| Methodology |
|
| The approach in this
assessment is to determine spatial representativeness of existing IMPROVE
and IMPROVE protocol monitors. The broad definition of spatial
representativeness that will be accepted for this assessment is an area
within which pollutant concentrations are roughly constant. |
|
| When IMPROVE sites were being
chosen to be representative of Class I areas of the WRAP, elevation and
distance criteria were developed to create groupings (clusters) around the
original monitoring stations. Sites considered being representative had to
be between a maximum and minimum elevation (+/- 10%) of the Class I area and
at a criteria distance of no more than 100-kilometers. Avoidance of large
obstructions, locality to local emission sources, and small valleys was
encouraged. Information and comments to reflect siting concerns were
incorporated on a technical, political and expert level. Therefore, network
efficiency would be assumed to already exist. |
|
| Six years of IMPROVE data
between 1997 and 2002 was utilized for this representativeness assessment.
The six major aerosol species known to be the main contributors to light
extinction were analyzed; sulfates, nitrates, organic carbon (also expressed
as organic mass by carbon or OMC) , elemental carbon ( also expressed as
light absorbing carbon or LAC), fine soil and coarse mass. Another important
component of understanding attributions of haze in a region is transport. On
a regional scale, slow-reacting primary pollutants such as sulfur dioxide
and nitrogen oxide that make up fine particulate undergo reactions and
secondary formation processes during transport that result in haze. The
spatial scale of these fine particles is known to be regional. |
|
| Assumptions |
|
| The assumptions made for
aerosols are as follows: |
 | All elemental sulfur is from sulfate that is from ammonium sulfate |
| All nitrate is from ammonium nitrate |
| Total organic carbon is equal to carbon released in four steps
(OC1-OC4) plus pyrolized organics (OP) determined through Thermal Optical
Reflectance analysis of the quartz filter |
| Elemental carbon (light absorbing carbon) is the equal to the sum of
elemental carbon fractions minus pyrolized organics determined through
Thermal Optical Reflectance analysis of the quartz filter. |
| Fine soil is the sum of Al, Si, K, Ca, and Ti determined by
particle-induced X-ray emission, and Fe determined by X-ray fluorescence. |
| Coarse mass is the total mass minus the fine mass |
|
|
| Procedures |
|
| The tribal assessment was
conducted within the COHA physiographic regions that were based upon similar
terrain and defined areas of similar aerosol characteristics. There are 18
separate regions within the WRAP. The assessment also reflects data analyzed
on the days of worst visibility. The upper 20% of calculated visibility
impairment values or 20% worst visibility days over the years in question
were investigated at each site. The clearer days or the lower 20% and the
median 60% calculated contributions to haze are available for each IMPROVE
station on this website. |
| Finally, the role of
meteorology, intervening terrain, emission sources, distance and elevation
differences were utilized to define regional characteristics and explain
variances in measurements. The percent contribution of each aerosol species
to light extinction was also studied to determine regional relationships. |
|
| The differences in distance
in kilometers were matched with paired regional site correlation
coefficients of each of the six main aerosol components and plotted. The
IMPROVE sites were divided into respective physiographic regions to plot
graphically for the individual case studies. Linear regression analysis was
incorporated into the plots to empirically show the strength of the
correlation relationship in association to an exponential trendline. The
coefficient of determination (r-squared= 0 being worst to 1 being the best)
was an indicator of the reliability of the trendline. |
|
A correlation between the
sample site and itself also had to be considered given that the exact same
measurement would not be obtained between two samplers at one site. Due to
the fact that measurement uncertainty occurs even if two samplers are at the
exact same location, an artificial concentration was generated utilizing the
measurement uncertainties of each chemical component. Random numbers were
generated using an Excel spreadsheet function of (=NORMSINV(RAND()). This
was put into the equation:
Artificial concentration= uncertainty * random number + measurement
The artificial concentration was then correlated against the actual
measurements to create a correlation coefficient at a distance of zero. |
|
| To better analyze the scatter
plots of correlation versus distance, a criteria cut-off correlation
coefficient was determined. The validation of spatial representativeness is
dependent upon the choice of cut-off correlation. To rationalize the
association between monitoring sites, an arbitrary correlation coefficient
of 0.70 was chosen to indicate acceptable agreement between measured
concentrations. This value was chosen as it lies between 0.5 (the median
point at which values equal and less to 0.5 would determine no true
relationship existed) and the best possible answer of 1.0. The distance, at
which the cut-off correlation exists from 0, will be assumed to represent
the area wherein a concentration is predictable and extrapolable from a
site. |
|
| For purposes of this
assessment, an overall regional zone of influence was determined. First a
regionally representative distance was found for each aerosol component.
Each aerosol distance was then weighted by its respective contribution to
light extinction on worst visibility days. The distances were then summed to
create a regional representative distance. An assumption was made that
although poor r-squared values did exist for some of the species spatial
correlation analysis trendlines, a standard error of the line was acceptable
in the assessment to determine regional representativeness. |
|
| Results |
|
| Regional zone of influence results are summarized in a
table for those regions that had four
or more sites available for analysis between 1997-2002. Regions
investigated with four or more sites include
Cascade Range,
Central Rocky
Mountains, Colorado Plateau,
Mexican Highlands,
Northern Great Plains,
Southern Rocky Mountains and the
Sierra Nevada. Zones of
representation ranged from 91km in the Colorado Plateau to 210 km in the
Northern Great Plains. For the remaining regions with less than four
IMPROVE sites, a zone of influence of 150 km was used (median between
smallest and largest distances of the seven regions with four sites). |
|
The regional representative
distances are utilized to map and determine which tribal areas within
respective regions are represented. Determinations are based upon a
simplified approach of
1) Yes, the tribal area is represented and by which IMPROVE site or sites
2) No, the tribal area is not represented. (Recommendations to be determined
in Task 3) |
|
| After applying the
methodology outlined above, portions of several tribes and 11 tribes in
their entirety are not represented
by an IMPROVE or IMPROVE protocol monitoring site. |
|
|
|
|
|
|
|
|
|
|
|
|
| The areas in red indicate
locations are outside of the zone of representation for that region. |
|
|
|
