Grantee Research Project Results

Final Report: Boise Valley Inversion and Air Pollution Study

EPA Grant Number: R829425
Title: Boise Valley Inversion and Air Pollution Study
Investigators: Dawson, Paul
Institution: Boise State University
EPA Project Officer: Chung, Serena
Project Period: May 27, 2002 through May 26, 2005
Project Amount: $256,266
RFA: EPSCoR (Experimental Program to Stimulate Competitive Research) (2001) RFA Text |  Recipients Lists
Research Category: EPSCoR (The Experimental Program to Stimulate Competitive Research)

Objective:

The Boise Valley (Treasure Valley [TV]), which has been experiencing a high rate of growth in the last decade, is particularly susceptible to prolonged winter valley inversion events. Atmospheric inversions often trap polluted air in the TV, see Figure 1, and make it unhealthy to breathe.

Figure 1. Map of Treasure Valley Region and Photograph of Winter Valley Inversion Event

In the winter, the major pollutant is fine particulate matter, PM_2.5 (PM less than 2.5 microns). Monitoring studies have shown that secondary aerosols, consisting mostly of ammonium nitrate and ammonium sulfate, form in the valley in cold, humid, stagnant air. These aerosols may account for more than 60 percent of the total PM_2.5 particles. During the study, the TV exceeded EPA PM_2.5 standards and reached unhealthy levels on December 4, 2002, as shown in Figure 2. The 24-hour average Air Quality Index (AQI) values reached 160 in Boise, the highest level in 10 years, and 153 in Nampa.

Air Quality Index (AQI)	0-50	51-100	101-150	151-200	201-300	301 and above
Air Quality Category	Good	Moderate	Unhealthy for Sensitive Groups	Unhealthy	Very Unhealthy	Hazardous

Figure 2. 24-hour Averaged AQI Values for the TV Between November 28 and December 8, 2002, and Pollutant Definitions

One goal of the research project was to better understand the meteorology and the associated evolution of PM in deep stable layers in the TV. Another goal was to apply a state-of-the-science meteorological model to simulate deep stable layer episodes successfully.

The objectives of the research project were to: (1) gain an understanding of the science and physics of atmospheric valley inversion development, persistence, and break-up; (2) analyze and model the 3-D airflow in the TV under various atmospheric conditions to provide a qualitative description of flow processes in the valley under various dynamic and thermal conditions; (3) monitor, model, and analyze the meteorology associated with deep stable layers in the TV during the project period; (4) monitor and analyze pollutant concentrations in the TV under winter inversion conditions during the project period to provide a better understanding of how air quality in the valley is related to the meteorology; (5) develop local training and expertise on local air quality problems in the TV; (6) compare results from all the winter inversion studies in the TV and make conclusions and suggestions for future studies, particularly for the air quality modeling; and (7) suggest possible improvements for mitigating air pollution events in the TV.

Summary/Accomplishments (Outputs/Outcomes):

Monitoring

Meteorological and air quality data for each multi-day inversion event were collected throughout the TV and the data were evaluated for quality assurance. The data were analyzed primarily using geographical information system (GIS) technology. Temperature and relative humidity contour plots were developed every 2 hours during the inversion events. The associated fog and wind patterns also were described and analyzed. These GIS analyses were useful in studying the evolution of the inversion events and in evaluating the accuracy of the numerical simulations. GIS software was written to implement these analyses, and a manual was developed to describe the implementation.

In 2 of the 3 years of the project, sodar systems were rented and used in the study. The sodar data, consisting of wind speed and wind direction up to 1,000 meters, were useful in studying the inversion events and in initializing and evaluating the model simulations.

National Weather Service (NWS) radiosonde data, surface data, and analysis maps also were important in the study. Similar to all inversion events that were studied, the maps in Figure 3 show an upper level ridge over the Pacific Northwest and a dominant surface high pressure system in the Great Basin area of the western United States. The depth and intensity of the inversions were analyzed using the maps, surface observations, and radiosonde data. Some examples of these analyses are presented later with the model results.

Figure 3. 500 millibar (mb) Analysis Map (right) and the Associated Surface Analysis Map (left) for 12 Z (12 GMT) on December 16, 2004

Modeling

Research activity focused on detailed meteorological modeling of the multiday inversion periods. The Pennsylvania State University/National Center for Atmospheric Research Mesoscale Model MM5 and its successor, the Weather Research and Forecast (WRF) model, were implemented on Boise State University’s (BSU’s) new Beowulf 128-node computer cluster. The modeling results were input into a GIS format and compared with the observations. The model results also were compared with radiosonde and sodar observations and with observations from four surface sites in the TV. A long-term goal of the modeling effort is to simulate the meteorology and air quality of the winter inversions in real time and to display the real-time simulations on the Web site.

Time plots of temperature, relative humidity, and wind speed at the NWS surface site at the Boise airport for December 14–17, 2004, are shown in the left column of graphs in Figure 4 along with the model results for the same time period. Plots of the observed temperature, relative humidity, and wind speed from the Boise radiosonde at 12Z on December 16, 2004, are shown with the simulated results in the right column of graphs in Figure 4.

Figure 4. Example Plots of Observations and Model Results

The “vanilla” run in the figures represents our best effort to date of simulating the observations using WRF but without using the data assimilation technique called 3DVAR in WRF. The 3DVAR is a modeling tool that was developed to bring observational data into limited area models to update initial and boundary conditions during the model run. The WRF simulation consisted of 3 nested grids, 35 vertical levels, and a horizontal resolution of 3 km in the innermost grid. The “sodar” run was exactly like the vanilla run, but it incorporated WRF’s 3DVAR data assimilation technique. The sodar data were brought into the model simulation every 12 hours to update initial and boundary conditions. The sodar data were expected to improve model results, particularly low-level wind results.

The results for the vanilla run were encouraging. The model simulated the time record of the BOI temperature well and satisfactorily simulated the relative humidity time record, particularly the diurnal trends, (left column of figures). Wind results were variable. The sodar run, however, was not very accurate. It was concluded that more practice and more evaluations were needed in using 3DVAR. Similarly, the vanilla run simulated the observed radiosonde data (right column of figures) better than the sodar run. It is interesting to note that the “vanilla” run simulated the surface inversion adequately, but it did not simulate the observed elevated inversion at 1,500 m accurately. This was a general weakness in the model simulations and, at times, resulted in erroneous fog and low stratus cloud patterns. The “vanilla” run also underpredicted the humidity at 1,200-1,500 meters. The “vanilla” run also overpredicted the observed low-level wind speeds with height in the December 16 wind speed profile. This is a common complaint of MM5 and WRF simulations. This was the result that sodar data assimilation was meant to correct.

GIS Analysis and Visualization

Surface temperatures and relative humidities were analyzed using color contours and the results are shown in Figure 5. Blue to red contours represent colder to warmer temperatures and lower to higher relative humidities. The colors in the relative humidity plots represent various saturation levels of fog with the blue contour being 85 percent relative humidity. The TV sites where observations were obtained are represented by the bold symbols. These sites are shown in both the observation analysis and in the model analysis. Elevation contours also are shown in the figures. Note that the temperature increases with elevation over the TV and that the upper eastern side of the TV is warmer than the lower western end of the valley. Also note that the model did not simulate the fog in the TV completely and that the model results in the fog were more saturated than in the observations.

VIS5D has been used to obtain animations depicting the evolution and dissipation of fog patterns in the valley. Figure 6 depicts fog in the valley in the first box and a clearing trend in the eastern side of the valley several hours later.

Conclusions:

We made significant progress in understanding the meteorology and evolution of winter inversions in the TV. We also did well at collecting data and performing GIS analysis of the inversion events. Undergraduate students developed a GIS manual to explain the GIS operation. GIS has had limited use in prior meteorological analyses. In the future, we would like to learn to use the GIS Statistical Analysis Tool to help us compare model results with observations and to compare various model results after performing sensitivity analysis with model parameters.

Observations of Temperature	Simulation of Temperature
Observations of Relative Humidity	Simulation of Relative Humidity

Figure 5. Examples of GIS Analyses of the Observations (left) and Model Results (right)

Figure 6. Example of Model Results Using the Visualization Package, VIS5D

We also made significant progress in implementing and executing both the MM5 model, the WRF model, which is the successor to MM5, and associated model visualization packages. Our objective in model evaluation and visualization was to use GIS for both the observations and the model results. We developed software programs and data formatting programs in conjunction with using MM5, WRF, GIS, and some of the visualization packages. Learning that low-level, low-speed winds were difficult to simulate with MM5, we tried to use data assimilation to improve the model wind results. We have not yet achieved success, however, with this assimilation technique, which is called 3DVAR in WRF.

We did a successful simulation of a December 2004, inversion event as described earlier. We simulated the temperature and relative humidity reasonably well at three sites in the upper TV. The fourth site in the lower TV was not as well simulated because fog persisted in the lower terrain while the model was producing a diurnal trend in temperature and humidity. We found that, besides low-level, low-speed winds, it is difficult to successfully simulate subtle capping inversions above the surface. These elevated inversions are important in defining fog and low-level stratus cloud features over the valley.

The sodar system proved to be valuable in understanding and evaluating the meteorology. We thought that the sodar system observations, used in conjunction with the model, would help initialize, simulate, and evaluate low-level winds. We believe that more experience in data assimilation modeling is needed, and the reporting system of commercial aircraft observations, called ACARS, should be incorporated into the 3DVAR operation.

We are continuing to collaborate on the winter inversion study with the Idaho Department of Environmental Quality (IDEQ), with NWS personnel, and with AIRQUEST scientists. We also are continuing a study with an IDEQ scientist that correlates the AQI to the strength and depth of the winter inversions. We are collaborating with scientists in the U.S. Federal Aviation Administration Air Transportation Center of Excellence for Aircraft Noise and Aviation Emissions Mitigation, with scientists in the BSU Center for Environmental Sensors, with scientists involved with the BSU Beowulf cluster, with scientists involved in wind energy studies, and with teachers and students at a local high school.

We successfully trained students and the local public on air quality science. Some graduate students, several undergraduate students, and some high school students participated and learned from the project. Some of our students had opportunities for internships at IDEQ. Also, the media interviewed us about the inversions, including a public radio news broadcast involving the principal investigator and an official from IDEQ. We also invited speakers to BSU to present seminars about inversions, modeling, and air quality.

A future objective of the researchers is to model the chemistry and the transport and dispersion of pollutants in the TV under winter inversion conditions. A continuing goal in the research effort is to develop an operational real-time air quality forecast model for the TV and elsewhere. We are pursuing that goal.

Supplemental Keywords:

mountain valley inversions, cool air pools, mesoscale modeling, particulate matter monitoring, air quality episodes,, RFA, Scientific Discipline, Air, particulate matter, Air Quality, Ecology, Environmental Monitoring, Atmospheric Sciences, aerosol formation, atmospheric particles, meteorology, air quality models, airborne particulate matter, air pollution, atmospheric aerosol particles

Progress and Final Reports:

Original Abstract

2002

2003 Progress Report