The Mystique, Mystery, Magic, or Marvel of Models
Case In Point - Predicting Ground Water Contamination Nitrate-Nitrogen and Our Health
James W. Bauder
Soil and Water Quality Specialist,
Montana State University
Background Paper
I think I began my interest in models about life below ground as a kid,
digging holes in the "dirt" in a cornfield across the road from my parents'
home in Tewksbury, MA. Maybe it was even before that, while picking rocks
from my uncle's farm fields on a dairy in upstate New York. Anyway, wherever
it started, today, four decades later, I find great fascination in trying to
figure out just what it took to cause something to happen to our ground
water or surface water. And, here in Montana, at the headwaters watersheds
of the Yellowstone, the Madison, the Jefferson, the Gallatin, the Missouri,
the Clark's Fork, there's always something going on that affects the water.
The challenge for me today - learning the cause and effect relationships and
then putting together "models", tools and templates so to speak, that explain
- how processes and practices on the land surface interact with
climate, soil, geology, geography, plants, and social systems to affect
ground and surface water quality; and
- how we can manage the watersheds in the present and future to sustain
their desirable qualities and characteristics.
My interest - in the large scale land use practices that affect the ground
and surface water quality.
I really got interested in modeling when I realized that most events in
society affected physical environment which deal with land use are repeats -
they have been done before and they will be done again. If only we could
figure out the physical cause and effect relationship, then we could use the
knowledge of these relationships to both prevent the occurrence of
'undesirables' in the future and modify our present activities and behaviors
to produce predictable - and desirable - outcomes relative to surface and
ground water quality.
Today my interest in water quality centers on much the same water quality
parameters as it did in the early 1970s when I began work on an irrigation
management project at Utah State University. My interest then - and now -
was the behavior and characteristics of nitrate-nitrogen. Actually, the
nitrogen cycle, but nitrate has a fascination of its own. Considering the
significance of nitrogen - 79% percent of the earth's atmosphere, required
by green plants, a structural component of proteins, amino acids, and many
other life-essential compounds and structure - it has the potential to cause
both acute and chronic toxicity when present as nitrate-nitrogen in drinking
water. It also plays a significant role in estuarian eutrophication.
Nitrate-nitrogen is an intermediate product in the complete nitrogen cycle.
The primary form of nitrogen taken up by plants, nitrate-nitrogen is anionic
and hence readily mobile in soil systems. Nitrate-nitrogen is an
intermediate product of composting, organic matter mineralization,
atmospheric electrical storms; it is the intermediate product of the most
abundantly applied plant nutrient on a global scale - nitrogen. Hence, being
able to define and use the cause-and-effect relationships associated with
nitrogen and nitrate-nitrogen in the environment to develop predictive
models of nitrogen behavior has been a career of many scientists.
The "model" has become a popularly used tool to characterize and predict
nitrogen and specifically nitrate-nitrogen behavior and occurrence. But,
models serve more than just the single purpose of predictive tools. Some
other outcomes or models:
- assessing impact, i.e., sensitivity analyses
- road maps for organizing information
- approaches to organizing thoughts and sequences
- emphasizing cause-and-effect relationships
- identifying gaps in information and understanding
- providing a central focus for discussion
- prioritizing the significance of relationships
- predicting future occurrences or consequences of actions
Along with models have come many problems, too. Case in point: no single
model applies universally to all conditions or situations, i.e., certainly
there will be some situation where the model provides inaccurate predictions
or mis-representation.
And, there are is a wide variety of approaches that can be used in modeling.
I myself have relied heavily on the statistically based, probablistic model
in the past. If two events or parameters, i.e., a) independent variable and
b) dependent variable, can be statistically related or correlated and
scientifically argued or justified, based on solid knowledge, then the
resulting probability has significance. What that means to me is that if
water runs down hill and if streams rise when water enters them from upslope
areas, there must be some correlation between the amount of water applied to
upslope areas and the degree to which a stream rises in a given period of
time after the water is applied upslope, i.e., physically based cause and
effect.
Well.... That's my case with nitrate-nitrogen and models. For nearly three
decades I have been trying to determine the cause and effect relationships
that impact nitrate occurrence in groundwater. When I moved to Montana in
1980 I began studying fallowing - the practice of leaving land idle in
preparation for the next crop. Sort of like composting in your garden. A few
years ago, I started studying the occurrence and distribution of
nitrate-nitrogen in Montana ground water. I then discovered a correlation
(statistical at least) which existed between the extent of fallowing
(composting as you would call it) and unexplainably high concentrations of
nitrate-nitrogen in widely distributed geographic regions of Montana.
Back to measuring, monitoring, predicting and modeling nitrate-nitrogen
occurrences in ground water.
It turns out that these areas that I discovered or reported for
nitrate-nitrogen contamination of ground water were not the same as those
predicted by some others using models based on farming practices and
fertilizer use. Their models were based on the number of farms, the amount
of fertilizer sold, and the amount of wheat production. My numbers were
based on an actual sampling of nearly 2000 private wells across Montana.
Using the information I had available on high nitrate-nitrogen wells and
MAPS, the Montana Agricultural Potentials System, a GIS based data overlay
system, I assembled a statistically supported, probabilistic model, that
allowed me to explain nearly 70% of the variability in my observations about
distribution of nitrate-nitrogen in Montana ground water. The parameters
that I found to be significantly correlated (which could be modeled) with
nitrate-nitrogen concentration were:
- the distribution of high permeability soils in a region
- the percentage of a geographic area with slope greater than 8%
- the percentage of a geographic area with slope less than 2%
- the percentage of farming area 'not' irrigated
- the percentage of a geographic area 'fallowed' (left idle)
- the average 'water holding capacity' of the soil
- the magnitude of difference between rainfall and evaporation
In the end, we eventually developed a model that allowed for the accounting
of 50 to 70% of the variability in nitrate-nitrogen concentration in ground
water using only two independent parameters, those being:
The percentage of a location in selected soil groups or types; and the
percentage of a geographic area in land with slope greater than 2% and less
than 8%.
Today, our investigations are moving ahead in a different direction. As you
will see from a visit to a couple of the WEB sites posted here, statistics
are beginning to surface (no pun intended) which suggest a potential
correlation between non-Hodgkins lymphoma and Parkinson's disease and high
nitrate-nitrogen concentrations in drinking water.
What is so important about nitrates in drinking water?
The following excerpts from a 1996 National Cancer Institute should get your
interest sparked about modeling the occurrence of nitrate-nitrogen in ground
water and surface water systems.
Date: Friday, September 6, 1996
NCI Press Office (301) 496-6641
Nitrate in Drinking Water Associated with Increased Risk for Non-Hodgkin
Lymphoma (NHL)
"Contamination of drinking water with nitrate, a chemical in fertilizers,
may be associated with an increased risk of non-Hodgkin lymphoma (NHL),
particularly in agricultural areas, a National Cancer Institute study
suggests. In a study published in the September issue of the journal
Epidemiology, scientists from NCI, the University of Nebraska Medical Center
in Omaha, and Johns Hopkins University in Baltimore assessed the average
amount of nitrate consumed daily in tap water by Nebraska residents
diagnosed with NHL, a cancer of the lymphatic system, and by a control group
of persons without the disease who lived in the same area. Both groups used
public water supplies....
The more nitrate they consumed in their water, the greater was their
probability of developing NHL. Persons with NHL were twice as likely to be
in the group that consumed the highest levels of nitrate as those without
the cancer..... In Montana, high nitrate concentrations have repeatedly been
detected in some locations in Fergus, Judith Basin, Roosevelt, Daniels,
Sheridan, and Valley counties. However, it is uncertain whether the findings
truly reflect the effect of nitrate, she added. An alternate possibility is
that nitrate exposure is simply a surrogate or a Amarker variable@ that is
correlated with another NHL risk factor that was not directly measured in
the study.
Nitrate levels in ground and surface waters of agricultural regions have
increased over the past 40 years as a result of increases in the use of
nitrogen fertilizers and as a result of the long-term effects of intensive
tillage and organic matter breakdown. Nitrate contamination occurs in
geographic patterns related to the amount of nitrogen contributed by
fertilizers, manure, and airborne sources such as automobile and industrial
emissions, and to soil drainage characteristics. Areas with well-drained
soils and high nitrogen inputs have the highest nitrate levels in the water
supply. In particular, large areas of the Midwestern corn belt states have
nitrate levels above natural levels. The U.S. Environmental Protection
Agency's regulatory limit for nitrate is 10 mg per liter of drinking water."
This document is available through the NCI's CancerNet services on the
Web (http://cancernet.nci.nih.gov) and through Cancer Fax (dial 301-402-5874
from the handset on your fax machine).
Discussion
About the Author
Resources
Science Seminars Archive
|
