TREATMENT CATEGORIES FOR COAL
MINE DRAINAGE
kl .
Hittman Associates, Inc.
-------
TREATMENT CATEGORIES FOR COAL
MINE DRAINAGE
by
H. Lee Schultz
Donald Koch
Carolyn Thompson
Dr. Kathleen Hereford
Hittman Associates, Inc.
Columbia, Maryland 21045
Contract No. 68-03-2566
Project Officer
Roger C. Wilmoth
Resource, Extraction, and Handling Division
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Industrial Envi-
ronmental Research Laboratory, Extraction Technology Branch,
U.S. Environmental Protection Agency, and approved for pub-
lication. Approval does not signify that the contents
necessarily reflect the views and policies of the U.S. Envi-
ronmental Protection Agency, nor does mention of trade names
or commercial products constitute endorsement or recommenda-
tion for use.
-------
FOREWORD
When energy and material resources are extracted, pro-
cessed, converted, and used, the related pollutional impacts
on our environment and even on our health often require that
new and increasingly more efficient pollution control methods
be used. The Industrial Environmental Research Laboratory -
Cincinnati (lERL-Ci) assists in developing and demonstrating
new and improved methodologies that will meet these needs
both efficiently and economically.
This report details the results of an effort to deter-
mine if a suitable basis existed to establish a separate
Effluent Guidelines subcategory for coal mining point source
discharges to differentiate between eastern and western coal
mining. Unfortunately, problems with the existing data base
prevented resolution of the issue. The information and
sources contained herein would be of primary interest to the
EPA regulatory offices. For further information, contact
the Resource Extraction and Handling Division.
David G. Stephan
Director
Industrial Environmental
Research Laboratory
Cincinnati
111
-------
ABSTRACT
This effort involved the organization and statistical
analysis of a large amount of data characterizing over 300
surface and underground coal mines and quantification of the
pre- and post-treatment quality of their wastewaters. Only
existing data, supplied to Hittman Associates by EPA, was
utilized in this evaluation. The study objective was to
determine whether the data supported the development of a
separate Effluent Guidelines subcategory for coal-mining
point source discharges to differentiate between eastern and
western coal mining activities. An extensive effort was
required to convert the data into a computerized format
which would allow for expeditious statistical analysis.
Following computerization, a variety of statistical analyses
were conducted to determine if substantiation existed within
the data base for development of treatment subcategories
based on any or all of the following factors: precipita-
tion, effluent flow, effluent source, influent concentration
of pollutant parameters, treatment type, and acidity/alkalin-
ity. The results of the statistical analyses illustrated
that the data base provided to Hittman Associates was not
adequate to provide a verifiable basis for subcategorization
of the coal mining point source category. This inadequacy
stemmed from numerous problems which existed within the
data base concerning the non-uniformity population and the
non-comparability of much of the data.
IV
-------
CONTENTS
Foreword iii
Abstract iv
1. Introduction 1
2. Summary of Conclusions 2
3. Technical Approach 3
Data obtained 3
Problems in data utilization 4
Methodology 6
4. Results and Discussion 9
Results 9
Discussion 12
Appendices
A. Statistical computer runs completed .... 16
B. Illustrative t-test results 22
v
-------
SECTION 1
INTRODUCTION
The purpose of this study was to conduct computerized
statistical analyses to answer the following interrelated
questions:
o Does a given coal mine wastewater treatment
technology provide significantly different results
in the West than it does in the East?
o If so, what are the differences?
o Should different, more stringent treatment cri-
teria (and hence different treatment technologies)
be required for mines in different parts of the
country?
The analyses conducted in addressing these questions
were carried out based on existing data that were supplied
to Hittman Associates, Inc. (HAI) by the U.S. Environmental
Protection Agency (EPA). These data quantitatively charac-
terized wastewaters originating at approximately 300 surface
and underground coal mines located in the continental United
States. The data also contained information characterizing
the mining operation conducted at each site. Statistical
analyses were conducted to determine if the data supplied to
HAI were adequate to serve as a basis for addressing the
study questions. The approach taken in evaluating these
data, the results of the analyses, and conclusions drawn
therefrom are discussed in the following sections.
-------
SECTION 2
SUMMARY OF CONCLUSIONS
The data supplied for this analysis were not adequate
to support a decision regarding the need for subcategoriza-
tion of the coal mining point source category. This con-
clusion is based on the results of the analyses described
here, and the funding and time limitations placed on the
project. Results obtained in this analysis were inconclu-
sive and support neither the need nor the lack of need
for such subcategorization.
Data obtained through a statistically designed sampling
program, or through such a program in concert with utiliza-
tion of existing National Pollutant Discharge Elimination
System (NPDES) data, could eliminate most of the problems
identified in the data supplied to HAI and could, therefore,
provide conclusive results.
-------
SECTION 3
TECHNICAL APPROACH
DATA OBTAINED
Data were obtained by HAI from four sources:
EPA-IERL/Cincinnati: Mathematica report (unpub-
lished at time of acquisition)
o EPA-EGD/Washington: Mine effluent and production
data in computer cards (with corresponding coding
sheets)
EPA/Dallas: National Pollution Discharge Elimina-
tion System (NPDES) files
EPA/Denver: National Pollutant Discharge Elimina-
tion System (NPDES) files.
The data contained in each of these sources are
described below.
Mathematica Report
The Mathematica data, An Inventory of Western Mines,
are in computer printout form. The data are primarily opera-
tional in nature, although ancillary information such as
precipitation levels, surface rights data, land use informa-
tion, coal characteristics, etc. are also included. The
report covers 37 active and 7 planned surface coal mines in
the West that had more than 100,000 tons of production in
1975 (which accounts for about 99.5 percent of the western
surface coal tonnage produced in 1975).
EPA/Washington Computer Cards
EPA Effluent Guidelines Division (EGD) supplied some
20,000 computer cards containing data on mine identification,
mine characteristics, and 60 water quality parameters. This
data base covers 225 mines in all. Coverage is mainly for
eastern mines, although a signficant number of western mines
are also included. These data constitute the most comprehen-
-------
sive coverage of eastern mines available. Because the data
are already in computerized form, significant time and
effort are saved.
Versar Data
These unpublished data are in paper form (tabular) and
contain water quality data for 23 mines located in the East
and the West. The data primarily cover effluent concentra-
tions of trace metals, although a few traditional water
quality parameters are also included.
EPA Regions VI and VIII NPDES/DMR Files
Files from the National Pollutant Discharge Elimination
System/Discharge Monitoring Report (NPDES/DMR) were made
available to HAI in Dallas, Texas, and Denver, Colorado,
so that the necessary data for mines located in EPA Regions
VI and VIII could be obtained. Accessing these data in-
volved identifying those files that contained pertinent
data, copying the data, and coding and keypunching the data
for computerization. The files for Regions VI and VIII make
up the largest, most comprehensive data source for western
mines available at this time.
PROBLEMS IN DATA UTILIZATION
It should be noted that the coverage of the data sup-
plied to HAI was rather variable. That is, data from
different sources did not cover the same mine character-
istics, water quality parameters, etc., in all instances.
Thus, the ability to compare the data from different mines
was significantly impaired.
Major problems were encountered as computerized data
processing activities were initiated. Specifically, a
number of problems of varying severity were identified in
the EPA-EGD/Washington computerized data base. First, data
appeared to be coded in a somewhat haphazard fashion, and
data fields were found to be variable among the cards. Data
items such as mine name, location, etc., were not coded in
the same columns on all cards. This means that when the
computer looks for information at a given location, it will
in many instances find part of one data entry and part of
another (instead of an entire entry). This would cause
those data entries to which the problem applies to be use-
less and therefore be eliminated from evaluation.
An additional problem relating to coding of information
in this data base was that all numerical data were not re-
corded in the same units. For example, in the mine size
4
-------
category, tonnages were usually recorded in terms of "ton
per year"; however, in some instances such data were re-
corded in terms of "tons per day" or "tons per hour." While
a relatively simple subroutine could be written to convert
all such data entries to equal terms, the assumptions neces-
sary to do so (such as the number of hours per day worked at
the mine, the number of workdays per week, the number of
work-weeks per year, etc.) were not provided. Therefore,
any attempt at conversion would result in questionable
outputs since without mine-specific operational data, any
conversion would be arbitrary. However, neither this nor
the previously defined problem were insurmountable. While
they did require additional time and effort, and cut down on
the sample of useable data in certain circumstances, they
did not in themselves render the initial evaluation impossible
A third problem relative to the EGD cards uncovered
later in the project did effect a serious impediment to the
evaluation. The cards had been converted to tape form, and
the tape read into the computer system. The taped data was
printed out via terminal link prior to conducting analytical
runs in order to assure the compatibility (in terms of
format, etc.) with the other data to be used. At this time
it was determined that this data bank was not useable in its
initial form, as the data it contained were not logical.
When the data bank was printed out, it was impossible to
correlate the various data entries (that is, one could not
ascertain which data went with which mine, etc.). It was
determined that this problem was probably due to one of two
causes:
Blank data cards were not included in the data
bank when a blank entry was encountered on the
coding sheet. Unfortunately the computer does not
know if some entries have been left out. It can
only read the cards in the order in which they are
put in and extract data from the cards based on
this ordering. Thus, if some cards were deleted
from the deck, all following cards would be out of
order.
All the necessary cards may have been present,
but out of order. This could have occurred if the
cards had been dropped or otherwise mixed up.
Unfortunately, no sequential numbering system was
used when keypunching the cards, so that there was
no quick way to determine if all cards were both
present and in order.
In order to resolve these problems, it was decided that
the EGD computerized data base had to be completely receded,
then keypunched and entered into the computerized data
-------
management system. The original contract was modified to
allow for this "receding. HAI was 'directed to conduct the
receding effort using the original coding sheets (from which
the EGD-supplied computer cards were prepared) as the data
source. During this effort data fields and units were
standardized, and any other problems not yet identified
relative to the data contained in this data base were to be
resolved as far as was possible. A major portion of the
time and funding allocation for this project was subse-
quently dedicated to the resolution of the EGD data base
problems.
METHODOLOGY
In order to address this project in a comprehensive
fashion, any major mine characteristic which would be likely
to affect effluent quality were to be considered as a likely
candidate basis for a treatment category. For example, mine
grouping could be based upon average annual precipitation
levels generally considered to be indicative of various
ecological classifications, such as the four categories
presented below*:
0 to 10 inches of precipitation per year - desert
11 to 30 inches of precipitation per year - grass-
land, savanna, or open woodland
31 to 50 inches of precipitation per year - dry
forest
51 inches of precipitation per year or more - wet
forest
Alternately, such categories could be based upon the natural
bisection of the United States by the 20-inch average annual
precipitation isopleth which trends vertically through
eastern North and South Dakota, central Nebraska, and
western Kansas, Oklahoma, and Texas. This delineation would
yield two categories**: 0 to 20 inches of precipitation per
year and 21 inches of precipitation per year or more.
* Odum, Eugene P., Fundamentals of Eaolopi/. Second Edition
W.B. Saunders Company, Philadelphia, Pennsylvania 1959
p. 112.
** Geraghty, Miller, Van Der Leeden, and Troise, Water
Atlas of the United States, Water Information Center",Port
Washington, flew lork, 1973.
-------
Effluent flow is another mine characteristic which
could serve as a logical basis for categorization. While at
first glance it might be assumed that this characteristic
would be directly proportional to rainfall, this is not
necessarily the case. Coal mines often intercept aquifers
(water bearing strata), especially underground mines. When
this occurs, water generally must be pumped from the area of
active mining. Therefore, it is conceivable that a mine
located in an arid area could have significant flows of
water requiring treatment (flows greater than that expected
based on rainfall alone).
Effluent source obviously plays a significant role in
the characteristics of the wastewaters to be treated, and
thus in the expected treatment efficiency. This considera-
tion provides three potential categories for evaluation:
Surface mines
Underground mines
Preparation plants.
With the above considerations in mind, evaluation of
the supplied data indicated the appropriateness of subcate-
gorization based on the following six characteristics:
Precipitation
Effluent flow
Effluent source
Influent concentration of pollution parameters
Type of treatment
Acidity/alkalinity.
These categories served as the basis for all subsequent
evaluation. In addition, EPA chose to concentrate on nine
of the water quality parameters contained in the overall
data base during the evaluation process. The nine para-
meters included in analytical efforts were the following:
pH
Total suspended solids
Total dissolved solids
Zinc
-------
Aluminum
Total iron
Total manganese
Acidity
Alkalinity.
Once the data to be utilized in this evaluation had
been entered into the computerized data management system
and debugged, final computer runs were conducted and the
results evaluated to arrive at a determination as to the
appropriateness of one or more subcategorizations for the
coal mining effluent limitation point source category. The
output from the data management system consisted initially
of mean and standard deviation values for the water quality
parameters under study for all mines in the six predeter-^
mined categories. This information was then evaluated, with
the aid of computerized techniques, to determine its statis-
tical significance.
Multiple regression analysis was used to examine dif-
ferences in water quality among the mines in the data base.
An analysis of covariance approach was used in that some of
the explanatory variables included in the regression equa-
tion were defined categorically (e.g., soil characteristics,
effluent source, type of treatment) while other variables
were numerically measured (e.g., precipitation, effluent
flow, mine production). Standard statistical tests based on
the t-statistic were applied to the estimated regression
coefficients to determine which variables are statistically
significant in explaining variations in water quality among
mines.
The regression analysis was performed not only for the
sample as a whole but also for selected subsamples defined
by certain of the category variables. (For example, the
data observed for surface mines defined one subsample, with
the data for underground mines and for preparation plants
defining other subsamples.) This approach permitted an
assessment of interaction effects between the numerical
variables and the categorical variables used to define the
subsamples. A statistical test based on the F-statistic was
used to determine if the relationship between water quality
levels and the explanatory variables could be regarded as
the same for different subsamples.
-------
SECTION 4
RESULTS AND DISCUSSION
RESULTS
Appendix A presents a listing of all computerized
statistical analysis runs conducted. This section presents
a discussion of the results obtained from these computer
runs.
t-test
The first type of statistical analysis run was a test
of differences between groups, a t-test (at a 95% + 99%
significance level). Groupings were constructed by dividing
the data into the following categories:
Mean average annual precipitation:
<15 in. vs. >15 in.
<20 in. vs. >20 in.
<30 in. vs. >30 in.
Effluent source:
Underground mine vs. surface mine
Underground mine vs. preparation plant
Surface mine vs. preparation plant
Acidity vs. alkalinity of sample
Effluent flow: <1 mgd vs. >1 mgd.
The results of these analyses are discussed below.
Precipitation--
Significant differences were found to exist among means
of all water quality parameters tested based on rainfall for
<20 in. vs. >20 in. Similar results were found to exist,
however, when the test was conducted for the 15-in. and 30-
in. demarcation categories as well (see Appendix B). In
fact, the mean values changed little among the three cases
tested.
-------
Effluent Source--
Significant differences among means were found to exist
for surface vs. underground mines for Al, Zn, Fe, Mn, and
alkalinity, and also for log Al, log Fe, and log alkalinity.
Acidity vs. Alkalinity--
Significant differences among means were found to exist
for all parameters based on the alkalinity or acidity of the
sample.
Effluent Flow--
Significant differences among means found for only pH
and log SS values when flows were grouped at <1 mgd and >1
mgd.
Influent Concentration of Pollutant Parameters vs. Effluent
Concentration--
The data base contains only 74 influent cases, 70 of
which are located in the East. Also, the water quality
information corresponding to these cases was found to be
variable (not all parameters were reported in all cases).
Therefore, the influent sample data were found to be inade-
quate to provide any statistically significant results.
Type of Treatment--
Within the data base there exist 25 distinct treatment
categories. These categories are made up of single treat-
ment methods as well as various combinations of such methods
In addition, in many cases the type of treatment utilized,
as well as data quantifying pollutant loadings before and
after treatment, were not reported for all mines. This
large number of potential treatment categories, combined
with the non-uniformity in data coverage concerning pollu-
tant output and method of treatment, results in a very small
data sample for evaluation of each potential treatment cate-
gory. Thus, the level of statistical significance which
could be drawn from an analysis based on treatment type
would yield results of minimal significance. It was deter-
mined, therefore, that the results of such an analysis would
not constitute substantiation for a decision to develop an
effluent treatment category based on type of treatment. Due
to this problem, no computer runs were conducted based on
treatment type.
Regression Analysis
The second type of statistical analysis run was a test
of correlation, a regression analysis. A set of independent
variables was chosen and tested to see if these independent
variables had any predictive value, hopefully explaining the
group differences. The independent variables tested were
average annual precipitation, effluent flow, effluent
10
-------
source, and acidity/alkalinity. Although a number of sta-
tistically significant relationships were found, none of
these had much predictive value. For example, the simple
regression of Al (dependent) on average annual precipitation
(independent) had an r2 of .012, meaning that 1.2 percent of
the variance in Al is explained by rainfall. The signfi-
cance of this relationship is indicated by the F score, if
one assumes that both samples are from normal populations.
The F score was 10.2, indicating a highly significant rela-
tionship, i.e. there was a less than 0.01 percent probabil-
ity that the indicated relationship occurred by chance.
Outlier Exclusion
Performance of a rigorous outlier exclusion exercise
was not possible due to time and funding constraints.
However, in order to illustrate the effects which such
exclusion would have on analytical results, outliers were
excluded from a number of computer runs based on best engi-
neering judgement. Upper concentration limits for the water
quality parameters under study were defined based on expe-
rience. Values in excess of the upper limits cited below
were then eliminated:
Al: 5 mg/1
Zn: 5 mg/1
Fe: 100 mg/1
Mn: 100 mg/1
Acidity: 5,000 mg/1
Alkalinity: 5,000 mg/1
TDS: 10,000 mg/1
SS: 1,000 mg/1
Comparison of mean values of those cases with outliers in-
cluded versus those with outliers excluded resulted in two
observations:
Excluding data points in excess of the above-cited
limits significantly reduced the magnitude of the
mean values of pollutant parameters.
In each case a very small number of data points
were excluded.
11
-------
Thus it can be concluded that a small number of suspiciously
large sample values were responsible for a significantly
large increase in mean values. It is decided, therefore,
that this best engineering judgement outlier exclusion
exercise was reasonable. The majority of computer runs were
then conducted incorporating this best engineering judgement
outlier exclusion.
DISCUSSION
The tests conducted here show that basic differences
exist among different groups of data, but the reasons for
the differences cannot be explained by the data. Specifically,
the following conclusions were drawn from the statistical
analysis for each category tested.
Precipitation
Although significant differences were found to exist
among means for all parameters based on average annual
precipitation (at a 95% + 99% level of significance) , the
mean values obtained were numerically similar in all three
cases tested, i.e., 15-in., 20-in., and 30-in. precipitation
demarcation (see Appendix B). Therefore, although higher
precipitation seems to be associated with lower pH and
higher concentrations of pollutants (and although the oppo-
site is true for lower precipitation), no obvious demar-
cation point was discovered. Also, the regression results
indicated a lack of predictive ability.
Effluent Source
Although significant differences were found among means
for a number of parameters, the regression indicated that no
predictive value could be identified. Therefore, the re-
sults while significant, are inconclusive.
Alkalinity vs. Acidity
Although significant differences were consistently
found among means of the water quality parameters tested at
a 95% + 99% level of significance, the regression again
indicated a lack of predictive ability as to the cause of
these differences.
Effluent Flow
The general lack of significant differences among means
obtained based on effluent flow (at a 1-mgd demarcation)
would seem to indicate that this category would not serve as
an adequate basis for subcategorization. However, a number
12
-------
of general considerations relative to the adequacy of this
data base which are presented in following discussions
indicate that even this conclusion cannot be supported by
the data.
General Considerations
In addition to the above-mentioned considerations, a
number of general inadequacies and limitations have also
been identified in relation to the data base.
The first problem regards the validity of the sample,
which should incorporate a certain degree of uniformity.
For example, data should be taken for all mines at approxi-
mately the same time of year so as to minimize seasonal
variation in the effluent data population obtained. This
was not the case with the data supplied for utilization in
this study which was obtained during different seasons over
a number of years. In addition, valid statistical analyses
should be run on a random sample from the uniform population
(all pollutant concentrations from all coal mines sampled
during a given season). Our sample consisted of all NPDES
data from Region VI, almost all the NPDES data from Region
VIII, and a large amount of data from EGD of unknown origin
(assumed to be a compilation of both previously recorded
data and collected data). A more valid sample could be
obtained by randomly selecting several coal mines in each
Region, recording all NPDES data from these mines, and then
weighting the data according to the total number of coal
mines in each Region. Even this design is somewhat flawed
because NPDES data may not be a timely representative sample
(i.e., many mines may never bother to send in data, and
these may be some of the worst in terms of pollution).
Also, the scope of data reported by mine operators is vari-
able (i.e., the same parameters are not always reported by
all mines). Therefore, an ideal data acquisition effort
would involve either obtaining additional water quality data
so as to supplement NPDES data, or obtaining all data
independent of the NPDES, thereby assuring both consistency
in the data base and adequacy of the type of information
obtained relative to the specific needs of the analysis to
be conducted.
A second consideration regards the distribution of the
variables. The t-test procedure and the F-test of the
regression procedure are predicated on having a normal
distribution in all variables. Water quality data does not
come from a normal distribution. A normal distribution has
zero skew. Water quality data is almost always skewed
because the mean is often close to zero (and a figure of
less than zero suspended solids is inaccurate). A Kolmogorov-
Smirnov test of goodness-of-fit against a normal distribution
13
-------
indicated that none of the parameters under consideration
were anywhere near a normal distribution.
The problem of normality can often be alleviated by
transforming the data. The transformations should be based
on experience or some knowledge about the behavior of the
variable. Average annual precipitation should be from a
normal population according to the central limit theorem
which states that the sampling distribution of means of
random samples of a population will be approximately normal
if the sample size is sufficiently large. Since the average
annual precipitation is a mean, and we had a reasonable
sample size, we assumed in this analysis that precipitation
is normally distributed. This also assumes that rainfall
throughout the United States is a homogeneous population.
Flow and the water quality parameters are probably not
normally distributed. Thus, a transformation is needed to
make these parameters approximate a normal distribution. It
is often assumed in hydrologic work that flow is lognormally
distributed, therefore taking the logarithm of flow would
yield a normal distribution. The same assumption was made
for all the water quality parameters in this evaluation
except pH (which is already the log of the hydrogen ion
concentration) for which a logarithmic transformation was
conducted. The validity of these transformations, however,
is open to question. A Kolmogorov-Smirnov test on the
transformed variables indicated that although they were
closer to a normal distribution than were the untransformed
variables, they were still far enough from a normal dis-
tribution to put the absolute validity of the t-test, F-test,
and regression analyses in question.
A third problem is not one of statistical theory but of
the data. There are many values in the data that are
questionable (and some that are highly unlikely). In addi-
tion, the data base contains a large number of missing
values. The data base is also unsuited to the objective of
proving a significant difference between eastern and western
mines. Data on influents to treatment works are needed to
prove this hypothesis. It has been determined that there
are only 74 influent values in the data base and these are
almost all from the East. Comparison of treatment tech-
nologies within the data base has also proven to be of
little value. For instance, the treatment technology of
settling includes a variety of undersized and poorly main-
tained ponds as well as highly effective installations.
Thus, in order to identify and quantify the role which
utilization of specific treatment technologies plays in
determining effluent quality, that portion of the data base
which contains data identifying the type of technology used
should ideally also contain data specifically characterizing
each of the treatment systems at each mine (design factors,
14
-------
for example). Also, of the samples which were not identi-
fied as influent, some were identified as effluent, and
the remainder had no designation at all. This latter case
existed for more than half of the entire data base. If only
those samples for which the source is identified were to be
included in the analysis the sample would be severely lim-
ited, thus reducing the statistical validity of any results
obtained. If data with no designation as to source were
included in the effluent category there would be no assur-
ance that a large amount of influent data was not thereby
being mixed with effluent data. If this were the case it
would negate the validity of any results obtained. In
conducting the analyses reported, runs were made based on:
Cases identified as influent only
Cases identified as effluent only
Cases identified as effluent only plus those cases
which are unidentified as the source.
It was determined that too few data points are contained in
the influent-only sample to provide any significant results.
In addition, essentially identical results were obtained
when the effluent-only run results were compared with those
from the effluent-only plus undesignated cases (this com-
parison was conducted for precipitation only, as the sta-
tistical results obtained based on precipitation, inclusive
as they may be, were among the best obtained). The fact
that little difference occurred in this comparison is prob-
ably a result of the non-uniformity of the data sample.
15
-------
APPENDICES
APPENDIX A. STATISTICAL COMPUTER RUNS COMPLETED
1. Regression: (All Data):
pH
Zn
Al
Fe
Mn
Acidity
Alkalinity
IDS
SS
PH
Zn
Al
Fe
Mn
Acidity
Alkalinity
IDS
SS
pH vs
Log Zn vs
Log Al vs
Log Fe vs
Log Mn vs
Log IDS vs
Log SS vs
(a) (b) (c)
vs precipitation ; flow ; Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
vs precipitation flow Alkalinity-Acidity
(d)
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
vs Underground, Surface, Preparation Plants
(e) (f)
Alkalinity, Precip., & Flow ; Log Alkalinity, Precip.,
Alkalinity, Precip., & Flow Log Alkalinity, Precip.,
Alkalinity, Precip., & Flow Log Alkalinity, Precip.,
Alkalinity, Precip., & Flow Log Alkalinity, Precip.,
Alkalinity, Precip., & Flow Log Alkalinity, Precip.,
Alkalinity, Precip., & Flow Log Alkalinity, Precip.,
Alkalinity, Precip., & Flow Log Alkalinity, Precip.,
Log Flow
Log Flow
Log Flow
Log Flow
Log Flow
Log Flow
Log Flow
2. Regression: (outliers excluded):
Ph
Log Zn
(a)
vs Log Flow, Precip. & Log Alkalinity
vs Log Flow, Precip. & Log Alkalinity
16
-------
Log Al
Log Fe
Log Mn
Log TDS
Log SS
Log Acid
Log Alk.
vs
vs
vs
vs
vs
vs
vs
Log
Log
Log
Log
Log
Log
Log
Flow,
Flow,
Flow,
Flow,
Flow,
Flow,
Flow,
Precip.
Precip.
Precip.
Precip.
Precip.
Precip.
Precip.
&
&
&
&
&
&
&
Log Alkalinity
Log Alkalinity
Log Alkalinity
Log Alkalinity
Log Alkalinity
Log Alkalinity
Log Alkalinity
3. Scatterplot
ph vs
Log Zn vs
Log Al vs
Log Fe vs
Log Mn vs
Log Acidity vs
Log Alkalinity vs
Log TDS vs
Log SS vs
(a)
Precipitation
Precipitation
Precipitation
Precipitation
Precipitation
Precipitation
Precipitation
Precipitation
Precipitation
4. Kolmogorov-Smirnov Goodness-of-Fit Test
Base
Base
Base
Base
Base
Base
Base
Base
Base
Base
Base
(a)
Data For
Data For
Data For
Data For
Data For
Data For
Data For
Data For
Data For
Data For
Data For
(b)
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
Log Transformed Data
For
For
For
For
For
For
For
For
For
For
PH
Al
Zn
Fe
Mn
Acidity
Alkalinity
TDS
SS
Precipitation
Flow
Al
Zn
Fe
Mn
Acidity
Alkalinity
TDS
SS
Flow
17
-------
5. t-Test (excluding outliers):
PH
Zn
Al
Fe
Mn
Acidity
Alkalinity
TDS
SS
vs
vs
vs
vs
vs
vs
vs
vs
vs
>20 in
>20 in
>20 in
>20 in
>20 in
>20 in
>20 in
>20 in
>20 in
(a)
Precip/<20 in.
Precip/<20 in.
Precip/<20 in.
Precip/<20 in.
Precip/<20 in.
Precip/<20 in.
Precip/<20 in.
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
>30 in.
>30 in.
>30 in.
>30 in.
>30 in.
>30 in.
>30 in.
>30 in.
>30 in.
(b)
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip/<30 in
Precip
Precip
Precip
Precip
Precip
Precip
Precip
Precip
Precip
Log Zn
Log Fe
Log Al
Log Mn
Log Acidity
Log Alkalinity
Log TDS
Log SS
(c)
vs ^20 in. Precip/<20 in. Precip; >30 in.
vs ^20 in. Precip/<20 in. Precip; >30 in.
vs >20 in. Precip/<20 in. Precip; >30 in.
vs >20 in. Precip/<20 in. Precip; >30 in.
vs >20 in. Precip/<20 in. Precip; >30 in.
vs >20 in. Precip/<20 in. Precip; >30 in.
vs >20 in. Precip/<20 in. Precip; >30 in.
vs >20 in. Precip/<20 in. Precip; >30 in.
(d)
Precip/<30 in. Precip
Precip/<30 in. Precip
Precip/<30 in. Precip
Precip/<30 in. Precip
Precip/<30 in. Precip
Precip/<30 in. Precip
Precip/<30 in. Precip
Precip/<30 in. Precip
(e)
pH vs Alkalinity/Acidity; vs
Zn vs Alkalinity/Acidity; vs
Al vs Alkalinity/Acidity; vs
Fe vs Alkalinity/Acidity; vs
Mn vs Alkalinity/Acidity; vs
Acidity vs Alkalinity/Acidity; vs
Alkalinity vs Alkalinity/Acidity; vs
TDS vs Alkalinity/Acidity; vs
SS vs Alkalinity/Acidity; vs
(f)
Surface/Underground;
Surface/Underground;
Surface/Underground;
Surface/Underground;
Surface/Underground;
Surface/Underground;
Surface/Underground;
Surface/Underground;
Surface/Underground;
pH
Zn
Al
Fe
Mn
Acidity
Alkalinity
TDS
SS
vs
vs
vs
vs
vs
vs
vs
vs
vs
(g)
Precip/<15 in. Precip
Precip/<15 in. Precip
Precip/<15 in. Precip
Precip/<15 in. Precip
Precip/<15 in. Precip
Precip/<15 in. Precip
Precip/<15 in. Precip
Precip/<15 in. Precip
in. Precip/<15 in. Precip
in
in
in
in
in
in
in
in
18
-------
Log Zn vs
Log Al vs
Log Fe vs
Log Mn vs
Log Acidity vs
Log Alkalinity vs
Log TDS vs
Log SS vs
(h)
Alkalinity/Acidity
Alkalinity/Acidity
Alkalinity/Acidity
Alkalinity/Acidity
Alkalinity/Acidity
Alkalinity/Acidity
Alkalinity/Acidity
Alkalinity/Acidity
(i)
Surface/Underground
Surface/Underground
Surface/Underground
Surface/Underground
Surface/Underground
Surface/Underground
Surface/Underground
Surface/Underground
Log Zn
Log Al
Log Fe
Log Mn
Log Acidity
Log Alkalinity
Log IDS
Log SS
(j)
vs >15 in. Precip/<15 in. Precip
vs >15 in. Precip/<15 in. Precip
vs >15 in. Precip/<15 in. Precip
vs >15 in. Precip/<15 in. Precip
vs >15 in. Precip/<15 in. Precip
vs >15 in. Precip/< 15 in. Precip
vs >15 in. Precip/<15 in. Precip
vs >15 in. Precip/<15 in. Precip
6. t-Test (outliers not excluded)
pH
Al
Zn
Fe
Mn
Acidity
Alk.
TDS
SS
vs
vs
vs
vs
vs
vs
vs
vs
vs
(a)
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
Acidity/Alkalinity
(b)
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
Surface/Prep Plant
(e)
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
; Underground/Prep Plant
pH
Al
Zn
Fe
Mn
Acidity
vs
vs
vs
vs
vs
vs
Alkalinity vs
TDS vs
SS vs
(d) (e)
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
Surface/Underground ; Flow <1 MGD/Flow >1 MGD
19
-------
PH
Al
Zn
Fe
Mn
<15 in
<15 in
<15 in
<1
L5 in
<15 in
Acidity
Alkalinity
IDS
SS
PH
Al
Zn
Mn
<15 in
<15 in
<15 in
<15 in
Acidity
Alkalinity
IDS
SS
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Log
Al
Zn
Fe
Mn
Acidity
vs
vs
vs
vs
vs
vs
vs
vs
Alkalinity
IDS
SS
Al
ZN
Fe
Mn
Acidity
Alkalinity
IDS
SS
<20
<20
<20
<20
<20
<20
<20
<20
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
vs
(f)
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
. Precip/>15 in. Precip
(g)
in. Precip/>20 in. Precip
in. Precip/>20 in. Precip
in. Precip/>20 in. Precip
in. Precip/>20 in. Precip
in. Precip/>20 in. Precip
in. Precip/>20 in. Precip
in. Precip/>30 in. Precip
in. Precip/>30 in. Precip
(i)
Underground/Prep. Plant ;
Underground/Prep. Plant ;
Underground/Prep. Plant ;
Underground/Prep. Plant ;
Underground/Prep. Plant ;
Underground/Prep. Plant ;
Underground/Prep. Plant ;
Underground/Prep. Plant ;
(k)
Ac id ity /Alkalinity
Acidity /Alkalinity
Acidity /Alkalinity
Ac id ity /Alkalinity
Acidity /Alkalinity
Ac id ity /Alkalinity
Acidity /Alkalinity
Ac id ity /Alkalinity
(h)
; <30 in. Precip/>30 in.
; <30 in. Precip/>30 in.
; <30 in. Precip/>30 in.
; <30 in. Precip/>30 in.
; <30 in. Precip/>30 in.
; <30 in. Precip/>30 in.
; <30 in. Precip /> 30 in.
; <30 in. Precip /> 30 in.
(j)
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Surface/Prep. Plant ;
Precip
Precip
Precip
Precip
Precip
Precip
Precip
Precip
20
-------
Log Al vs
Log Zn vs
Log Fe vs
Log Mn vs
Log Acidity vs
Log Alkalinity vs
Log TDS vs
(1)
in. Precip/>15 in.
in. Precip/>15 in.
in. Precip/>15 in.
in. Precip/>15 in.
in. Precip/>15 in.
in. Precip/>15 in.
in. Precip/>15 in.
Log SS
vs <15 in. Precip/>15 in.
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
Precip;
<20 in
<20 in
<20 in
<20 in
<20 in
<20 in
<20 in
<20 in
(m)
Precip/>20 in. Precip
Precip/>20 in. Precip
Precip/>20 in. Precip
Precip/>20 in. Precip
Precip/>20 in. Precip
Precip/>20 in. Precip
Precip/>20 in. Precip
Precip/>20 in. Precip
(n) (o)
Log Al vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log Zn vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log Fe vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log Mn vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log Acidity vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log Alkalinity vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log TDS vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow >1 MGD
Log SS vs <30 in precip/>30 in precip ; Flow <1 MGD/Flow ^1 MGD
21
-------
APPENDIX B. ILLUSTRATIVE t-TEST RESULTS
TABLE B-l. t-TEST: PRECIPITATION
EXCLUDING OUTLIERS* (95% + 99% LEVEL OF SIGNIFICANCE)
<15 in >15 in <20 in >2Q in <30 in >30 in
pH
Zn(ug/l)
Al(uS/l)
FeO/g/1)
Mn(ug/l)
Acid(ug/D
Alk(mg/l)
TDS(mg/l)
SS(mg/l)
8.0
69
479
591
246
16
495
2724
47
6.7
406
812
5884
11230
328
190
2143
66
8.0
71
449
834
233
16
512
2511
40
6.6
409
819
6326
11309
328
170
2176
70
7.9
70
573
871
217
49
461
2462
42
6.5
421
817
6591
11552
335
167
2171
71
Log ZNQjg/1)
Log AKiug/1)
Log Fe((jg/l)
Log Mn(/ug/l)
Log Acid (mg/1)
Log Alk (mg/1)
Log IDS (mg/1)
Log SS (mg/1)
1.51
2.20
2.42
1.96
0.89
2.60
3.31
1.21
2.03
2.58
2.90
3.0
1.92
1.98
3.11
1.33
1.54
2.17
2.34
1.93
0.89
2.62
3.27
1.18
2.03
2.59
2.95
3.0
1.92
1.94
3.11
1.36
1.57
2.38
2.39
1.94
-
2.55
3.27
1.20
2.04
2.57
2.97
3.02
-
1.93
3.10
1.36
*Mean values (rounded off) are presented in Table 1 for all cases wherein
a statistically significant difference in means was identified by the
t-test.
22
-------
TABLE B-2. t-TEST SCORES: PRECIPITATION
EXCLUDING OUTLIERS* (95% + 99% LEVEL OF SIGNIFICANCE)
Parameter 15-in. case 20-in. case 30-in. case
PH
Zn
Al
Fe
Mn
Acid
Alk
TDS
SS
-17.4
10.0
2.5
13.2
12.5
11.2
- 9.0
- 3.4
2.4
-24.2
10.1
3.4
10.0
12.5
11.2
-10.6
- 2.1
4.6
-22.7
10.7
2.8
11.0
12.6
9.9
-10.2
- 3.3
4.5
Log Zn
Log Al
Log Fe
Log Mn
Log Acid
Log Alk
Log TDS
Log SS
8.4
4.2
8.7
12.0
3.2
3.1
- 6.0
2.9
8.1
4.9
14.6
13.8
3.36
-23.4
- 4.8
5.4
8.8
2.7
14.8
15.3
2.4
-20.8
- 6.1
4.9
*t-scores are presented in Table 2 for all cases in which a
statistically significant difference in means was identified
by the t-test.
23
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
REPORT NO.
2.
3. RECIPIENT'S ACCESSIOONO.
4. TITLE AND SUBTITLE
5. REPORT DATE
TREATMENT CATEGORIES FOR COAL MINE DRAINAGE
April 17. 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
H. Lee Schultz,
Donald Koch,
8. PERFORMING ORGANIZATION REPORT NO.
Carolyn Thompson,
Dr. Kathleen Hereford
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Hittman Associates, Inc.
9190 Red Branch Road
Columbia, Maryland 21045
10. PROGRAM ELEMENT NO.
1NE623
11. CONTRACT/GRANT NO.
68-03-2566
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Resource Extraction and Handling Division
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268 *
13. TYPE OF REPORT AND PERIOD COVERED
Final (1/78 - 1/79)
14. SPONSORING AGENCY CODE
EPA 600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This effort involved the organization and statistical analysis of a large
amount of data characterizing over 300 surface and underground coal mines and
quantification of the pre- and post-treatment quality of their wastewaters. Only
existing data, supplied to Hittman Associates by EPA, was utilized in this
evaluation. The study objective was to determine whether the data supported the
development of a separate Effluent Guidelines subcategory for coal-mining point
source discharges to differentiate between eastern and western coal mining
activities. An extensive effort was required to convert the data into a
computerized format which would allow for expeditious statistical analysis.
Following computerization, a variety of statistical analyses were conducted to
determine if substantiation existed within the data base for development of
treatment subcategories based on any or all of the following factors: precipita-
tion, effluent flow, effluent source, influent concentration of pollutant
parameters, treatment type, and acidity/alkalinity. The results of the statistical
analyses illustrated that the data base provided to Hittman Associates was not
adequate to provide a verifiable basis for subcategorization of the coal mining
point source category. This inadequacy stemmed from numerous problems which
existed within the data base concerning the non-uniformity population and the
non-comparability of much of the data.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Data Analysis
Field Data
Statistical Tests
Coal Mining
Water Quality
Water Treatment
Water Pollution
Mine Water
b.IDENTIFIERS/OPEN ENDEDTERMS
Effluent Guidelines
COSATI Field/Group
13B
13. DISTRIBUTION STATEMEN1
RELEASE TO PUBLIC
19. SECURITY CLASS (Thil Report)
UNCLASSIFIED
21. NO. OF PAGES
29
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
24
------- |