Transcript The combined effects of land use and climate change on river and

Current and future effects of
land use and climate change on
river and stream salinity
John Olson
Desert Research Institute
Society for Freshwater Science Annual Meeting
23 May 2016
River & stream salinity
• Varies over four orders of magnitude,
specific electrical conductivity (EC) from < 10
to > 12,000 µS/cm
• Human activities can increase salinity, to
greater than seawater (> 54,000 µS/cm)
Photo: PBS
Photo: Wikipedia
Photo:
Wikipedia
Photo: Kalfka et al. 2001
Photo: Simon Fraser University
River & stream salinity
Evaporative Cooling
Mohseni et al. 2003
River & stream salinity
Loss of water resources
Loss of low salinity habitat :
• ↓ freshwater biodiversity
• Alters ecosystems
function
>3,000 µS/cm unsuitable:
• Irrigation
• Industry
• human consumption
Proportion of
Invertebrate Genera
Ecosystem impairment
USEPA 2011
EC (µS/cm)
To effectively address this challenge, we
need to know:
• How much has salinity in streams already
been altered by human activities?
• How much more alteration is likely with
increasing development and climate
change?
Objectives
Climate & Atmospheric
Inputs
Biota
Human
Rock
Weathering
Soil
Runoff Water
1. Model natural background: Natural EC = f(geology, climate, …)
2. Compare to current salinity: Observed – Natural = EC Alteration
3. Model salinity alteration: EC Alteration = f(urban, ag, mining, …)
4. Predict salinity in 2100: f(future climate, urban, ag, …) = Future EC
Response Data
Models based on in situ EC measurements at
minimally impacted sites by multiple agencies
• 1935 sites
• Measurements
made at “baseflow”
Predictor Data
Watershed averages of 24 static predictors
% CaO
Soils
Deposition
Topography
Geophysical
Geochemical
UCS (MPa)
1-5
1-5
5 - 12
5 - 35
12 - 18
% CaO
18 - 25
1-5
75 - 145
1-5
25 - 35
5 - 12
145 - 215
5 - 35
215 - 250
35 - 75
35 - 46
12 - 18
18 - 25
25 - 35
35 - 46
35 - 75
UCS (MPa)
75 - 145
145 - 215
215 - 250
Predictor Data
Watershed averages of 12 dynamic predictors
from published climate & land use projections
• IPCC A2 mid-high emission scenario
• Current (2010) and end of century (2100)
Climate
Land use / Land cover
• From Hawkins et al. 2013 • From Sohl et al. 2014
• Climate model: NCAR
• USGS FORE-SCE
Community Climate
model
System Model v3.0
Predictor Data
Climate
• Dynamically downscaled (WRF Regional Climate Model)
• Statistically downscaled (PRISM Climate Data)
• Corrects GCM biases, incorporates topography
Temperture
Increase,
deg C
1-3
3-6
6-9
>9
Change in mm/yr
<-200
-200 - -100
-100 - -50
-50 - -25
-25 - 0
0 - 25
25 - 50
50 - 100
100 - 200
> 200
Predictor Data
Land use / Land cover
• FORE-SCE model spatially allocates land use
change using logistic regression models
• Based on biogeophysical and socioeconomic
2005
conditions
IPCC scenarios
Forest Age
Protective
Status
Historical
Land Cover
Spatial
Allocation
(FORE-SCE)
Suitability Surface
Annual
LU/LC
Maps
2100
Natural Background Salinity Model
Observed EC, S/cm (log scale)
5
20
100
500
Random Forest Model Performance
EC Natural Background Model
RMSE = 67 µS/cm
R2 = 0.78
1:1 line
5 10
50
200
1000
Predicted EC, S/cm (log scale)
Natural Background Salinity Model
Random Forest Model Predictors
20
100
0
2
4
3000
% Conifer
250
250
250
0.6
0
100
Rck Strength
100
250
100
0.0
0
30
Rock % S
Atm Ca Dep
100
100
10
100
20 40
100
100
0
Yearly Precip
250
Min Precip
250
Max Temp
250
250
Rock % CaO
0
100
0.0
0.6
Change from Natural Background Salinity
• 2,001 sites, part of the
U.S. EPA’s National River
and Stream Assessment
Sites chosen using a
probability-based sample
design
• Statistically represent
U.S. streams
Current Observed
EC (µS/cm)
Change from Natural Background Salinity
Predicted Natural
Background EC (µS/cm)
Current Observed
EC (µS/cm)
Change from Natural Background Salinity
Change from Natural Background EC
Photo: Simon Fraser University
Kaushal et al. 2005
Alteration → Response Data
Photo: Wikipedia
Percent change
Photo: PBS
Photo: Wikipedia
Absolute change in µS/cm
Salinity Alteration Model
Random Forest Model Performance
Observed Alteration, S/cm
1
10 100
10000
EC Alteration Model
RMSE = 604 µS/cm
R2 = 0.60
1:1 line
1
10
100
1000 10000
Predicted Alteration, S/cm
Salinity Alteration Model
Random Forest Model Predictors
0.6
800
200
0.0
0.6
800
0
100
0.0
% Mining
200
200
800
200
0.0
8
Rck Strength
800
% Urban
4
0.6
Soil Bulk D
800
0
80
% Crops
200
40
200
200
800
200
0
Atm Ca Dep
800
Rock % S
800
Min Precip
0.0
0.3
0.8
1.4
End of Century Salinity
Current
Observed
Natural
background
model:
EC (µS/cm)
Natural EC = f(geology, climate)
Climate
Projected
Land use
Alteration model :
EC Alteration = f(urban, ag)
Predicted End of
Century EC (µS/cm)
End of Century Salinity
Change in EC from current to 2100
Percent change
Absolute change in µS/cm
Change in Salinity
Median: use
524 threshold
µS/cm
Human
> 3,000
µS/cm
Median:
319 µS/cm
Currently - 3% of sites
Median: 214 µS/cm
2100 – 11% of sites
Change in Salinity
Invertebrate Extirpation Thresholds
< 300 µS/cm [5% Extirpation]
Background – 65% of sites
Currently – 48% of sites
2100 – 26% of sites
< 2000 µS/cm [50% Extirpation]
Background – 100% of sites
Currently – 93% of sites
2100 – 88% of sites
1500
1000
500
0
Climate Change
Only
Climate & Land
Use Change
• We calculated the
effects of climate
change on salinity
separate from the total
change
• Climate change
accounts for < 10% of
changes in salinity in
2100
Change in EC ( S/cm)
Change in Salinity
Conclusions
• Still work to be done:
• Quantify uncertainty
• Better parse climate & LU impacts
• Assess vulnerability
• Human activities have already increased median
salinity by > 100 µS/cm, potentially doubled by
end of century
• Climate change only accounts for < 10% of
change by itself
• Half of remaining low salinity habitat may be lost
by end of century
Questions?
Acknowledgments:
GIS analysis: Megan Dettenmaier,
Ben Poulsen, and Nate McPherron
Funding: Sulo and Aileen Maki
Endowment