Colorado State Universityviews.cira.colostate.edu/.../WSAQS...09Dec2014.docx · Web viewThe P-X uses 2-m temperature and relative humidity for indirect soil moisture and deep soil

Western-State Air Quality Modeling Study (WSAQS)

Weather Research Forecast (WRF)Winter Modeling Workplan

Prepared by:

University of North CarolinaInstitute for the Environment

100 Europa Drive, Suite 490Campus Box 1105

Chapel Hill, NC 27599-1105

ENVIRON International Corporation773 San Marin Drive, Suite 2115

Novato, California, 94945415-899-0700

December 2014

CONTENTS

1.0 INTRODUCTION 31.1 Background 31.2 Overview 31.3 Organization of the Modeling Workplan 4

2.0 MODEL SELECTION 52.1 Description of WRF 5

3.0 EPISODE SELECTION 63.1 Episode Selection Criteria 63.2 Episode Selection RESULTS 6

4.0 METEOROLOGICAL MODELING 104.1 Model Selection and Application 104.2 WRF Domain Definition 104.3 Topographic Inputs 104.4 Vegetation Type and Land Use Inputs 104.5 Atmospheric Data Inputs 134.6 Water TemPerature 134.7 Time Integration 134.8 Diffusion Options 134.9 Lateral Boundary Conditions 134.10 Top and Bottom Boundary Conditions 134.11 FDDA Data Assimilation 144.12 WRF Physics and INput Sensitivity 144.14 WRF OUTPUT VARIABLES 154.15 WRF application Methodology 16

5.0 METEOROLOGICAL MODEL PERFORMANCE EVALUATION 205.1 Meteorological Process Evaluation using Field Campaign Observations 215.2 Quantitative Evaluation using Surface Meteorological Observations 22

6.0 REFERENCES 26

TABLESTable 3.1. Intense Observations Periods (IOPs) during the PCAPS in the Salt Lake

basin. 8Table 4-1. 37 Vertical layer interface definition for WSAQS WRF meteorological

model simulations. 12Table 4-2. Physics options used in the initial 3SAQS WRF simulation that would be

used as starting configuration for WSAQS WRF runs. 17Table 4-3. Proposed WRF Wintertime Sensitivity Experiments 17Table 5-1. Meteorological model performance benchmarks for simple and

complex conditions. 24

WSAQS Draft WRF Winter Modeling Workplan

FIGURES

Figure 3-1. Time series of 8-hour average ozone concentrations at all monitoring sites in the Uintah Basin, winter 2012-2013. EPA NAAQS of 75 ppb is shown as a red dashed line. 9

Figure 4-1. Proposed WRF 36/12/4 km grid structure for the WSAQS meteorological. The domains chosen are identical to those used in the 3SAQS. 11

Figure 4-2. Comparison of snow cover from the NAM-12km (left) and SNODAS-1km (right). 18

Figure 4-3. Illustration of “iterative nudging” for the P-X land surface model. The P-X uses 2-m temperature and relative humidity for indirect soil moisture and deep soil temperature nudging. 18

Figure 4-4. 2-m temperature analysis used for soil nudging with the P-X land surface model with standard 2m temperature analyses (left) and iterative nudging (right). 19

Figure 5-1. Conceptual model of meteorological processes that influence cold air pool formation and demise. 20

Figure 5-2. Locations of MADIS surface meteorological modeling sites with the WRF 4-km 3SAQS modeling domain. 22

Figure 5-3. Example quantitative model performance evaluation display using soccer plots that compare monthly temperature performance (colored symbols) for the three states: Colorado (top left), Utah (top right) and Wyoming (bottom) with the simple and complex model performance benchmarks (rectangles). 25

WSAQS Draft WRF Winter Modeling Workplan

1.0 INTRODUCTION

1.1 BACKGROUND

The Western-State Air Quality Study (WSAQS) includes cooperators from U.S. Environmental Protection Agency (EPA), United States Forest Service (USFS), Bureau of Land Management (BLM), National Park Service (NPS), and the state air quality management agencies of Colorado, Utah, and Wyoming. The WSAQS is intended to facilitate air resource analyses for federal and state agencies in the states of Wyoming, Colorado, and Utah toward improved information for the public and stakeholders as a part of the National Environmental Policy Act (NEPA) and potentially other studies. Funded by the Environmental Protection EPA, BLM, and USFS and with in-kind support from NPS and Colorado, Utah, and Wyoming state air agencies, by working closely with cooperators and overseeing the various agreements, the main focus of the study is on assessing the environmental impacts of sources related to oil and gas development and production. In particular, the cooperators will use photochemical grid models (PGMs) to quantify the impacts of proposed oil and gas development projects within the WSAQS region on current and future air quality, including ozone and visibility levels in the National Parks and Wilderness Areas.

The WSAQS has hired the University of North Carolina (UNC) at Chapel Hill and ENVIRON International Corporation (ENVIRON) to develop the technical data needed to perform the modeling for WSAQS as well as populate the Western Data Warehouse (WSDW). UNC/ENVIRON is working closely with the NPS and other cooperators to develop technical capacity and expertise, and train NPS staff.

A key component for the WSAQS PGM modeling is the meteorological inputs. Meteorological inputs for PGMs are generated using prognostic meteorological models. Wintertime ozone formation near oil and gas development areas in the Rocky Mountain region is driven by a unique combination of emissions, dynamical, and radiative conditions that are challenging to simulate in PGMs. The conventional WRF modeling configurations used to simulate summer ozone formation do not work well for wintertime modeling. While different groups have documented WRF configurations for simulating specific winter ozone episodes (e.g Ahmadov et al., 2014), there has not been an effort to develop and test a wintertime WRF configuration that yields acceptable performance across all of the Western states. This document is a workplan for developing and testing an operational WRF wintertime configuration for simulating the meteorological conditions conducive to ozone formation.

1.2 OVERVIEW

The pilot (2013 through September 2014) Three State Air Quality Study (3SAQS) performed meteorological modeling for the year 2011 using the Weather Research and Forecasting Model version 3.5.1 focused on the intermountain west states of Colorado, Utah, and Wyoming. The WRF 2011 meteorological fields were subsequently used in photochemical grid model (PGM) application and model performance evaluation (MPE). The MPE for 2011 identified areas of poor model performance within the study region (UNC and ENVIRON, 2014a,b). Of particular concern was the wintertime model performance in two locations in the intermountain west. High winter ozone events have occurred in the Jonah-Pinedale Anticline Development (JPAD) area in the Upper Green River area of southwest Wyoming and the Uinta Basin, Utah that are both rural oil and gas development areas. Lessons learned from the prior modeling sensitivities and input from the U.S. EPA, the University of Utah (UUtah), the Utah Department of Environmental Quality Division of Air Quality (UTDAQ), Wyoming Department of Environmental Quality (WDEQ) and others are being used to develop model sensitivities to improve the meteorological model performance during the winter months. In particular, UUtah and UTDAQ have tested various WRF model configurations to improve model performance

WSAQS 3 DRAFT Winter Modeling Workplan

in simulating cold air pools in the Uintah Basin and Salt Lake Valley, which are important contributors to poor air quality in oil and gas development regions (Neeman et al. 2014).

The inability to model or the misrepresentation of the cold air pools is a serious limitation for PGM studies investigating the air quality impacts of natural gas and oil development in the West U.S. In particular, snow cover is a key parameter that needs to be properly represented within a meteorological model for a number of reasons. An increase in snow cover increases the surface albedo and near-surface actinic flux leading to elevated photolysis rates (Schnell et al., 2009) favoring high near surface ozone concentrations. The increase in snow cover also favors a highly stable boundary layer. These conditions generate cold air pools that may persist for many days. Cold pool events are often associated with complexity in other synoptic and mesoscale atmospheric circulations, in particular warming in the free atmosphere associated with warm air advection and subsidence (Lareu et al. 2013). Other meteorological model deficiencies include sensitivity to initialization time (personal communication Erik Crossman and Lance Avery), representation of low cloud cover (Gultepe et al. 2014), and turbulence/mixing within the PBL. The WRF PBL schemes allow too much turbulent mixing in the winter favoring deeper than observed boundary layers (Baklanov et al. 2011; Holstag et al. 2013).

This document presents the meteorological modeling workplan to develop an operational winter modeling configuration that: 1) optimizes the WRF meteorological model performance for simulating winter ozone events across the three-state region and 2) tests the best winter model configuration for a summer month relative to the 2011 annual 3SAQS WRF model configuration to provide guidance as to the model configuration to use outside of the winter months. The WRF configuration improvements developed here will be implemented in an improved 2011 air quality modeling platform, especially during the winter months, and in a preliminary 2014 modeling platform.

1.3 ORGANIZATION OF THE MODELING WORKPLAN

The WSAQS winter meteorological modeling will leverage lessons learned from the meteorological modeling from the 3SAQS and input from other studies and scientists working on similar issues, in particular as part of studies of the winter ozone events in the Uinta Basin and JPAD areas. Details of the 3SAQS WRF modeling technical approach can be found in the 2011 WRF Modeling Protocol (ENVIRON and UNC, 2013) and in the 2011 WRF Application/Evaluation report (UNC and ENVIRON, 2014a). Although the WSAQS modeling analysis is not currently being performed to fill any particular regulatory requirement, such as a State Implementation Plan (SIP) attainment demonstration or an Environmental Impact Statement (EIS) or Resource Management Plan (RMP) as part of the National Environmental Policy Act (NEPA), it is being conducted with the same level of technical rigor as a SIP-type analysis and the databases developed may ultimately be used as a basis for regulatory air quality modeling. This WSAQS winter meteorological modeling workplan has the following sections:

1. Introduction : Presents an overview and objectives of the study.2. Model Selection : Introduces the model selected for the study.3. Episode Selection : Describes the modeling period for the study.4. Meteorological Modeling : Describes how the meteorological modeling will be conducted

including modeling domains, model options and application strategy.5. Meteorological Model Performance Evaluation : Provides the procedures for conducting the

model performance evaluation of the meteorological model.6. References : References cited in the document.


2.0 MODEL SELECTIONThis section discusses the selection of the Weather Research and Forecasting (WRF) meteorological model for the WSAQS. The selection methodology follows EPA’s guidance for regulatory modeling in support of ozone and PM2.5 attainment demonstration modeling and showing reasonable progress with visibility goals (EPA, 2007). EPA recommends that models be selected for regulatory ozone, PM and visibility studies on a “case-by-case” basis with appropriate consideration being given to the candidate models’:

Technical formulation, capabilities and features;

Pertinent peer-review and performance evaluation history;

Public availability; and

Demonstrated success in similar regulatory applications.

The 3SAQS modeling protocol (ENVIRON and UNC, 2013) provides an overview of WRF capabilities, performance, and success in prior regulatory applications.

2.1 DESCRIPTION OF WRF1

The non-hydrostatic version of the Advanced Research version of the Weather Research and Forecast (WRF-ARW2) model (Skamarock et al. 2004; 2005; 2006; Michalakes et al. 1998; 2001; 2004) is a three-dimensional, limited-area, primitive equation, prognostic model that has been used widely in regional air quality model applications. The basic model has been under continuous development, improvement, testing and open peer-review for more than 10 years. It has been used world-wide by hundreds of scientists for a variety of mesoscale studies, including cyclogenesis, polar lows, cold-air damming, coastal fronts, severe thunderstorms, tropical storms, subtropical easterly jets, mesoscale convective complexes, desert mixed layers, urban-scale modeling, air quality studies, frontal weather, lake-effect snows, sea-breezes, orographically induced flows, and operational mesoscale forecasting. WRF is a next-generation mesoscale prognostic meteorological model routinely used for urban- and regional-scale photochemical, fine particulate and regional haze regulatory modeling studies. Developed jointly by the National Center for Atmospheric Research and the National Centers for Environmental Prediction, WRF is maintained and supported as a community model by researchers and practitioners around the globe. The code supports two modes: the Advanced Research WRF (ARW) version and the Non-hydrostatic Mesoscale Model (NMM) version. WRF-ARW has become the new standard model used in place of the older Mesoscale Meteorological Model (MM5) for regulatory air quality applications in the U.S. It is suitable for use in a broad spectrum of applications across scales ranging from hundreds of meters to thousands of kilometers.

1 http://www.wrf-model.org/index.php 2 All references to WRF in this document refer to the WRF-ARW


http://www.wrf-model.org/index.php

3.0 EPISODE SELECTIONEPA’s ozone, PM2.5, and visibility SIP modeling guidance (EPA, 2007)3 contains recommended procedures for selecting modeling episodes, while also referencing EPA’s 1-hour ozone modeling guidance for episode selection (EPA, 1991)4. This Chapter presents the modeling period selected for performing the WSAQS and the justification and rationale for its selection.

3.1 EPISODE SELECTION CRITERIA

EPA’s modeling guidance lists primary criteria for selecting episodes for ozone, PM2.5, and visibility SIP modeling along with a set of secondary criteria that should also be considered.

3.1.1 Primary Episode Selection Criteria EPA’s modeling guidance (EPA, 2007) identifies four specific criteria to consider when selecting episodes for use in demonstrating attainment of the 8-hour ozone or PM2.5 NAAQS:

1. A variety of meteorological conditions should be covered, including the types of meteorological conditions that produce 8-hour ozone and 24-hour PM2.5 exceedances in the western U.S.;

2. Choose episodes having days with monitored 8-hour daily maximum ozone and 24-hour PM2.5 concentrations close to the ozone and PM2.5 Design Values;

3. To the extent possible, the modeling data base should include days for which extensive data bases (i.e. beyond routine aerometric and emissions monitoring) are available; and

4. Sufficient days should be available such that relative response factors (RRFs) for ozone projections can be based on several (i.e., > 10) days with at least 5 days being the absolute minimum.

3.1.2 Secondary Criteria EPA also lists four “other considerations” to bear in mind when choosing potential 8-hour ozone episodes including:

1. Choose periods which have already been modeled;2. Choose periods that are drawn from the years upon which the current Design Values are based;3. Include weekend days among those chosen; and4. Choose modeling periods that meet as many episode-selection criteria as possible in the maximum

number of nonattainment areas as possible.

EPA suggests that modeling an entire summer ozone season for ozone or an entire year for PM2.5 would be a good way to assure that a variety of meteorological conditions are captured and that sufficient days are available to construct robust relative response factors (RRFs) for the 8-hour ozone and PM2.5 Design Value projections.

3 http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rh-guidance.pdf 4 http://www.epa.gov/ttn/scram/guidance/guide/uamreg.pdf


http://www.epa.gov/ttn/scram/guidance/guide/uamreg.pdf

http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rh-guidance.pdf

3.2 EPISODE SELECTION RESULTS

To improve the WRF modeling performance for the wintertime conditions compared to the 3SAQS routine (i.e., summer configuration) 2011 WRF application, we selected two wintertime periods to test sensitivities with the WRF model: December 2010 – March 2011 (includes overlap from the 3SAQS) and December 2012 – March 2013. Extensive field measurement campaigns operated during these periods in both the Uinta Basin and JPAD region to characterize the meteorology, emissions, and the resulting air quality, especially during cold air pool events. We did not select the December 2011-Februrary 2012 period because during this period the ozone concentrations were well below the NAAQS and cold air pools were virtually absent in the Uintah Basin due to low snow cover (Utah DEQ, 2014). The field campaign measurements that we propose to use to help develop and evaluate the WSAQS WRF winter modeling configuration include, but are not limited to:

1) Persistent Cold Air Pool Study (PCAPS) within the Salt Lake basin5

2) Uinta Basin Winter Ozone Studies (UBWOS)6 3) Upper Green River Winter Ozone Study (UGRWOS)7

The measurements from these field campaigns will allow us to characterize a WRF configuration for two different winter seasons and three different basins. The range of high ozone events covered by these studies will provide a sufficient number of days to analyze the meteorological conditions that produce high winter ozone. During these periods, there are many known meteorological events that contribute to both high ozone and PM2.5. The UBWOS and UGRWOS focus on high ozone in the Uinta and JPAD regions, respectively. The PCAPS campaign includes a focus on high PM2.5 around Salt Lake City. Below we include a brief description of the field campaigns and select air pollution episodes captured in the measurements.

PCAPS is a three-year project to investigate the processes leading to the formation, maintenance and destruction of persistent cold air pools within the Salt Lake basin. The PCAPS study helped to identify processes that lead to and the buildup, maintenance and breakup of the cold air pools, the consequences of these processes on air pollution, and to improve meteorological models for more accurate simulations of these events. Table 3-1 shows the Intense Observational Periods (IOPs) from December 1, 2010 through February 7, 2011 for PCAPS. We will use the relevant observations within these episode periods to evaluate the meteorological model performance.

UBWOS was launched to gain a better understanding of wintertime ozone formation within the Uinta Basin. Some objectives of this field campaign included creating a long-term record of atmospheric chemistry and meteorology, and understand the meteorological and chemical processes active in wintertime ozone production in the Uinta Basin (Utah State, 2013). Observations are available for both of the winter seasons that we propose to model. In particular, 2013 experienced numerous 8-hr average ozone events that exceeded the NAAQS of 75 ppb for several days, as seen in Figure 3-1. Several of these events are persistent and begin in January 2013. We will use information on the timing of the high ozone events to identify meteorological episodes for model evaluation.

UGRWOS is a set of field studies initiated by the Wyoming Department of Environmental Quality (WDEQ) to better understand baseline and elevated wintertime ozone within the JPAD region. These

5 http://pcaps.utah.edu/6 http://www.deq.utah.gov/NewsNotices/annualreport/Planning/s12.htm. 7 http://deq.state.wy.us/aqd/Upper%20Green%20Winter%20Ozone%20Study.asp


field studies were initiated after the first measured high ozone values in 2005, with the area since being designated as “non-attainment” with respect to the national ozone standard.8 The UGRWOS field studies consist of various meteorological and air quality measurements. Of particular interest for this workplan are observed high ozone events in early 2011. February and March 2011 saw the highest 8-hr average observed ozone concentrations since 2007 with 26 days exceeding 65 ppb. We will use these high ozone events in February and March 2011 for selecting episodes to use for evaluating the WRF winter modeling.

Table 3.1. Intense Observations Periods (IOPs) during the PCAPS in the Salt Lake basin.9

IOP Dates Begin (DD-MM-YYYY) End (DD-MM-YYYY) CommentIOP1 01-12-2010 12:00 UTC 07-12-2010 02:00 UTC Long lived cold pool.

PCAP began on 29 Nov, before PCAPS operations began

IOP2 07-12-2010 12:00 UTC 10-12-2010 15:00 UTCIOP3 12-12-2010 12:00 UTC 14-12-2010 21:00 UTC Short and sweet."

Followed by great troughiness

IOP4 24-12-2010 12:00 UTC 26-12-2010 21:00 UTC "Short and sweet." again. The Christmas Cold Pool

IOP5 01-01-2011 00:00 UTC 09-01-2011 12:00 UTC Likely a biggieIOP6 11-01-2011 12:00 UTC 17-01-2011 20:00 UTCIOP7 20-01-2011 12:00 UTC 22-01-2011 06:00 UTCIOP8 23-01-2011 12:00 UTC 26-01-2011 12:00 UTC Somewhat uneventful,

observations were minimal

IOP9 26-01-2011 12:00 UTC 31-01-2011 06:00 UTC Very intensive operations, sidewall and canyon ops

IOP10 02-02-2011 18:00 UTC 05-02-2011 18:00 UTC Intensive lake based operations

8 http://rd.usu.edu/files/uploads/2012_summary_ugrb.pdf9 IOPs obtained from http://pcaps.utah.edu/about/status.php


http://pcaps.utah.edu/about/status.php

Figure 3-1. Time series of 8-hour average ozone concentrations at all monitoring sites in the Uintah Basin, winter 2012-2013. EPA NAAQS of 75 ppb is shown as a red dashed line.10

10 Figure 3-1 from http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-Final-Report/UBOS_2013Sec_3_DistribMon.pdf


http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-Final-Report/UBOS_2013Sec_3_DistribMon.pdf

http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-Final-Report/UBOS_2013Sec_3_DistribMon.pdf

4.0 METEOROLOGICAL MODELINGThe WRF meteorological model will be applied for the winter months of 2010-2011 and 2012-2013 using a 36/12/4 km domain structure. However, cold air pools can be forced by small-scale processes (e.g. turbulence; Laureau and Horel, 2014) and may require higher resolution simulations. We propose adding a higher-resolution modeling domain for a single configuration and episode if the 4-km simulation fails to improve the representation of cold air pool events. The higher resolution simulation will include a 1.33-km domain covering the oil and gas basins in northeast Utah, southwest Wyoming, and northern Colorado.

4.1 MODEL SELECTION AND APPLICATION

The latest publicly available version of WRF (version 3.6.1 released August 14, 2014) will be used in the WSAQS meteorological modeling. The WRF preprocessor programs including GEOGRID, UNGRIB, and METGRID will be used to develop model inputs.

4.2 WRF DOMAIN DEFINITION

The WRF computational grid was designed so that it can generate CAMx/CMAQ meteorological inputs for the same 36 km CONUS , 12 km western US, and 4 km three state domains as used in the 3SAQS. The WRF modeling domain was defined to be slightly larger than the CAMx/CMAQ PGM modeling domains to eliminate boundary artifacts in the meteorological fields used as input to CAMx/CMAQ. Such boundary artifacts can occur as the boundary conditions (BCs) for the meteorological variables come into dynamic balance with WRF’s atmospheric equations and numerical methods for downscaling from coarser to finer-resolution domains. Figure 4-1 depicts the WRF horizontal modeling domains to be used in the WSAQS. The outer 36 km continental U.S. (CONUS) domain (D01) has 165 x 129 grid cells, selected to be consistent with the existing Regional Planning Organization (RPO) and EPA modeling CONUS domain. The projection is Lambert Conformal with the “national RPO” grid projection center of 40o, -97o with true latitudes of 33o and 45o. The 12 km western U.S. domain has 256 × 253 grid cells with offsets from the 36 km grid of 15 columns and 26 rows. The 4 km three state domain has 301 x 361 grid cells with offsets from the 12 km grid of 75 columns and 55 rows.

The WRF model will use the same 37 vertical layers sigma-pressure surfaces as defined in the 3SAQS, Table 4-1. This vertical layer configuration considers approximately the same number of vertical levels in the lowest 200 meters as in other meteorological models studies resolving cold air pools (i.e. Neemann et al. 2014).

4.3 TOPOGRAPHIC INPUTS

Topographic information for the WRF modeling will be developed using the standard WRF terrain databases available from the National Center for Atmospheric Research (NCAR).11 The 36 km CONUS domain was based on the 10 min. (~18 km) global data. The 12 km western US domain will be based on the 2 min. (~4 km) data. The 4 km three state domain will be based on the 30 sec. (~900 m) data.

4.4 VEGETATION TYPE AND LAND USE INPUTSVegetation type and land use information will be derived from two different land use databases: the USGS 24-category and NLCD 2006 40-category databases. These land use databases are distributed with WRF version 3.6.1 and are proposed for consistency with other WRF modeling efforts within other groups, including the U.S. EPA and the UUtah. These efforts are described in more detail in Section 4.12.

11 http://dss.ucar.edu/


http://dss.ucar.edu/

The standard WRF surface characteristics corresponding to each land use category will be employed with the exception of snow cover. For snow, the albedo will be adjusted for each land use cover type. This adjustment is important because the surface albedo can depend on vegetation type/coverage. For example, conifer forests have a lower snow albedo because they consist of dark trees covering the high albedo snowpack. Grasslands have a high snow albedo because they could be completely covered by the snowpack. This level of detail can have important implications for the surface energy budget and dependent issues.

Figure 4-1. Proposed WRF 36/12/4 km grid structure for the WSAQS meteorological. The domains chosen are identical to those used in the 3SAQS.


Table 4-1. 37 Vertical layer interface definition for WSAQS WRF meteorological model simulations.

WRF Meteorological Model

WRF Layer Sigma

Pressure (mb)

Height (m)

Thickness(m)

37 0.0000 50.00 19260 205536 0.0270 75.65 17205 185035 0.0600 107.00 15355 172534 0.1000 145.00 13630 170133 0.1500 192.50 11930 138932 0.2000 240.00 10541 118131 0.2500 287.50 9360 103230 0.3000 335.00 8328 92029 0.3500 382.50 7408 83228 0.4000 430.00 6576 76027 0.4500 477.50 5816 70126 0.5000 525.00 5115 65225 0.5500 572.50 4463 60924 0.6000 620.00 3854 46123 0.6400 658.00 3393 44022 0.6800 696.00 2954 42121 0.7200 734.00 2533 40320 0.7600 772.00 2130 38819 0.8000 810.00 1742 37318 0.8400 848.00 1369 27117 0.8700 876.50 1098 17716 0.8900 895.50 921 17415 0.9100 914.50 747 17114 0.9300 933.50 577 8413 0.9400 943.00 492 8412 0.9500 952.50 409 8311 0.9600 962.00 326 8210 0.9700 971.50 243 829 0.9800 981.00 162 418 0.9850 985.75 121 247 0.9880 988.60 97 246 0.9910 991.45 72 165 0.9930 993.35 56 164 0.9950 995.25 40 163 0.9970 997.15 24 122 0.9985 998.58 12 121 1.0000 1000 0


4.5 ATMOSPHERIC DATA INPUTS

The WRF initialization fields will be taken from the 12 km (Grid #218) North American Model (NAM) archives available from the National Climatic Data Center (NCDC) NOMADS server.

4.6 WATER TEMPERATURE

The water temperature data will be taken from the NCEP RTG global one-twelfth degree analysis.12

4.7 TIME INTEGRATION

Third-order Runge-Kutta integration will be used (rk_ord = 3). The maximum time step, defined for the outer-most domain (36 km) only, should be set by evaluating the following equation:

dt= 6dxFmap

Where dx is the grid cell size in km, Fmap is the maximum map factor (which can be found in the output from REAL.EXE), and dt is the resulting time-step in seconds. For the case of the 36 km RPO domain, dx = 36 and Fmap = 1.08, so dt should be taken to be less than 200 seconds. Longer time steps risk CFL errors, associated with large values of vertical velocity, which tend to occur in areas of steep terrain (especially during very stable conditions typical of winter).

For the current modeling, adaptive time stepping will be used to decrease total runtime for the annual simulation. We suggest using target_cfl = 0.8, starting_time_step = -1, and min_time_step = -1 (for each domain). The max_time_step = 180, 60, 20 should be used, with the default values for max_step_increase_pct. Any initializations that fail due to CFL errors should be re-run with use_adaptive_time_step = .false., with time_step set to 180 or 120 seconds.

4.8 DIFFUSION OPTIONS

Horizontal Smagorinsky first-order closure (km_opt = 4) with sixth-order numerical diffusion and suppressed up-gradient diffusion (diff_6th_opt = 2) will be used.

4.9 LATERAL BOUNDARY CONDITIONS

Lateral boundary conditions will be specified from the initialization dataset on the 36 km domain with continuous updates nested from the 36 km domain to the 12 km domain and from the 12 km domain to the 4 km domain, using one-way nesting (feedback = 0).

4.10 TOP AND BOTTOM BOUNDARY CONDITIONS

The top boundary condition will be selected as an implicit Rayleigh dampening for the vertical velocity. Consistent with the model application for non-idealized cases, the bottom boundary condition will be selected as physical, not free-slip.

12 http://polar.ncep.noaa.gov/sst


4.11 FDDA DATA ASSIMILATION

The WRF model will be run with a combination of analysis and observational nudging (i.e., Four Dimensional Data assimilation [FDDA]). Analysis nudging will be used on the 36 km and 12 km domain. For winds and temperature, analysis nudging coefficients of 5x10-4 and 3.0x10-4 will be used on the 36 km and 12 km domains, respectively. For mixing ratio, an analysis nudging coefficient of 1.0x10-5 will be used for both the 36 km and 12 km domains. The nudging uses both surface and aloft nudging with no nudging for temperature and mixing ratio in the lower atmosphere (i.e., within the boundary layer). Observational nudging will be performed on the 4 km grid domain using the Meteorological Assimilation Data Ingest System (MADIS)13 observation archive. The MADIS archive includes the National Climatic Data Center (NCDC)14 observations and the National Data Buoy Center (NDBC) Coastal-Marine Automated Network C-MAN15 stations. The observational nudging coefficients for winds and temperatures will be 1.2x10-4, 6.0x10-4, respectively and the radius of influence will be set to 60 km.

4.12 WRF PHYSICS AND INPUT SENSITIVITY

The WRF sensitivities proposed here are based on knowledge gained from the prior 3SAQS meteorological modeling, recent studies working to improve WRF model performance during the wintertime over complex terrain, and advice from meteorological modelers working on similar problems. The preliminary WRF wintertime configuration will use the same physics as in the final 3SAQS configuration but with the latest release of WRF (v3.6.1). Table 4-2 lists the 3SAQS WRF physics options to be used as a base configuration for the WSAQS wintertime modeling. Our goal is to improve upon the 3SAQS WRF model configuration, especially during the wintertime. The WRF configurations will be run and evaluated for two full winter seasons.

Snow cover is known to be an important component of wintertime ozone events and that improving its representation in the model improves the simulation of cold air pools. We will directly use the Snow Data Assimilation System (SNODAS16, Barrett 2003) product for snow cover in WRF. SNODAS integrates a physically based, spatially distributed energy and mass balance model with observed snow data from satellite and airborne platforms, and ground stations. SNODAS output has high spatial resolution (~1 km) and temporal resolution (1 hour) over the United States. Our implementation of SNODAS in WRF will include adjustments to the snow albedo that are based on land use type. SNODAS will replace the NAM snow cover used in the previous 3SAQS WRF configuration. Figure 4-2 illustrates that there are large differences in snow cover between NAM and SNODAS.

The 2008 and 2011 3SAQS WRF simulations used the Noah land surface model (LSM), which is consistent with prior WRF simulations conducted by UUtah and UTDAQ. When WRF is run for operational air quality simulations, it is typically reinitialized every 5.5 days. In particular, the Noah LSM in WRF includes a snow sub-model that is initialized using a snow cover database (i.e. from NAM or SNODAS) and reinitializing the snow in addition to the atmospheric fields is important. Discussions with UUtah and UTDAQ provided insight into a potential problem with the timing and frequency of reinitializing in WRF. In particular, initializing the model in the middle of a cold air pool event may create problems when trying to represent these events. This is a difficult operational problem given timing of these events and that they are likely happening at different times throughout the domain. Reinitializing during a cold pool event is a problem that may be avoided by carefully constructing the reinitialization time for known events or by using the Pleim-Xiu (P-X)

13 Meteorological Assimilation Data Ingest System. http://madis.noaa.gov/14 National Climatic Data Center. http://lwf.ncdc.noaa.gov/oa/ncdc.html 15 National Data Buoy Center. http://www.ndbc.noaa.gov/cman.php16 Snow Data Assimilation System. http://nsidc.org/data


http://www.ndbc.noaa.gov/cman.php

http://lwf.ncdc.noaa.gov/oa/ncdc.html

http://madis.noaa.gov/

land surface model (LSM). The P-X LSM directly assimilates SNODAS data, thus avoiding reinitialization. If reinitialization happens during a known cold pool and associated poor air quality event, we will run at least one of these cold pool events continuously to evaluate the impacts of reinitializing the model during a cold pooling episode.

The P-X LSM is a well-known option in WRF to drive retrospective meteorological simulations for regulatory air quality applications (Gilliam and Pleim, 2010). As the P-X LSM does not include a sub-snow model, it can directly assimilate the SNODAS dataset using FDDA. WRF modeling by the U.S. EPA no longer reinitialize the model every 5.5 days (per conversation Rob Gilliam, US EPA) and they have recently demonstrated significant model performance improvements for near surface fields when recycling the WRF model 2-m temperature as a background field to Obsgrid for soil nudging (Gilliam et al. 2014). Figures 4-3 and 4-4 and illustrate this “iterative nudging” approach with an example of differences between the iterative and standard nudging techniques. We will test the impacts of continuous integration, directly assimilating SNODAS, and iterative nudging with a WRF sensitivity run for both winter seasons. Note that this P-X sensitivity test will use the NLCD 2006 LULC data and the Asymmetric Convective Model (ACM version 2; Gilliam and Pleim, 2010). We will use the ACM PBL scheme for consistency with ongoing efforts at the U.S. EPA. Additionally, WRF PBL sensitivity tests performed by both the U.S. EPA and UTDAQ have shown have shown little improvement in near surface fields relative to the magnitude of observed errors (per conversation Chris Misenis, U.S. EPA and Lance Avey, UTDAQ).

Table 4-3 lays out the WRF configuration experiments that we will conduct to optimize the performance of the model for winter conditions. The first configuration will use the Noah LSM with USGS LULC data. The novel winter configuration that we will test will initialize the Noah LSM using SNODAS and include land-use specific albedo adjustments. For this configuration we will develop a special data processor to format the SNODAS data for input to the WRF-REAL program. This configuration will reinitialize every 5.5 days as in the prior 3SAQS, which includes reinitializing back to SNODAS.

In the second configuration we will use the P-X LSM with NCLD LULC data and the ACM2 PBL scheme. WRF-REAL will also be used to create SNODAS snow-cover inputs to WRF, which will be assimilated directly into P-X. This simulation will not reinitialize but rather use a continuous integration initialized on December 1st.

We will build on the second configuration and apply an iterative nudging approach for the final WRF configuration experiment. All three experiments will be run for the two winter season episodes described above and evaluated against the available field campaign meteorology observations. A possible fourth sensitivity test includes a high-resolution 1.33-km WRF simulation for a single configuration and episode. This simulation will only be performed if the prior 4-km simulations fail to improve the representation of cold air pool events, especially for near surface fields. The higher resolution simulation will include a 1.33-km covering the oil and gas basins in northeast Utah, southwest Wyoming, and northern Colorado.

The wintertime configuration chosen with the smallest overall error within the Uinta, Upper Green River, and Salt Lake basins will be used to simulate July of 2011. This additional test will help determine if different winter and summer configurations are necessary for simulating meteorology to support air quality assessments in the Western U.S.

4.14 WRF OUTPUT VARIABLES

The WRF model will be configured to output additional variables in order to support air quality modeling with the Community Multiscale Air Quality Model (CMAQ). The following fields will be activated in the WRF output history files for the 3SAQS: fractional land use (LANDUSEF), aerodynamic resistance (RA), stomatal


resistance (RS), vegetation fraction in the Pleim-Xiu LSM (VEGF_PX), roughness length (ZNT), inverse Monin-Obukhov length (RMOL)

4.15 WRF APPLICATION METHODOLOGY

The WRF model when using the Noah land surface option will be executed in 5-day blocks initialized at 12Z every 5.5 days. Twelve (12) hours of spin-up will be included in each 5-day block before the data will be used in the subsequent evaluation. The WRF model when using the P-X land surface option will be executed with a continuous integration with 12 hours of spin-up. Model results will be output every 60 minutes and output files will be split at 12-hour intervals. Additional experiments will include testing using a longer runtime given the length of known cold air pool events. WRF will be configured to run in distributed memory parallel mode.


Table 4-2. Physics options used in the initial 3SAQS WRF simulation that would be used as starting configuration for WSAQS WRF runs.

WRF Treatment Option Selected NotesMicrophysics Thompson A scheme with ice, snow, and

graupel processes suitable for high-resolution simulations.

Longwave Radiation RRTMG Rapid Radiative Transfer Model for GCMs includes random cloud overlap and improved efficiency over RRTM.

Shortwave Radiation RRTMG Same as above, but for shortwave radiation.

Land Surface Model (LSM) NOAH Two-layer scheme with vegetation and sub-grid tiling.

Planetary Boundary Layer (PBL) scheme

YSU Yonsie University (Korea) Asymmetric Convective Model with non-local upward mixing and local downward mixing.

Cumulus parameterization Kain-Fritsch in the 36 km and 12 km domains. None in the 4 km domain.

4 km can explicitly simulate cumulus convection so parameterization not needed.

Analysis nudging Nudging applied to winds, temperature and moisture in the 36 km and 12 km domains

Temperature and moisture nudged above PBL only

Observation Nudging Nudging applied to surface wind and temperature only in the 4 km domain

moisture observation nudging produces excessive rainfall

Initialization Dataset 12 km North American Model (NAM)

Table 4-3. Proposed WRF Wintertime Sensitivity ExperimentsWinter Season

Exp 1. SNODAS w/ Albedo Adjustments

Exp. 2. Reinitialization

Exp 3. P-X Land Surface Model

Exp 4. Iterative Nudging

Exp. 5 High Resolution

Dec. 2010 – Mar. 2011 Noah LSM with

USGS17 LULC data and SNODAS initialization

NA P-X LSM & ACM2 PBL with

NLCD 200618 LULC data and

assimilated SNODAS

Same as Exp 2. with iterative

nudging of soil temperature and

moisture

N/A

Dec. 2012 – Mar. 2013

NA

2011 Episode N/A Exp. 1 N/A N/A Exp. 1, 2, or 3

17 USGS Land Cover Database. http://landcover.usgs.gov/18 National Land Cover Database 2006. http://www.mrlc.gov/nlcd2006.php


http://landcover.usgs.gov/

Figure 4-2. Comparison of snow cover from the NAM-12km (left) and SNODAS-1km (right).19

Figure 4-3. Illustration of “iterative nudging” for the P-X land surface model. The P-X uses 2-m temperature and relative humidity for indirect soil moisture and deep soil temperature nudging.

19 Images courtesy of Rob Gilliam, US EPA.


Met Analysis (NAM)

OBSGRIDObs. Data

WRF analysis (4-km)

4-km WRF

WRF output (4-km)

Figure 4-4. 2-m temperature analysis used for soil nudging with the P-X land surface model with standard 2m temperature analyses (left) and iterative nudging (right).


5.0 METEOROLOGICAL MODEL PERFORMANCE EVALUATIONThe WRF model evaluation approach will primarily use a process-oriented evaluation. The WRF evaluation will focus on time and locations with high wintertime ozone. Figure 5-1 is a conceptual model of known meteorological processes important for wintertime ozone. The simulated meteorological processes seen in Figure 5-1 will be compared against field campaign data such as those obtained from PCAPS, UBWOS, and UGRWOS. The following are some of the meteorological processes to be evaluated, where possible:

synoptic scale processes such as the preconditioning of cold air near the surface, warm air advection aloft, and large-scale atmospheric subsidence.

mesoscale processes such as the spillover of warm air into the valley creating dynamic instability and thermally driven upslope and downslope flows.

surface and boundary layer processes such as the amount of snow cover and/or cloud cover, temperature inversions, boundary layer heights, wind speed and direction, and net solar radiation.

Below we discuss the data used for the process oriented evaluation and additional quantitative evaluation of near surface meteorological observations.

Figure 5-1. Conceptual model of meteorological processes that influence cold air pool formation and demise.


5.1 METEOROLOGICAL PROCESS EVALUATION USING FIELD CAMPAIGN OBSERVATIONS

A number of software tools including R20, GRADS21, NCL22, and AMET23 will be used to compare field campaign measurements and observations at the surface and aloft. Below is a list of the types of the observational datasets and the meteorological processes being evaluated.

915 Wind Profiler with RASS (Radio Acoustic Sounding System) and 449 Wind Profilero Vertical profiles of temperature o Vertical profiles of wind speed

Ceilometer – aerosol optical deptho Cloudso Relative humidity

Mobile soundings from glidero Temperature o Relative Humidity

HOBO mobile temperature logger Surface meteorological tower sites

o Winds o Temperature o Relative humidity o Barometric Pressure o Solar radiationo Soil Parameters

SODAR/Sonic anemometero Boundary layer height

Doppler Lidaro Vertical profiles of winds

Surface meteorological data (ASOS, MESONET sites, field sites)o Temperature o Relative Humidityo Winds

Upper-air soundings (RAOBs) o Vertical profiles of temperatureo Vertical profile of dewpointo Vertical profile of winds

SNOTEL datao Snow depth, air temperature, precipitation

20 http://www.r-project.org/21 http://iges.org/grads/22 http://www.ncl.ucar.edu/23 http://www.cmascenter.org


Overall, the meteorological data from the field studies will be used to evaluate the meteorological processes important for simulating the timing, strength, and length of the cold air pool episodes important for wintertime ozone.

5.2 QUANTITATIVE EVALUATION USING SURFACE METEOROLOGICAL OBSERVATIONS

The statistical evaluation approach examines tabulations and graphical displays of the model bias and error for surface wind speed, wind direction, temperature, and mixing ratio (humidity) and compares the performance statistics to benchmarks developed based on a history of meteorological modeling as well as past meteorological model performance evaluations. Interpretation of bulk statistics over a continental or regional scale domain is problematic, and it is difficult to detect if the model is missing important sub-regional features. For this analysis the statistics will be performed on the three states of Colorado, Wyoming, and Utah and select valley basins within the 4-km domain, with additional domain-wide basis for the 36 km CONUS and 12 km modeling domains. The WRF evaluation procedures used by 3SAQS study will be used in the WSAQS (ENVIRON and UNC, 201324).

The observed database for winds, temperature, and water mixing ratio to be used in this analysis are the National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (ESRL) Meteorological Assimilation Data Ingest System (MADIS). The locations of the MADIS monitoring sites within the 4 km 3SAQS WRF modeling domains are shown in Figure 5-2.

Figure 5-2. Locations of MADIS surface meteorological modeling sites with the WRF 4-km 3SAQS modeling domain.

24 http://vibe.cira.colostate.edu/wiki/Attachments/Modeling/3SAQS_2011_WRF_MPE_v8_draft_Aug04_2014.pdf


The quantitative model performance evaluation of WRF using surface meteorological measurements will be performed using the publicly available METSTAT25 and AMET26 evaluation tools. Both tools calculate statistical performance metrics for bias, error, and correlation for surface winds, temperature, and mixing ratio and can produce time series of predicted and observed meteorological variables and performance statistics. A full annual model evaluation is very difficult to summarize in a single document, especially a simulation that could be used for many different purposes. With this in mind, the WRF model evaluation will present results for several subregions and even at the individual site level within the three-state area, leaving potential data users to independently judge the adequacy of the model simulation. Overall comparisons are offered to judge the model efficacy, but this review does not necessarily cover all potential user needs and applications.

Statistical metrics will be presented for each state, for each RPO, and for the United States portion of the 36 km, 12 km, and 4 km modeling domains. To evaluate the performance of the WRF 2011 simulation for the U.S., a number of performance benchmarks for comparison will be used. Emery et al. 2001 derived and proposed a set of daily performance “benchmarks” for typical meteorological model performance. These standards were based upon the evaluation of about 30 MM5 and RAMS meteorological simulations of limited duration (multi-day episodes) in support of air quality modeling study applications performed over several years. These were primary ozone model applications for cities in the eastern and Midwestern U.S. and Texas that were primarily simple (flat) terrain and simple (stationary high pressure causing stagnation) meteorological conditions. More recently these benchmarks have been used in annual meteorological modeling studies that include areas with complex terrain and more complicated meteorological conditions; therefore, they must be viewed as being applied as guidelines and not bright-line numbers. That is, the purpose of these benchmarks is not to give a passing or failing grade to any one particular meteorological model application, but rather to put its results into the proper context of other models and meteorological data sets. Recognizing that these simple conditions benchmarks may not be appropriate for more complex conditions, McNally (2009) analyzed multiple annual runs that included complex terrain conditions and suggested an alternative set of benchmarks for temperature under more complex conditions. As part of the WRAP meteorological modeling of the western U.S., including the Rocky Mountain Region, as well as the complex conditions up in Alaska, Kemball-Cook (2005) also came up with meteorological model performance benchmarks for complex conditions. Table 5-1 lists the meteorological model performance benchmarks for simple (Emery et al., 2001) and complex conditions, where we have adopted the complex benchmarks from Kemball-Cook (2005) since they covered more of the meteorological parameters.

The key to the benchmarks is to understand how well the model performs in this case, relative to other model applications run for the U.S. These benchmarks include bias and error in temperature, wind direction and mixing ratio as well as the wind speed bias and Root Mean Squared Error (RMSE) between the models and databases. The benchmark for each variable to judge whether predictions from a meteorological model are on par with previous meteorological modeling studies are as follows:

25 26 http://www.cmascenter.org


Table 5-1. Meteorological model performance benchmarks for simple and complex conditions.Parameter Simple ComplexTemperature Bias ≤ ±0.5 K ≤ ±2.0 KTemperature Error ≤ 2.0 K ≤ 3.5 KMixing Ratio Bias ≤ ±1.0 g/kg NAMixing Ratio Error ≤ 2.0 g/kg NAWind Speed Bias ≤ ±0.5 m/s ≤ ±1.5 m/sWind Speed RMSE ≤ 2.0 m/s ≤ 2.5 m/sWind Direction Bias ≤ ±10 degrees NAWind Direction Error ≤ 30 degrees ≤ 55 degrees

The equations for bias and error are given below, with the equation for the Root Mean Squared Error (RMSE) similar to the unsigned error metric only being the square of the differences between the prediction and observation and a square root is taken of the entire quantity.

Bias =

Error = Figure 5-3 displays an example model performance soccer plot graphic from the 3SAQS 2011 WRF run for monthly temperature performance within the 4 km domains. Shown in the “soccer plot” is the monthly temperature bias (x-axis) versus error (y-axis) as colored symbols with the simple and complex performance benchmarks27 represented by the rectangles. When the WRF monthly performance achieves the benchmark, it falls within the rectangle (i.e., scores a goal). In this example we see the 2011 WRF simulation for Colorado, Wyoming, and Utah demonstrates that nearly all of the near surface temperatures results fall within the complex benchmark with exception of Utah during the winter. However, during the wintertime model error increases.

27 Note that Figure 5-3 is using the McNally (2009) versions of the complex benchmark for temperature whereas we have adopted the Kemball-Cook (2005) versions for the temperature and wind benchmarks.


N

iii OP

N 1

1

N

iii OP

N 1

1

Figure 5-3. Example quantitative model performance evaluation display using soccer plots that compare monthly temperature performance (colored symbols) for the three states: Colorado (top left), Utah (top right) and Wyoming (bottom) with the simple and complex model performance benchmarks (rectangles).


6.0 REFERENCESAhmadov, R., et al. 2014. Understanding high wintertime ozone pollution events in an oil and

natural gas producing region of the western US. Atmospheric Chemistry Physics, 14, 20295-20343.

Baklanov, A.A., B. Grisogono, R. Bornstein, L. Mahrt, S.S. Zilitinkevich, P. Taylor, S.E. Larsen, W.M. Rotach, and H.J.S. Fernando. 2011. The nature, theory, and modeling of atmospheric boundary layers, Bulletin of the American Meteorological Society, 92, 123-128, doi: 10.1175/2010BAMS2791.1.

Barrett, A. 2003. National Operational Hydrologic Remote Sensing Center Snow Data Assimilation System (SNODAS) Products at NSIDC. NSIDC Special Report 11. Boulder, CO.

Emery, C., E. Tai, and G. Yarwood, 2001. Enhanced Meteorological Modeling and Performance Evaluation for Two Texas Ozone Episodes. Prepared for the Texas Natural Resource Conservation Commission, prepared by ENVIRON International Corporation, Novato, CA. 31-August. (http://www.tceq.texas.gov/assets/public/implementation/air/am/contracts/reports/mm/EnhancedMetModelingAndPerformanceEvaluation.pdf ).

ENVIRON and UNC . 2013. Three-State Air Quality Study (3SAQS): Weather Research and Forecasting Model Draft Modeling Protocol. ENVIRON International Corporation, Novato, California. University of North Carolina, Chapel Hill, NC. May 2013. (Add link).

UNC and ENVIRON. 2014a. Three-State Air Quality Study (3SAQS): Weather Research and Forecasting Model – Year 2011 Modeling Application/Evaluation. University of North Carolina, Chapel Hill, NC. ENVIRON International Corporation, Novato, California. August 2014. (http://vibe.cira.colostate.edu/wiki/Attachments/Modeling/3SAQS_2011_WRF_MPE_v8_draft_Aug04_2014.pdf).

UNC and ENVIRON. 2014b. Three-State Air Quality Study (3SAQS): CAMx Photochemical Grid Model Draft Model Performance Evaluation – Simulation year 2011. University of North Carolina, Chapel Hill, NC. ENVIRON International Corporation, Novato, California. November 10. (http://views.cira.colostate.edu/tsdw/Demo/Documents.aspx).

EPA. 1991. Guideline on the Regulatory Application of the Urban Airshed Model. U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards, technical Support Division, Source Receptor Analysis Branch, Research Triangle Park, North Carolina. July. (http://www.epa.gov/ttn/scram/guidance/guide/uamreg.pdf).

EPA. 2007. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5 and Regional Haze. U.S. Environmental Protection


http://www.epa.gov/ttn/scram/guidance/guide/uamreg.pdf

http://www.tceq.texas.gov/assets/public/implementation/air/am/contracts/reports/mm/EnhancedMetModelingAndPerformanceEvaluation.pdf

http://www.tceq.texas.gov/assets/public/implementation/air/am/contracts/reports/mm/EnhancedMetModelingAndPerformanceEvaluation.pdf

Agency, Research Triangle Park, NC. EPA-454/B-07-002. April. (http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rh-guidance.pdf).

Gilliam, R., J. Pleim, W. Appel, C. Misenis, K. Baker. 2014. Improving WRF for Fine-scale Air Quality Applications. Atmospheric Modeling and Analysis Division Seminar, March 26.

Gilliam. R.C., and J.E. Pleim. 2010. Performance Assessment of New Land Surface boundary Layer Physics in the WRF-ARW. Journal of Applied Meteorology and Climatology, 49, 760-774. doi:http://dx.doi.org/10.1175/2009JAMC2126.1.

Gultepe, I., T. Kuhn, M. Pavolonis, C. Calvert, J. Gurka, A.J. Heymsfield, P.S. K. Liu, B. Zhou, R. Ware, B. Ferrier, J. Milbrandt, B. Bernstein. 2014. Ice fog in arctic during FRAM-ICE fog project: aviation and nowcasting applications. Bulletin of the American Meteorological Society, 211-226. doi: 1175/BAMS-D-11-00071.1

Holtslag, A.A.M. G. Svensson, P. Baas, S. Basu, B. Beare, A.C.M. Beljaars, F.C. Bosveld, J. Cuxart, J. Lindvall, G.J. Steenveld, M. Tjernstrom, B.J.H. Van De Wiel. 2013. Stable atmospheric boundary layers and diurnal cycles: challenges for weather and climate models. Bulletin of the American Meteorological Society. 94, 1691-1706. doi:10.1175/BAMS-D-11-00187.1.

Kemball-Cook, S., Y. Jia, C. Emery and R. Morris. 2005. Alaska MM5 Modeling for the 2002 Annual Period to Support Visibility Modeling. Prepared for Western Regional Air Partnership (WRAP). Prepared by Environ International Corporation. September. http://pah.cert.ucr.edu/aqm/308/docs/alaska/Alaska_MM5_DraftReport_Sept05.pdf

Laureau, N.P., E.T. Crosman, C.D. Whiteman, J.D. Horel, S.W. Hoch, W.O.UJ. Brown, T.W. Horst. 2013. The persistent cold-air pool study, Bulletin of the American Meteorological Society, 94, 51-63. doi:10.1175/BAMS-D-11-00255.1.

Laureau, N. and J. Horel, 2014. Turbulent Erosion of Persistent Cold Air Pools: Numerical Simulations. Journal of Atmospheric Science. doi:10.1175/JAS-D-14-0173.1, in press.

McNally, D. E., 2009. 12km MM5 Performance Goals. Presentation to the Ad-Hoc Meteorology Group. 25-June. http://www.epa.gov/scram001/adhoc/mcnally2009.pdf

Michalakes, J., J. Dudhia, D. Gill, J. Klemp and W. Skamarock. 1998. Design of a Next-Generation Regional Weather Research and Forecast Model. Mesoscale and Microscale Meteorological Division, National Center for Atmospheric Research, Boulder, CO. (http://www.mcs.anl.gov/~michalak/ecmwf98/final.html).

Michalakes, J., S. Chen, J. Dudhia, L. Hart, J. Klemp, J. Middlecoff and W. Skamarock. 2001. Development of a Next-Generation Regional Weather Research and Forecast Model. Developments in Teracomputing: Proceedings of the 9th ECMWF Workshop on the Use of High Performance Computing in Meteorology. Eds. Walter Zwieflhofer and Norbet Kreitz. World Scientific, Singapore. Pp. 269-276. (http://www.mmm.ucar.edu/mm5/mpp/ecmwf01.htm).


http://www.epa.gov/scram001/adhoc/mcnally2009.pdf

http://pah.cert.ucr.edu/aqm/308/docs/alaska/Alaska_MM5_DraftReport_Sept05.pdf

http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rh-guidance.pdf

Michalakes, J., J. Dudhia, D. Gill, T. Henderson, J. Klemp, W. Skamarock and W. Wang. 2004.

The Weather Research and Forecast Model: Software Architecture and Performance. Proceedings of the 11th ECMWF Workshop on the Use of High Performance Computing in Meteorology. October 25-29, 2005, Reading UK. Ed. George Mozdzynski. (http://wrf-model.org/wrfadmin/docs/ecmwf_2004.pdf).

Neeman, E.M., E.T. Crosman, J.D. Horel, L. Avey. 2014. Simulations of a cold-air pool associated with elevated wintertime ozone in the Uintah Basin, Utah. Atmospheric Chemistry and Physics. 15, 15953-16000. doi:10.5194/acpd-14-15953-2014

Schnell, R.C. S.J. Oltmans, R.R. Neely, M.S. Endres, J.V. Molenar, A.B. White. 2009. Rapid photochemical production of ozone at high concentrations in a rural site during winter, Nat. Geosci., 2, 120-122. Doi:10.1038/NGEO415

Skamarock, W. C. 2004. Evaluating Mesoscale NWP Models Using Kinetic Energy Spectra. Mon. Wea. Rev., Volume 132, pp. 3019-3032. December. (http://www.mmm.ucar.edu/individual/skamarock/spectra_mwr_2004.pdf).

Skamarock, W. C., J. B. Klemp, J. Dudhia, D. O. Gill, D. M. Barker, W. Wang and J. G. Powers. 2005. A Description of the Advanced Research WRF Version 2. National Center for Atmospheric Research (NCAR), Boulder, CO. June. (http://www.mmm.ucar.edu/wrf/users/docs/arw_v2.pdf).

Skamarock, W. C. 2006. Positive-Definite and Monotonic Limiters for Unrestricted-Time-Step Transport Schemes. Mon. Wea. Rev., Volume 134, pp. 2241-2242. June. (http://www.mmm.ucar.edu/individual/skamarock/advect3d_mwr.pdf).

Utah State, 2013. 2012 Uintah Basin Winter Ozone Study and Air Quality Study Final Report. Utah State University. Feb.

(http://rd.usu.edu/files/uploads/ubos_2011-12_final_report.pdf).

Utah DEQ. 2014. Uintah Basin Ozone Study Final Report – Distributed Measurements of Air Quality and Meteorology. Utah Department of Environmental Quality, (DEQ), Division of Air Quality (DAQ). (http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-Final-Report/UBOS_2013Sec_3_DistribMon.pdf).


http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-Final-Report/UBOS_2013Sec_3_DistribMon.pdf).

http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-Final-Report/UBOS_2013Sec_3_DistribMon.pdf).

http://www.mmm.ucar.edu/individual/skamarock/advect3d_mwr.pdf