D3.1 - CAMELS

advertisement
WP3 Modeling of the 20th century land carbon balance
Studies based on observations of atmospheric CO2, remote sensing, and on carbon process models,
have all indicated that vegetation activity in the Northern Hemisphere is increasing, and this leads to
significant carbon sinks in these regions. A number of factors, such as fertilization effect of the
increase in atmospheric CO2 concentration and nitrogen deposition, interannual climate variability,
and lengthening growing season duration, have been supposed contributing to such increasing trends,
but the main cause of these trends still remain uncertain (Schimel et al., 2001). One of the primary
objective of CAMEL project is to improve our knowledge about the mechanisms of such increasing
trends and recent carbon sinks. To achieve this goal, we use a terrestrial carbon cycle model
ORCHIDEE forced by observed climate (Mitchell and Jones, 2005) and atmospheric CO2 variability to
simulate terrestrial ecosystem carbon fluxes (NPP, HR and NEP) for the period from 1901 to 2002.
Only modeling results from 1980 to 2002 were saved at every daily step to define timing of
phenological events.
Comparison interannual variability in LAI between ORCHIDEE and NOAA/AVHRR
In order to validate ORCHIDEE model simulation results, we used a continuous LAI dataset
(ftp://primavera.bu.edu/pub/datasets/AVHRR_DATASETS/PATHFINDER/VERSION3_DATA/) for
the period 1982-2000 deployed by Global Inventory Monitoring and Modeling Studies (GIMMS)
group to compare interannual variations of remote sensing observed LAI for spring and autumn with
those of ORCHIDEE model-simulated LAI. Fig. 1 shows that the ORCHIDEE model captures
surprisingly well the observed year-to-year LAI fluctuations both in spring and autumn. As a result of
the cooling induced by aerosols emitted by the volcanic eruption of Mount Pinatubo in June 1991,
temperature greatly declined and thus caused a considerable LAI decline during 1992 in both the
simulation and the NOAA/AVHRR observation. However, the fluctuation derived from time series of
NOAA/AVHRR (cv=8.5% for spring and 3.7% for autumn) is much stronger than the result from
ORCHIDEE model (cv=2.8% for spring and 1.0% for autumn) because of the facts that some
variations due to satellite drift/changeover and incomplete corrections for calibration loss and
atmospheric effects (clouds, aerosols, etc.) still remain in the GIMMS LAI dataset (Slayback et al.,
2003). Interestingly, the difference of autumn LAI fluctuation between the two simulations mainly
occurs in 1985, 1989 and 1996, during or one year after which satellite sensor changed.
Interannual variability in C flux during the last century
Figure 2 shows the interannual variabilities in global annual NPP, HR and NEP from 1901 to
2002. The global annual NPP increased by 9.5 Pg C from 67.5 Pg C in 1901 to 77.0 Pg C in 2002,
with a 14.1 % increase or an annual rate of 0.11 Pg C yr-1. The period from 1980 to 2002 shows the
most rapid increase in NPP with increase a rate of 0.4% yr-1, which is consistent with the estimate of
0.34 % yr-1 observed by the remote sensing (Nemani et al., 2003). Global annual heterotrophic
respiration (HR) is also significantly increased during the last century, especially since the 1970s. As
the residue between NPP and HR, global annual NEP is fluctuated around 0 during 1901-1960, while
1
often showed negative values during the last 40 years because the increase rate of NPP is lager than
that of HR. Annual NEP during 1980s and 1990s is -1.3 Pg C yr-1 and -2.4 Pg C yr-1, respectively.
Phenology change during the last two decades
We consider two phonological phenomena: growing season duration (GSL) and carbon uptake period
(CUP). Over the last two decades, both the mean beginning date of GSL and CUP over the Northern
Hemisphere (> 25oN) have advanced at fast rates of 0.16 days yr-1 (R2= 0.20, P=0.03) and 0.20 days
yr-1 (R2= 0.15, P=0.07), respectively (Fig. 3a). Such an advanced spring phenology trend was coupled
with a significant increase in spring temperature (R2= 0.35, P=0.003). For the autumn, there are opposite
trends between the end dates of GSL and CUP (Fig. 3b). Though the mean onset date of vegetation
dormancy has dramatically delayed by 0.14 days yr-1 (R2= 0.30, P= 0.007), the mean end of CUP tend to be
earlier with an advancement rate of 0.07 days yr-1 (R2= 0.31, P=0.006), indicating that extension of GSL
does not necessarily lead to increase in CUP.
We also calculate trends in the onset timing of the phenological events at each pixel using linear least
squares method (Figure 4). Phenology trends are substantially different between North America and
Eurasia, owing to the different patterns of temperature change. Most of northern terrestrial ecosystem
experienced increase in GSL (Figure 4c), usually as a result of earlier beginning of GSL in Eurasia (Figure
4a) and later end of GSL in North America (Figure 4b). For the CUP, areas characterized by extension of
CUP are located mainly in Eurasia (Figure 4f), where CUP beginning shows advanced trend (Figure 4d).
Most parts of North America exhibited declining trend in CUP.
Relationships between C flux and phenology
The correlation analysis between growing season length and different annual C flux (GPP, NPP
and NEP) over the entire study area shows that growing season length is closely and positively
correlated with annual GPP (R2=0.53, P<0.001) and NPP (R2=0.40, P=0.001), but strikingly unrelated
to annual NEP (R2=0.01, P=0.67) (Fig. 5). Regression analysis also show that at continental scale,
extending of growing season will cause an increase in annual GPP by 0.6 % per day (or 5.2 gC m-2 yr-1
per day) and annual NPP by 0.5 % per day (or 2.3 gC m-2 yr-1 per day). Applying estimates of growing
season extension observed in this study and from NOAA/AVHRR NDVI data (0.97 days yr-1 in
Eurasia and 0.64 days yr-1 in North America), the annual NPP of northern ecosystems could be
increased by 0.3-0.5% yr-1 in Eurasia and 0.1-0.3% yr-1 in North America since 1980. This increase
is comparable to the one reported from remote sensing data analysis using empirical light-use
efficiency models, i.e. 0.47 % yr-1 in North America (Hicke et al., 2002), 1.03% yr-1 in China (Fang et
al., 2003) and 0.34 % yr-1 globally (Nemani et al., 2003). Furthermore, the dependence of annual NEP
on carbon uptake period (R2=0.12, P=0.10) is much stronger than that on growing season length
(R2=0.01, P=0.67).
Reference
Schimel, D. S., et al. (2001), Recent patterns and mechanisms of carbon exchange by terrestrial
ecosystems, Nature, 414, 169-172.
Mitchell TD, Jones PD, (2005), An improved method of constructing a database of monthly climate
observations and associated high-resolution grids. International Journal of Climatology, 25,
693-712
Fang JY, Piao SL, Field CB, et al. (2003), Increasing net primary production in China from 1982 to
1999. Frontiers in Ecology and the Environment, 1, 293-297.
2
Hicke, J. A., G. P. Asner, J. T. Randerson, C. J. Tucker, B. Los, R. Birdsey, J. C. Jenkins, C. B. Field,
and E. Holland (2002), Satellite-derived increases in net primary productivity across North
America, 1982-1998, Geophysical Research Letters, 29(10), 1427,
doi:10.1029/2001GL013578.
Nemani RR, Keeling CD, Hashimoto H, Jolly WM, Piper SC, Tucker CJ, Myneni RB, and Running
SW, (2003), Climate-driven increases in global terrestrial net primary production from 1982 to
1999, Science, 300, 1560-1563.
Slayback D, Pinzon J, Los S, Tucker CJ (2003) Northern Hemisphere photosynthetic trends 1982-1999.
Global Change Biology, 9, 1-15.
3
Figure legends
Fig. 1. Interannual changes in anomalies of LAI in the Northern Hemisphere (north of 25 oN) from
ANHRR satellite time series and ORCHIDEE model for (a) spring and (b) autumn.
Fig. 2. Interannual variabilities in global annual NPP, HR, and NEP from 1901 to 2002.
Fig. 3. Interannual changes in (a) dates of spring phenology events and spring temperature and (b)
dates of autumn phenology events and autumn temperature for the Northern Hemisphere (north
of 25oN) from 1980 to 2002.
Fig. 4. Spatial distribution of trends in dates of (a) growing season beginning, (b) growing season end,
(c) length of growing season, (d) carbon uptake beginning, (e) carbon uptake end, and (f) length
of carbon uptake duration.
Fig. 5. The relationships between (a) growing season length and annual GPP, (b)growing season length
and annual NPP, (c) growing season length and annual NEP, and (d)l ength of carbon uptake
period and annual NEP for entire study area.
4
Figure 1
0
0
-0.25
-0.5
1982
-0.1
1985
1988
1991
1994
1997
ORCHIDEE
0.1
ORCHIDEE
0.25
Satellite
Anomalies of LAI
(a) Spring
0.2
Anomalies of LAI
Satellite
ORCHIDEE
Anomalies of LAI
0.5
-0.2
2000
Year
0.5
Satellite
ORCHIDEE
0.25
Satellite
Anomalies of LAI
(b) Autumn
0.1
0
0
-0.25
-0.5
1982
0.2
-0.1
1985
1988
1991
1994
1997
-0.2
2000
Year
5
Figure 2
6
NPP
HR
NEP
80
3
75
0
70
-3
65
1901
NPP (Pg C )
NPP, RH (Pg C )
85
-6
1911
1921
1931
1941
1951
1961
1971
1981
1991
2001
Year
6
Fig. 3
8
6
y = -0.1966x + 101.23
R2 = 0.15
4
90
2
y = -0.1601x + 89.486
R2 = 0.20
80
70
1980
0
y = 0.0568x - 0.2834
R2 = 0.35
1984
Spring Temperature (oC)
Growing season
C arbon uptake
Spring temperature
(a) Spring
100
(Julian day)
Date of spring beginning
110
-2
1988
1992
1996
2000
Year
Growing season
C arbon uptake
Autum temperature
(b) Autumn
y = 0.1364x + 273.97
R2 = 0.30
11
10
270
y = 0.031x + 8.6345
R2 = 0.27
9
260
250
1980
y = -0.0731x + 255.27
R2 = 0.31
Autumn Temperature (oC)
280
(Julian day)
Date of autumn end
290
8
1984
1988
1992
1996
2000
Year
7
Fig. 4
8
Fig. 5
500
a
y = 5.2089x - 101.67
R2 = 0.53
950
NPP (g C m-2 yr-1)
GPP (g C m-2 yr-1)
1000
900
850
800
180
183
186
189
192
b
460
440
420
400
180
195
183
Growing season length (days)
20
c
y = 0.2458x - 70.598
R2 = 0.01
0
-20
-40
-60
180
183
186
189
192
Growing season length (days)
186
189
192
195
Growing season length (days)
195
NEP (g C m-2 yr-1)
NEP (g C m-2 yr-1)
20
y = 2.3394x + 3.1467
R2 = 0.40
480
d
y = -0.9963x + 130.56
R2 = 0.12
0
-20
-40
-60
150
153
156
159
162
165
Length of carbon uptake period (days)
9
Download