Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Using GIS to Predict Arsenic Over 10 ppb in Drinking Water
OrHow to lie with statistics and maps
Rich SouleMDH
All Blue
All Red
Michael Gastner, Cosma Shalizi, and Mark Newman University of Michigan
Michael Gastner, Cosma Shalizi, and Mark Newman University of Michigan
Make a Statewide Map of Arsenic in Groundwater in Less Than 3 Hours
1) Come up with a single physical process.It must be mapped at state scale.
2) Find the data sets representing this process and arsenic.
Evaluate data set bias.3) Generate a Probability Map for Arsenic
over 10 ppb.
Clay
Sand
Sand withIron
ReducedBlack, green and grey
Rotten eggs odor.
Red, brown and, tanLittle to no odor.
Deep
Shallow Oxygen+
Carbon+
MicrobesHAsO4HAsO4--22
HAsO4HAsO4--22
As+5
HAsO4-2
As+3
H2AsO3 Ars
enic
Confining Thickness
HydrogeologicModel
Ars
enic
Time
BiologicalModel
GeochemicalModel
Oxidized
Conceptual Model• Pr(As) > 10 ppb ≈ Clay Thickness.
• Groundwater is reduced by microbes at a constant rate and is limited by electron acceptors (O2, FeNO3, Fe ).
• The flow of electron acceptors is determined by vertical groundwater recharge. – Q=KiA, Assume average “K” and “i” .
• Assume Steady State.
Why?: Spatial RelationshipsArsenic & DM Lobe Wood & DM Lobe
Arsenic >10 ppbArsenic <10 ppb
Data Sets: Geology (CWI)VerifiedLocations
RawLocations
170,000Wells
200,000Wells
Use it All.
VerifiedLocations
RawLocations
Bias: Well Density
Summarizing Well Logs
37 ft
56 ft
21 ft
15 ft
Clay Thickness
+Total 138 ft
Clay
Clay
Clay
Clay
Do this370,000 Times.Takes 20 min.
5 kmGrid
10 kmGrid
Grid Clay Thickness
> 100 Ft.
None
Clay Thickness
Grid Clay Thickness
1 kmGrid
1 kmGrid with Brakes
> 100 Ft.
None
Clay Thickness
No Data
MARSAreaLooks Really Bad….Population Declining….Co-incidence?
Data Sets: Arsenic
Yes
No
Yes
Location
High As?Deeper?
YesWest Central
MARS
State
State
Area
No
Yes
Geology Bias Data
No LocationShallower?
Construction?
CDC/MDH
Deeper?Construction?
GWMAPS
Use MARS Data?
Over 10 ppbNumber in AreaData Set
49%47CDC48%50GWMAPS52%893MARS
Yes, but cautiously.
Get Clay Grid Values
10
%
%
0 20 40 60 80 100Avg. Feet Clay Above Screen
0
20
40
60
80
Pr (
As>
10)
Probability As >10 ppb All Arsenic Data
All Data
MARS Data
GWMAPS
CDC Data
Low
Medium
High
Medium
0 20 40 60 80 100Avg. Feet Clay Above Screen
0
20
40
60
80
Pr(
As>
10)
Mapping Arsenic Using “Blocks”
All Data
MARS Data
GWMAPS
CDC Data
Results
10 ppb
Results
10 ppb
Low
Medium
High
Medium
Score Card
0 20 40 60 80 100Avg. Feet Clay Above Screen
0
20
40
60
80
Pr(
As>
10)
MARS
All
MARS
MARS
MARS
All
AllAll
GWMAPS
GWMAPS
GWMAPS
GWMAPS
CDC
CDC
CDC CDC
Implications• State Scale: the proposed processes may impact
the arsenic concentration in groundwater.• County Scale: These processes are probably
variable (i.e. K, infiltration…).• Section Scale: The change of the redox state of
groundwater through time may be the most important factor.
MPCA & others
Changing the Redox StateBorehole Chemistry & Microbial Ecosystems
“Reducing Factors” Increase Arsenic:Decreasing Infiltration (Tile Drains)Increasing Carbon Load (Monoculture)
“Oxidizing Factors” Decrease Arsenic:Increased Nitrate Loading
Acknowledgements• Arsenic Lunch Group:
Mindy Erickson Randal BarnesRoman Kanivetsky Karla PetersonDavid Rindal Mike Berndt
• MDHBruce Olsen Sheila GrowBrian Johnson Justin BlumAl Epp Mike Baker