Substantial effects of land cover uncertainty
on carbon and climate projections
Difference in forest area between
Max and Min Forest cases (1985-2004 avg)
Alan V. Di Vittorio
With special thanks
to the iESM team
Lawrence Berkeley National Laboratory
Jiafu Mao and Xiaoying Shi
Oak Ridge National Laboratory
LMWG/BGCWG/SDWG Meeting
1 March 2017
Earth Systems and Society Program
2
~15 ppmv CO2 bias in 2004
iESM v1 (CESM 1.1.2,CAM5,CLM4)
Obs: NASA GISS
~392
CO2 concentration (ppmv)
~377
Year
3
~15 ppmv CO2 bias in 2004
iESM v1 (CESM 1.1.2,CAM5,CLM4)
Obs: NASA GISS
~392
CO2 concentration (ppmv)
~377
~7M km2 more forest
increases veg C gain by ~54 Pg and
decreases CO2 gain by ~15 ppmv
over 90 years
Year
4
Different CMIP5 LULC/C can obscure
LULC/C change effects on regional climate
Model (1870)
Model (1992 - 1870)
North America
1974-2004 minus 1862-1891
(Lejeune et al., 2017)
Model
5
• What is the contribution of land cover
uncertainty to simulated carbon cycle
uncertainty?
• How does the land cover-driven carbon
uncertainty compare to the effects of CO2
concentration, nitrogen deposition, and
climate?
• What are the effects of land cover
uncertainty on local/regional climate?
6
iESM-CLM simulations: 1850 - 2004
• Identical CMIP5 land use inputs (* indicates fully coupled sim)
Case
LULCC Reference
LULCC assumptions
No LULCC
Constant 1850
No conversion
Default*
Year 2000
Proportional to PFTs
Max forest*
Previous year
∆ Pasture/crop maximizes forest area
Pasture rule*
Previous year
+ Pasture replaces grass/shrub PFTs first
Proportional*
Previous year
Proportional to PFTs; accounts for pasture
Crop rule
Previous year
+ Crop replaces tree PFTs first
Min Forest*
Previous year
∆ Pasture/crop minimizes forest area
Prop constant CO2
Previous year
Proportional to PFTs (max/min forest also)
Prop const CO2/clim
Previous year
Proportional to PFTs (max/min forest also)
Prop const N dep
Previous year
Proportional to PFTs (max/min forest also)
• Atmosphere: CRU-NCEP, transient CO2, N deposition, and aerosols
7
5.1 Million km2 range in forest area by 2005
2.1 M km2
5.1 M km2
4.9 M km2
3.9 M km2
-200
0
LC uncertainty substantially affects total
ecosystem carbon (Pg C)
43 Pg C
*(37 Pg C, +10 ppmv)
28 Pg C
*(22 Pg C, +6 ppmv)
33 Pg C
*(23 Pg C, ~5 ppmv)
Year
0
24 Pg C
41 Pg C
-200
8
11 Pg C
Year
9
Regional dependence of climate impacts
Max forest minus Min forest (1985-2004 avg)
60
+
-
30
+
0
-30
60
30
0
-30
~
+
~
+
+
~
+
~
0.8
0 (°C)
-0.8
7
% of
grid cell
0
-7
10
Temperature diff (C)
2.0
Raw Data
What drives
temperature sign?
8
66
Difference of Albedo (%)
4
Albedo diff (%)
Difference of Atmospheric
Air Temperature (oC)
1.5
1
1.0
84.50
0.5
67.00
49.50
0
32.00
0.0
14.50
-3.000
-20.50
-0.5
-38.00
-55.50
-1.0
2
-60
-40
-60
00
-2
84.50
0
20
0
Latitude
Latitude
40
60
60
80
67.00
-4
49.50
32.00
-6
14.50
-8
-10
-12
-20
R = -0.39, Sig.
Raw Data
Linear Fit
95% Confidence Band
95% Prediction Band
-3.000
-20.50
-38.00
-55.50
5
10
15
20
25
30
35
00
30
Tree Percentage (%)
DifferenceDifference
in Treeof Cover
(% of grid cell)
Max Forest minus Min Forest
(1985-2004 avg)
(tree diff > 0)
11
Sensible heat may dominate latent heat
Max Forest minus Min Forest
(1985-2004 avg)
(tree diff > 0)
10
10
Sensible Heat diff (W/m2)
Latent Heat diff (W/m2)
88
6
4
2
00
Raw Data
Linear Fit
95% Confidence Band
95% Prediction Band
-8
R = 0.26, Sig.
0
0
5
10
15
20
25
Difference of Tree Percentage (%)
30
30
2
49.50
-2
14.50
-6
4
00
32.00
-4
6
67.00
84.50
-2
-10
Difference of Latern Heat Flux (W/m-2)
8
-3.000
-4
-6
-20.50
-38.00
-55.50
35
-8
Raw Data
Linear Fit
95% Confidence Band
95% Prediction Band
-10
-12
R = 0.007, Sig.
0
5
10
15
20
25
30
0 Difference of Tree Percentage (%) 30
Difference in Tree cover (% of grid cell)
12
Albedo vs Evaporative Fraction
Max Forest minus Min Forest
(1985-2004 avg, tree diff > 0)
-0.10
12
-0.05
0.00
0.05
0.10
12
Difference of Albedo (%)
10
10
Temperature Difference
8
0.3
6
0
-0.3
8
6
4
4
2
2
0
0
-2
-2
-4
-4
-6
-6
-8
-10
-8
Tree Difference
5 10 15 20
-12
-0.10
25
-10
30
-0.05
0.00
0.05
Difference of Evaporative Fraction
-12
0.10
13
Latent heat and temperature
Max Forest minus Min Forest
(1985-2004 avg, tree diff > 0)
-6
-4
-2
0
Tree Difference
Tree
Difference
6
4
6
8
Temperature
Difference < 0
0.500
30
30%
-10
10
Temperature Differencre
10
10%
20
20%
8
2
0.00
10
10
8
8
6
4
4
2
2
0
0
-2
-2
-4
-4
-6
-6
-8
-8
-10
-10
-10
-8
-6
-4
-2
0
2
4
6
8
Difference of Latent Heat Flux (W/m -2) 2
Difference in Latent Heat (W/m )
10
Difference of Albedo
10
-8
Difference in Albedo (%)
-10
-8
-6
-4
-2
0
Tree Difference
Tree
Difference
4
6
8
10
Temperature
Difference > 0
Temperature Differencre
10
10%
20
20%
0.500
30
30%
6
2
0.00
10
8
6
4
4
2
2
0
0
-2
-2
-4
-4
-6
-6
-8
-8
-10
-10
-10
-8
-6
-4
-2
0
2
4
6
8
Difference of Latent Heat Flux (W/m -2) 2
Difference in Latent Heat (W/m )
10
14
Summary
•
Land cover uncertainty contributes ~5 ppmv
uncertainty to CO2 concentration
• This is about 1/3 of original bias
•
•
•
•
•
•
•
Land cover uncertainty can contribute
on the order of 1 degree to temperature
uncertainty
This is on the order of LULCC-attributed effects
Regional dependence
Sensible heat vs albedo vs latent heat?
Local vs distributed
LULCC-driven uncertainty is reducible
LULCC is an integrated process
14
Summary
•
Land cover uncertainty contributes ~5 ppmv
uncertainty to CO2 concentration
• This is about 1/3 of original bias
•
•
•
•
•
•
•
Land cover uncertainty can contribute
on the order of 1 degree to temperature
uncertainty
This is on the order of LULCC-attributed effects
Regional dependence
Sensible heat vs albedo vs latent heat?
Local vs distributed
LULCC-driven uncertainty is reducible
LULCC is an integrated process
14
Summary
•
Land cover uncertainty contributes ~5 ppmv
uncertainty to CO2 concentration
• This is about 1/3 of original bias
•
•
•
•
•
•
•
Land cover uncertainty can contribute
on the order of 1 degree to temperature
uncertainty
This is on the order of LULCC-attributed effects
Regional dependence
Sensible heat vs albedo vs latent heat?
Local vs distributed
LULCC-driven uncertainty is reducible
LULCC is an integrated process
Questions?
Difference in 2004 forest area between
the Max forest case and
the Proportional case
This work is supported by the Director, Office of Science,
Office of Biological and Environmental Research of the
U.S. Department of Energy under Contract No. DEAC02-05CH11231 as part of the Integrated Assessment
Research and Earth System Modeling Programs.
E1
Different LUCID LULC/C can obscure
LULC/C change effects on regional climate
Model (1870)
Model (1992-1870)
North America
1974-2004
minus
1862-1891
(Lejeune et al., 2017)
MM M
Model