dodla, H. P. Dasari, J. H. Young, C. Patrick, J. Baham, R.  L. Hughes, M. G. Hardy, and S. J. Swanier, 2010: Simulation of surface ozone pollution in the Central Gulf Region using WRF/Chem Model: Sensitivity to PBL and land surface physics, Advances in Meteor. ID 319138, pp24.
[15]    Wilczak, J. M. and co-authors, 2009: Analysis of regional meteorology and surface ozone during the TexAQS II field program and an evaluation of the NMM-CMAQ and WRF-Chem air quality models, J. Geophys. Research, 114, doi:10.1029/2008JD011675.
[16]    Gupta, H. V., L.A. Bastidas, S. Sorooshian, W. J. Shuttleworth, and Z. L. Yang (1999): Parameter estimation of a land surface scheme using multicriteria methods, J. Geophys. Res., 104, No. D16, 19491-19503.
[17]    Sen, O. L., L. A. Bastidas, W. J. Shutteworth, Z. L. Yang, H. V. Gupta, and S. Sorooshian (2001): Impact of field-calibrated vegetation parameters on GCM climate simulations, Q.J.R. Meteorol. 127, 1199-1223.

Topic: ,