How much do different land models matter for climate simulation?
Jiangfeng Wei
with support from Paul Dirmeyer, Zhichang Guo, Li Zhang, Vasu Misra, and James Kinter
COLA/IGES
Motivation
Uncertainty of land surface models
•
significantly different output at the same forcing (e.g., PILPS, GSWP)Complexity of land-atmosphere interaction
•
full of nonlinear processes•
uncertainties in land simulation may be brought to atmosphereSources of the signals are hard to trace in the complex system
3
Models
AGCM SSiB
AGCM CLM
AGCM Noah
AGCM
Experiments
Exp I (individually coupled runs)
Exp C (combined run, interactive ensemble)
Average fluxes from 3 land models
Same atmospheric forcing for 3 land models
5
All the simulations start from April 1, 1982 and end on January 1, 2005 (close to 23 years). The last 18 years of data is used for analysis.
The atmospheric initial condition is from NCEP/NCAR
reanalysis, and land initial conditions are from long-term
offline simulations.
Tropical land (25S-25N)
Northern land
(25N-70N)
7
Inter-model differences of JJA climatology
Var(I ) Var(C) Var(I )
Var(I) and Var(C) are the inter-model (3 cases) variances of fluxes from land to atmosphere in experiments I and C, respectively.
Var(I): land model difference + land-atmosphere feedback Var(C): land model difference only
: the percentage of inter-model variance caused by land-atmosphere feedback If Var(I)Var(C), 01, else, <0.
Define
Can land-atmosphere interaction amplify the uncertainties
from land models?
9
Var(I )Var(C) Var(I )
: the percentage of inter- model variance caused by land- atmosphere feedback
01: land-atmosphere interaction can amplify the spread caused by land model differences
<0: the spread decreases when coupled to the AGCM
For the colored area (>0): (LH) has much larger value than for (SH) because LH is more strongly influenced by
precipitation.
The largest value of (LH) is generally over semi-arid regions, where precipitation uncertainties influence LH most.
For SH, only about half of the inter-model spread is caused by the different forcings over land and another half is from LSS differences.
Var(I )Var(C) Var(I )
: the percentage of inter- model variance caused by land- atmosphere feedback
11
1987-2004 JJA interannual variation
LH-SWnet
Water limiting LHSWnet
Energy limiting
The evaporation regime largely determines how the spread of LH among LSSs changes.
Std dev of T
min, T
max, and DTR(=
Tmax- Tmin)among 3 LSSs
“Pure” influence of land model uncertainties
Land uncertainty + feedback
More impact on Tmax than on Tmin, mainly through LH
13
Summary for part I
1. The choice of LSS has significant impact on the model hydrological cycle.
2. The evaporation regime largely determines how the spread of LH among LSSs changes.
3. In coupled GCM simulations, most of the LH uncertainties over semi-arid areas are caused by the precipitation difference and LSS differences have very little influence, while only about half of the inter-model differences of SH over land are caused by the forcing difference and another half is from LSS differences.
4. The uncertainties of LH among the LSSs have strong influence on
surface temperature, and it has more influence on T
maxthan on T
min.
The influence is stronger in dry regions/seasons, where LH has more
uncertainty. Land-atmosphere interaction can weaken the influence of
LSS uncertainties in the tropics, but may strengthen their influence in
middle to high latitudes.
Memory of land models
15
Causes of low LH memory in Noah model:
Tropics: percentage of canopy interception is too high
Middle to high latitudes: high percentage of interception and low memory of vegetation transpiration
17
Global Land-Atmosphere Coupling Experiment
16-member ensembles for 1 June- 31 August of 1994 (SST prescribed)
Ensemble W: control integrations
Ensemble S: subsurface soil moisture is given the same as one member of W
16
2X
2X15
2X(0 1)
Ω measures the similarity (or predictability) of the time series in 16 ensemble members, and is equivalent to the percentage of variance caused by the slowly varying oceanic, radiative, and land surface processes.
(S)-(W)
is the predictability come from the prescribed subsurface soil moisture, and is a measure of land-atmosphere coupling strength in GLACE.
Koster et al. (2004, 2006)
Ω shows similar patterns for 3 models, with largest values in the tropical rain belt where the SST forcing has strongest influence.
19
Percentage of intraseasonal (30–100 day) precipitation variance calculated from the CMAP for the years 1982–2002.
(Pegion and Kirtman 2008).
global mean
spatial correlation with (S)-(W) spatial correlation with (W)
Obs.
intraseasonal variance?
(W)
21
Calculated with the intraseasonal component of precipitation time series
Most of the precipitation predictability () and land-atmosphere coupling strength ((S)-(W)) are associated with the intraseasonal component of precipitation in the models, although they only account for a small percentage (~20%) of the total variance.
Results from models participating in GLACE
23
Conceptual relationships
F(
0 )
Based on the above analysis, we can build a conceptual relationship:
0is a constant, and
0>>. Thus, the spatial variation of is largely determined by F. Then the coupling strength
(S) (W ) F( (S) (W ))
(S)- (W) is the difference of between the two ensembles, and is the “pure” impact of soil moisture on the coupling strength (i.e.
without the influence of external forcing).
It is evident that both F and (S)- (W) can impact the coupling strength greatly.
F: the impact of low-frequency external forcing
: the impact of soil moisture
(S) (W ) F( (S) (W ))
For the three individually coupled models, F is similar and (S)- (W) causes the large difference in coupling strength. Noah model should have the smallest (S)- (W).
If F is equivalent to the percentage of intraseasonal variance, it should have been overestimated by the models. We can adjust (S)-(W) as
((S) (W ))adjusted ((S) (W )) var %(obs) var %(model)
< 1
25
Adjustment of the GLACE coupling strength
Difficult to evaluate by comparing with
observation.
Hard to say whether it is more realistic.
Another physical explanation:
More low-frequency variation of precipitation (rains too frequently at reduced intensity)
Precipitation has prolonged impact on soil moisture
Less runoff, more ET, increased soil moisture memory
Overestimation of land- atmosphere coupling
Summary for part II
1. Different land models or subsurface soil moisture have little influence on the global pattern of precipitation predictability () and variance distribution because of the stronger control of other factors.
2. The regional effect of soil moisture can be highlighted by the difference of from two ensembles, which shows contrasting patterns for the three models.
3. Most of the precipitation predictability and land-atmosphere coupling strength are associated with the intraseasonal component of precipitation in the models.
4. Most models have overestimated the low-frequency variance percentage and underestimated the high-frequency variance percentage of precipitation.
5. Based on the findings, we adjust the land-atmosphere coupling strength
27
References
• Wei, J., P. A. Dirmeyer, Z. Guo, L. Zhang, and V. Misra, 2009: How much do different land models matter for climate simulation? Part I: Climatology and variability. COLA Tech. Rep. 273. 35pp. [Available online at
ftp://grads.iges.org/pub/ctr/ctr273_ms.pdf]
• Wei, J., P. A. Dirmeyer, and Z. Guo, 2009: How much do different land models matter for climate simulation? Part II: A decomposed view of land- atmosphere coupling strength. COLA Tech. Rep. 274. 27pp. [Available online at ftp://grads.iges.org/pub/ctr/ctr274_ms.pdf]
