Upload
warren-osborne
View
239
Download
0
Tags:
Embed Size (px)
Citation preview
MULTIPLE BIOME-CLIMATE EQUILIBRIA IN AMAZONIA MULTIPLE BIOME-CLIMATE EQUILIBRIA IN AMAZONIA AND PERSPECTIVES FOR THE FUTURE OF THE AND PERSPECTIVES FOR THE FUTURE OF THE
TROPICAL FORESTTROPICAL FOREST
MULTIPLE BIOME-CLIMATE EQUILIBRIA IN AMAZONIA MULTIPLE BIOME-CLIMATE EQUILIBRIA IN AMAZONIA AND PERSPECTIVES FOR THE FUTURE OF THE AND PERSPECTIVES FOR THE FUTURE OF THE
TROPICAL FORESTTROPICAL FOREST
Carlos A. NobreCarlos A. Nobre
CPTEC/INPE, Cachoeira Paulista, SP- BrazilCPTEC/INPE, Cachoeira Paulista, SP- Brazil
CCC Symposium: Amazon Climate and Hydrology9-10 May 2005, Duke University
Contents
• Biome-Climate interactions in Tropical South America: the CPTEC Potential Biome Model; multiple biome-climate stable equlibria; impacts of global warming on biome redistribution in South America.
• Are biomass burning aerosols in the atmosphere causing a reduction of dry season rainfall or a delayed onset of the rainy season in Southern Amazonia?
Vegetation-Climate Interactions
Climate Vegetation
Bidirectional on various times scales
Climate determines vegetation!
The Holdridge Life-Zone Classification System (Holdridge, 1947; 1964)
Savanna/Dry Forest
A Planet of Climate Determinist!
Atlantic rainforest
What does determine tropical forest and savanna distribution?
Area of Study
VegetationA, Aa, Ab, AsC, CsD, Da, Db, Dm, DsF, Fa, FsLO, La, Ld, LgONP, Pa, PfS, SM, SN, SO, ST, Sa, Sd, Sg, SpTd, TpWATERrm
N
EW
S
-45,-15
-45,-10
-45,0
-50,5-55,5-60,5-65,5-70,5
-70,0
VegetationTypes in Brazilian an AmazoniaRadam
200 100 0 200 M
100 100 0 200 km
Scale 1:16.000.000Projection
longitude of central meridian - 57 00 00
What enviromental factors do explain What enviromental factors do explain tropical savannas?tropical savannas?
• Fires?Fires?• Soils (biogeochemistry)?Soils (biogeochemistry)?• Climate?Climate?• Others?Others?
Various authors (e.g., Leopoldo Coutinho, IB/USP) postulate that the most important environmental factors driving the existence of tropical savannas are fires and soil biogeochemistry; the mean climate would be compatible with a tropical seasonal dry forest and not with the typical tropical savannas of South America and Africa.
Annualprecipitation
Meanclimaticequator
Arid Savanna Rainforest Savanna Arid
Growing season
Gro
win
g s
easo
n le
ng
th in
mo
nth
s
Mea
n a
nn
ual
pre
cip
itat
ion
in m
m
South Equator North Latitude
A scheme of the relationship between mean annual precipitation and growing season length in tropical climates (from Newman, 1977)
Sombroek 2001, Ambio
Map. no. 1
Annual rainfall (mm)
Annual Rainfall (mm)
DEZ-FEV
SET-NOVJUN-AGO
MAR-MAI
Precipitação (mm)
>900
600-900
300-599
<300
Nilo and Nobre, 1991, Climanálise
Seasonal rainfall totals shows strong seasonality of climate in Amazonia
Sombroek 2001, Ambio
Map of dry season length (DSL) (data after Sombroek, 2001), expressed as the number of months with <100 mm of rain.
Steege et al., Biodiversity and Conservation 12 (in press), © 2003 Kluwer Academic Publishers
TR
OP
ICA
L F
OR
ES
T C
OV
ER
CLIMATE STATE
FORESTFOREST
SAVANNASAVANNA
(DRY SEASON PRECIPITATION OR LENGTH OF THE WET SEASON )
BIOME DISTRIBUTION BIOME DISTRIBUTION RESPONDS TO CLIMATE !RESPONDS TO CLIMATE !
1
0
Fig. 3 Establishment of relative forest area in a savanna region as a function of precipitation.
Sternberg, 2001, Global Ecology & Biogeography, 10, 369–378
Climate conditions for tropical savannasClimate conditions for tropical savannas
• Tmean > 24 C
• 13 C < Tcoldest month < 18 C
• P3 driest months < 50 mm
• P6 wettest months > 600 mm
• 1000 mm < Pannual < 1500 mm
To what extent does vegetation determine climate?
PE
C
Simple Atmospheric Water Balance
P = E + C
P = PrecipitationE = EvapotranspirationC = Moisture Convergence
From forest to pasture...
Simulating the impacts of deforestation
Forest Pasture
Caatinga
Cerrado
CerradoAtlântic Ocean
Pacífic Ocean
P pasture - P forest ( annual, in mm)
EFFECTS OF LARGE SCALE DEFORESTTIONEFFECTS OF LARGE SCALE DEFORESTTION
Rocha, 2001.
Summary of Numerical Simulations of deforestation • 1 to 2.5 C surface temperature increase (verified by observations!)• 15% to 30% evapotranspiration decrease (verified by observations!)• 5% to 20% rainfall decrease (still inconclusive observations!)
PR
EC
IPIT
AT
ION
TROPICAL FOREST COVER
CLIMATE RESPONDS CLIMATE RESPONDS TO VEGETTION !TO VEGETTION !
PPpresentpresent
0.8 P0.8 Ppresentpresent
P present 2 to 2.5 m
0,6
0,7
0,8
0,9
1
1,1
1,2
0 20 40 60 80 100Desfl orestamento (%)
Prec
ipit
ação
Rel
ativ
a
. Avissar et al 2002
Conceptual models of regional deforestation in Amazonia
• A Potential Biome Model that uses 5 climate parameters to represent the (SiB) biome classification was developed (CPTEC-PBM).
• CPTEC-PBM is able to represent quite well the world’s biome distribution. A dynamical equilibrium vegetation model was constructed by coupling CPTEC-PBM to the CPTEC Atmospheric GCM (CPTEC-DEBM).
CPTEC Potential Biome Model
Geog
rap
hy
Ecolo
gy
Modeling of Geographical Distribution of Species
Points of occurrence
Algorithm Precipitation
Tem
pera
ture
Ecological Niche Model
Prediction of
distribution
Geog
rap
hy
Ecolo
gy
Modeling the Geographical Distribution of Biomes
Area of Occurrence
Algorithm Variable A
Var
iabl
e B
Biome Model
Prediction of
distribution
Five climate parameters drive theFive climate parameters drive the potential vegetation model potential vegetation model
Oyama and Nobre, 2002
Monthly values of precipitation and temperature
Water Balance Model
Potential Vegetation Model
SSiB Biomes
Figure 6. Environmental variables used in CPTEC PVM: growing degree-days on 0oC base (a), growing degree-days on 5oC base (b), mean temperature of the coldest month (c), wetness index (d), seasonality index (e). Growing degree-days in oC day month-1,
and temperature in oC.
growing degree-days on 5oC base
Oyama and Nobre, 2004
growing degree-days on 0oC base
Wetness index
mean temperature of the coldest month
Oyama and Nobre, 2004
Oyama and Nobre, 2004
seasonality index
The potential vegetation model algorithm
Oyama and Nobre, 2004
Tropical Forest
Visual Comparison of CPTEC-PBM Visual Comparison of CPTEC-PBM versus Natural Vegetation Mapversus Natural Vegetation Map
CPTEC-PBM
SiB BiomeClassification
Oyama and Nobre, 2004
62% agreement on a global 2 deg x 2 deg grid
Visual Comparison of CPTEC-PBM versus Natural Vegetation Map
SiB BiomeClassification
NATURAL VEGETATION POTENTIAL VEGETATION
Oyama and Nobre, 2004
Statistic (Monserud e Leemans 1992)
good agreement
poor agreement
Oyama and Nobre, 2004
agrement
perfect
excel.
v. good
good
regular
poor
v.poor
none
Objective verification of CPTEC-PBM
bioma nome p0 (%) concordância
1 floresta tropical 71 0,73 muito boa
2 floresta temperada 52 0,49 regula
3 floresta mista 26 0,26 pouca
4 floresta de coníferas 55 0,56 boa
5 lariços 70 0,65 boa
6 savana 56 0,60 boa
7 campos extratropicais 76 0,50 regular
8 caatinga 50 0,40 regular
9 semi-deserto 57 0,55 boa
10 tundra 62 0,67 boa
11 deserto 70 0,74 muito boa
média global 62 0,58 boa
literatura ~ 40 0,40 - 0,50 regular
Oyama and Nobre, 2004
Global Mean GoodVery Good
Good
Good
Good
Good
Good
Very Good
Regular
Poor
Regular
Regular
agreement
Tropical Forest
Temperate Forest
Mixed Forest
Boreal Forest
Larch
Savannas
Grasslands
Dry shrubland
Semi-desert
Desert
Literature
Improvements of the Biome Model under development:
• Add a simple vegetation carbon balance equation for biomes
• Add one more parameter to account for the effect of lightning-caused fire on Savanna/Tropical Dry forest separation
Motivation: Current potential vegetation models tend to represent some regions of dry forests as savannas
One reason for these differences can be the occurrence of fires
Considering fires in potential vegetation estimates
For example: in India and Southeast Asia
Considering fires in potential vegetation modeling
FiresAffect trees
Favor grasses
For similar climate conditions, fire occurrence may interfere in the establishment of mixed forests and favor savannas
Savanna instead of dry forest
Long-term fire occurrence in savannas may be explained by lightning
Lightning activity in the transition between dry and rain seasons is an important cause for fires in savannas in Brazil (Ramos-Neto e Pivelo 2000, Environ. Manag. )
In Amazonia, most of the lightning activity was detected during the transitional months of Sep-Nov (Christian et al. 2003, JGR)
Considering fires in potential vegetation modeling
In a first approximation, broad-scale lightning activity may be parameterized from patterns of wind direction
- “Petersen et al. (2002, J.Clim.) cite that low-level easterly wind flow is the most important factor modulating lightning activity over the Amazon Basin” (Christian et al. 2003, JGR)
- “During easterly regimes both relative increases in CAPE and convective forcing favor more lightning” (Petersen et al. (2002, J.Clim.)
For example, based on previous studies for Amazonia:
Map of total lightning activity (cloud-cloud and cloud-ground) detected using the TRMM-LIS, September to November 2001
Do multiple equilibrium climate-vegetation states exist in nature?
Figure 3 External conditions affect the resilience of multi-stable ecosystems to perturbation. The bottom plane shows the equilibrium curve as in Fig. 2. The stability landscapes depict the equilibria and their basins of attraction at five different conditions. Stable equilibria correspond to valleys; the unstable middle section of the folded equilibrium curve corresponds to a hill. If the size of the attraction basin is small, resilience is small and even a moderate perturbation may bring the system into the alternative basin of attraction. Shaffer et al, 2001. Nature
Biome-Climate Bi-Stability for the SahelBiome-Climate Bi-Stability for the Sahel
Current State Second State
SCHEFFER EL AL., NATURE | VOL 413 | 11 OCTOBER 2001
The second equilibriun state depend mostly on vegetation(albedo) feedback and secondarily on ocean feedbacks
Nobre et al. 1991, J. Climate
Modeling Deforestation and Biogeography in AmazoniaModeling Deforestation and Biogeography in Amazonia
Current Biomes Post-deforestation
“1” Tropical Forest“6” Savanna
Fig. 1 Regions of the Amazon basin that can potentially be converted to savanna after some deforestation. Black regions represent regions in the Amazon basin with tropical forest and having d.s. precipitation > 100 mm. Dark grey regions represent regions having tropical forest with d.s. precipitation ≤ 100 mm. Thisregion could potentially be converted to savanna, given enough deforestation. Light grey regions represent other types of vegetation but mainly savannas having precipitation during the dry season ≤ 100 mm. The dry season precipitation isoline was derived from Nix (1983).
Sternberg, 2001, Global Ecology & Biogeography, 10, 369–378
Searching for Multiple
Biome-Climate Equilibria
Climate Equilibrium States
Oyama, 2002
Vegetation = f (climate)
Climate = f (vegetation)
Vegetation = f1 (g0, g5, Tc, h, s)
Climate = f2 (AGCM coupled to vegetated land surface scheme)
• Climate-Vegetation Equilibrium States only when f1 = f2
• Since f1 and f2 are non linear functions, the possibility exists for multiple equilibrium states
• Such points can only be found numerically. If the system is well behaved, we should find at least one equilibrium state corresponding to the present day biome-climate state.
38 vertical levels
1ºlat 1º
long
200 km
200 km
CPTEC-INPE AGCMComputational code (hundreds of thousands of computer code), which represents numerical aproximations of mathematical equations. These equations represent the Laws of Physics that govern atmospheric motions and their interactions with the surface.
lateral interactions
Interactions with surface
Interactions among layers
Number of elementary volumes for calculation:
400 x 200 x 28= 2,24 milhõesE-W N-S Vertical
Claculation for each elementary volume:Temperature, humidity, wind speed and velocity geopotential height.
Number of elementary volumes for calculation:
400 x 200 x 28= 2,24 milhõesE-W N-S Vertical
Claculation for each elementary volume:Temperature, humidity, wind speed and velocity geopotential height.
Global Domain
www.cptec.inpe.br
• Systematic errors in model-calculated rainfall would result in wrong specification of vegetation.
• Need some kind of correction for that (analogous to ‘flux correction’ in coupled O-A models)
• The correction is to add model calculated anomaly fields of temperature and rainfall to the observed climatology of those quantities, and, then, calculate the new equilibrium biomes.
• That design implies that we can only search for stable equilibrium states that are not too far from present conditions!
How to find numerically Multiple Vegetation-Climate Equilibrium States?
Oyama, 2002
Results of CPTEC-DBM for two different Initial Conditons: all land areas covered by
desert (a) and forest (b)
Oyama, 2002
Biome-climate equilibrium solution with IC as forest (a) is similar to current natural vegetation (c); when the IC is desert (b), the final equilibrium solution is different for Tropical South America
a
b
c
Initial Conditions
Oyama and Nobre, 2003
Two Biome-Climate Equilibrium States found for South America!
Soil Moisture
Rainfallanomalies
-- current state (a)-- second state (b)
Testing the robustness of these results with sensitivity analysis of AGCM to changes in land cover in Northeast
Brazil (desertification) and Amazonia (deforestation, “savannazation”)
Fig. 2 - Vegetation maps for the control (a) and desertification (b) runs.
Oyama and Nobre, 2004
Sensitivity experiment on “Desertification” in Northeast Brazil
Fig. 3 - Annual (a) and wet season (March-May, b) precipitation anomalies. Contour interval is 0.5 in pannel (a), and 1 mm day -1 in (b). Solid (dashed) lines refer to positive (negative) values; zero line is omitted. Dark and light shading refer to high and low statistical significance anomalies, respectively, for the sign test. NEB is enclosed by a thick contour line.
Oyama and Nobre, 2004
(Desertification – Control) Precipitation Anomalies (mm/day)
Annual Wet Season (March-May)
Sensitivity Analysis to ‘Savannazation’ of AmazoniaSensitivity Analysis to ‘Savannazation’ of Amazonia
Resolution: ~ 2ºx2º
Control 2033 All Savanna
2033 All Savanna
JJA 5,4% -21,7%
JJAS 1,9% -21,9%
Dry Season Precipitation*
* 12°S-3°N / 50°W-75°WOliveira et al., 2004
Oyama and Nobre, 2003
Is there a meteorological explanation for the second Equilibrium State?
Soil Moisture
Rainfallanomalies
-- current state (a)-- second state (b)
Unconditional probability of a wet day. a) Threshold of 1 mm, b) Weak rainfall (rainy days: 1 mm - 5 mm) and, c) Moderate rainfall (rainy days: 5 mm - 25 mm). The daily data spans 1979 to 1993.
P > 1 mm 1 mm < P < 5 mm 5 mm < P < 25 mm
Obregon 2001
Unconditional Probability of a Wet Day
Obregon 2001
1 mm < P < 5 mm
SACZ
Sea BreezesInstability lines
Annual Precipitation
Hydraulic Lift: Passive water transport by plants driven by the soil water potential gradient
Hydraulic redistribution increases the efficiency of deep root water transport ability.
• Nocturnal water transport from deep soil layer to upper soil layer by plants.• Increases water availability to the shallow roots.• Nutrient availability increases.• An important water source for neighboring plant growth.• Reverse hydraulic lift is also observed (Burgess et al., 1998): Hydraulic redistribution
+: water flow to the plant-: water flow away from the plant
a
b
bb
c
c
a
Sap-fl
ow
velo
city
Observation from the Amazon
a
rain
Before rain
c
After rain
b
Daytime
R. Oliveira
Oliveira et al., in press
Paleovegetation Reconstructions as Validation for the Second Stable
Equilibrium?
Application of CPTEC-PBM for Past Climate Change
Oyama, 2002
(a) PBM results with uniform cooling of 5 C and drying of 2 mm/day to emulate climate conditions of the LGM (21 ka BP);
(b) vegetation reconstruction for LGM;
a
b
- 5 C- 2 mm/day
Vegetation feedbacks in Amazonia at the Vegetation feedbacks in Amazonia at the last glacial maximum (21 ka BP)last glacial maximum (21 ka BP)
• GENESIS-IBIS coupled vegetation-climate model• 3 experiments: control, R, RPV• Control: present orbital forcing, 350 ppmv CO2 in both
radiative and physiological routines, modern vegetation cover
• R: 21 ka BP radiation forcing only (orbital forcing, 180 ppmv CO2 radiative forcing), modern vegetation cover
• RPV: 180 ppmv CO2 forcing for physiological routines, dynamic vegetation
Reference: Foley, J. A.; Levis, S.; Costa, M. H., Cramer, W.; Pollard, D. 2000: Incorporating dynamic vegetation cover within global climate models. Ecological Applications, v. 10, n. 6, p. 1620-1632.
Last glacial maximum: GENESIS+IBISThe importance of vegetation feedbacks
Reference: Foley, J. A.; Levis, S.; Costa, M. H., Cramer, W.; Pollard, D. 2000: Incorporating dynamic vegetation cover within global climate models. Ecological Applications, v. 10, n. 6, p. 1620-1632.
R = radiation of 21 ka BP, fixed modern vegetation, [CO2] = 180 ppmvRPV = radiation of 21 ka BP + dynamic vegetation + physiology at [CO2] = 180 ppmv
Decrease of AmazoniaRainfall!
Results for Amazonia
What are the likely biome and species changes in Tropical South America due
to Global Warming?
Consequences of Global Warming Consequences of Global Warming Scenarios for Tree Species Scenarios for Tree Species Abundance for Savanna and Tropical Abundance for Savanna and Tropical Forest (Amazonia)Forest (Amazonia)
Geog
rap
hy
Ecolo
gy
Analysis of Species Redistribution for Climate Change
Points of Occurrence
Algorithm Precipitation
Tem
pera
ture
Ecological Niche Model
Prediction of distribution
Projection of new distribution
with climate change
Projections taking climate change into consideration
Impacts for the 162 species studiedImpacts for the 162 species studied((Biota Neotropica 2003, Thomas et al.,Nature 2004)
• Scenário + 0,5% CO2:
– 18 species extinct– 91 species with reduction > 90%
• Scenário + 1% CO2:
– 56 species extinct – 123 species with reduction > 90%
Projection for 2055: + 2 °C ►24% extinction!
Climate Scenarios from Hadley Center
Prediction of present (1961-90) Prediction of present (1961-90) and future (2055) distribution ofand future (2055) distribution of Acosmium subelegans Acosmium subelegans
Projected area of occurrence by 2055 (1% CO2)
Projected area opf occurrence by 2055 (0,5% CO2)
Prediction of species distribution for Prediction of species distribution for 162 tree species of the Brazilian 162 tree species of the Brazilian Cerrado: present (1961-1990)Cerrado: present (1961-1990)
Miles et al. 2004. The impact of global climate change on tropical forest biodiversity in Amazonia. Global Ecology and Biogeography, (Global Ecol. Biogeogr.)
43% of all 69 species of Angiosperms became non-viable by 2095
Climate Model:HADCM2GSa11% CO2 increase/yr
= SI/2
Can global warming cause a forest Can global warming cause a forest die-back in Amazonia? die-back in Amazonia?
Change in Global Climate in HadCM3LC
Interactive CO2 and Dynamic Vegetation
2090s - 1990s
Cox et al., 2000
Change in Amazon Climate and Hydrology in HadCM3LC
Lat: 15oS - 0oNLon: 70oW - 50oW
Amazon forest die-back!
Cox et al., 2000
Change in Amazon Carbon Balancein HadCM3LC
Lat: 15oS - 0oNLon: 70oW - 50oW
Amazon forest die-back!
Cox et al., 2000
Redistributions of biomes in South Redistributions of biomes in South America as response to global America as response to global climate changeclimate change
Anomalias médias entre 2070 e 2099 de precipitação e temperatura para cada modelo e para a média dos 6 modelos (AVG).
Temperature Anomalies (deg C) for 2070-2099
A2 High GHG Emissions Scenario
Anomalias médias entre 2070 e 2099 de precipitação e temperatura para cada modelo e para a média dos 6 modelos (AVG).
Precipitation Anomalies (mm/day) for 2070-2099
A2 High GHG Emissions Scenario
Geog
rap
hy
Ecolo
gy
Analysis of Biome Redistribution for Climate Change
Points of Occurrence
Algorithm Precipitation
Tem
pera
ture
Ecological Niche Model
Prediction of distribution
Projection of new distribution
with climate change
Projections taking climate change into consideration
Biomas para cada modelo com base nas anomalias médias entre 2070 e 2099; AVG se refere aos biomas obtidos com a média dos 6 modelos.
Projected Biome Distributions for South America for 2071-2099
B2 Low GHG Emissions Scenario
Nobre et al., 2004
Painel à esquerda, topo: as cores em vermelho estão relacionadas à quantidade de modelos que atribuem o mesmo bioma para um dado ponto de grade (por exemplo, a cor azul indica que 3 modelos concordam quanto ao bioma). Nos painéis à direita, representam-se os biomas somente se houver concordância entre um número mínimo de modelos. Do topo para a base, o número mínimo passa de 4 para 6. Ou seja, o bioma em um dado ponto de grade representa o que a “maioria” dos modelos prevê. O mapa de vegetação potencial atual é também colocado para facilitar a análise dos mapas.
Minimum Number of Biomes with Agreement among GCM
4
5
6
Synergistic effects of regional climate change (induced by land cover change) and global climate change (induced by global warming) over
Amazonia
Scenário 1
Global Climate Change warmer drierdrier
Regional Climate Change warmer drierdrier
Scenário 2
Global Climate Change warmer wetter
Regional Climate Change warmer drierdrier
High sensitivity to climate (fire, species diversity, etc.)
Moderate sensitivity to climate
Are biomass burning aerosols over southern Amazonia changing the
charecter of dry season precipitation or delaying the onset of the rainy season?
Fire...
Freitas Longo and Silva Dias, 1996
Biomass burning trajectories
SCAR-B, 1995
Biomass burning smoke covers millions of km2
Hei
ght
Aerosol Concentrations in AmazoniaChanges from very low values of5-12 μg/m³ to very high 500 μg/m³In areas affected by biomass burning
Alta Floresta Aerosol Mass Concentration 1992-2000
0
100
200
300
400
500
600
Mas
s co
nce
ntr
atio
n (
µg/m
³)
Coarse Mode
Fine Mode
Africa compared to Amazonia:•Smaller particle effective radius;•Half the rainfall per lightning.•More population and pollution•More desert dust•Less available moisture•Stronger updrafts ???
=> Less efficient rain processes0
5
10
15
20
25
30
6 8 10 12 14 16 18 20 22 24 26 28 29
AmazonAfrica
Rel
ati
ve f
req
uen
cy
Effective Radius [micron]
The relative frequency of effective cloud radiusfor clouds over the African Congo (red bars)and the Amazon (blue bars) (Danny Rosenfeld, personal communication).
Clouds are more continental over
Central Africa than over Amazonia
Smoking Rain Clouds over the Amazon
Andeae, M. O., et Al. ,Science 303,1337(2004)
Measurement of the Effect of Amazon Smoke on Inhibition of Cloud Formation.
Koren, Ila, et al., Science 303,1342 (2004)
CO
(p
pb
v)
Par
ticl
es #
/cm
3
Andreae et al., Science (2004)
Bio
mas
s B
urni
ng
Pris
tine
For
est
Fig. 4. The evolution of cloud drop diameter distribution (DSD) with height in growing convective clouds, in the four aerosol regimes of (A) blue ocean, 18 October 2002, 11:00 UT (universal time), off the northeast Brazilian coast (4S 38W); (B) green ocean, 5 October 2002 20:00 UT, in the clean air at the western tip of the Amazon (6S 73W); (C) smoky clouds in Rondonia, 4 October 2002, 15:00 UT (10S 62W); and (D) pyro-clouds, composite where clouds at height 4000 m are from 1 October, 19:00 UT (10S 56W), and clouds above 4000 m are from 4 October, 19:00 UT (10S 67W). The lowest DSD in each plot represents conditions at cloud base, except in (D), where a size distribution for large ash particles outside of the cloud is also shown. Note the narrowing of CDSD and the slowing of its rate of broadening with height for the progressively more aerosol-rich regimes from (A) to (D).
Cloud Drop Diameter Cloud Drop Diameter in four aerosol in four aerosol
regimesregimesAndreae et al., Science (2004)
Rain gauge stations with daily data used in this ongoing preliminary analysis
Stations Begin End
Cuiabá 01/01/1961 31/12/1996
Itiquira 01/01/1966 31/12/2002
Jaraguá 01/01/1964 31/12/2002
E. do Sul 01/01/1944 31/12/2002
Hipothesis to be tested: if aerosol effect is reducing dry season rainfall in the southern boundaries of Amazonia, one should see differences in August-September rainfall (intensity and/or frequency) by comparing a period with little biomass burning (1960-70) to a period with high biomass burning (1980-90).
Daily rainfall distribution. Red line: Smooth daily rainfall by two firts harmonics (Cum. Var. of them are placed at right top of each figure).Blue line: Represent the limits of the rainy seasnon (number in red). Tthe rainy season is defined to begin when rainfall exceeds its annual daily average.
Cumulative of mean daily rainfall.
Monthly Total Precipitation: August and September
Red line: Smoothed by LOWES.
August September
Smoothed curve (red) shows mostly decadal variability, but no decreasing trend
AUGUST
SEPTEMBER
Frequency of rainfall days for 5-year intervals.
INCONDITIONAL PROBABILITY OF RAINFALL
(EVERY 5 YEARS)
There is no clear observed trend in number of rainy days!
FREQUENCY OF RAINFALL INTENSITY
(Quality)
Frequency of daily rainfall intensity for 3 classes of intensity:
WEAK <5 mm
MODERATE 5-25 mm
INTENSE >25 mm
August
September
Again, there is no observed decrease of daily rainfall intensity with time!
CUMULATIVE RAINFALL DISTRIBUTION
Cumulative rainfall distribution for 5-day (pentad) total rainfall for 5-year intervals (colors).
There are no trends of pentadal rainfall being larger for the earlier period!
Correlation Coeficientes between the anomalies of rainfall and indexes of atmospheric circulation.
AOO IOS N3.4 PDO TSA
Cuiabá -- -.29 -- -- --
Itiquira -- -.42** .34* .27 --
Jaragua -- -- -- -- --
Estrela do Sul -- -- -- .27*
SEPTEMBER
AOO IOS N3.4 PDO TSA
Cuiabá -- -- -- -- -.33*
Itiquira -- -- -- -- -.32*
Jaragua .35* -- -- .28 --
Estrela do Sul .22 -- -- .38 ** --
**99 % of statistical significance * 95 % of statitstical significance
AUGUST
Antarctic oscillation (AOO)
Pacific Decadal Oscillation (PDO)
SST of South Atlantic (TSA)
SOI: Southern Oscillation Index
N3.4: SST anomaly for region Nino 3.4
Rainfall in the study area does not show, on cursory examination, any clear or consistent relationship to large scale atmospheric or oceanic circulation patterns!
Conclusions I
• This preliminary analysis indicates that interannual and decadal variability dominate the variability of dry, biomass burning season rainfall over Southern Amazonia.
• There are no observational evidence of any significat effect that could be attributed to biomass burning aerosol on daily raifall intensity or frequency.
• If such aerosol effect exists, it is likely to be much smaller than natural variability of dry season rainfall
• Further analyses are being carried out to investigate any possible impacts on onset of rainy season
Does climate variability play the key role linking together climate change, edaphic factors, and human use factors?
Resilience Stochastic Perturbations Gradual Perturbations affect Resilience (e.g., deforestation, fire, fragmentation global warming, etc.)
Can large interannual variability of climate (e.g., basin-wide rainfall) tip the balance
to a new state?
Amazon River Discharge ( mAmazon River Discharge ( m33/s)/s)station: Óbidos (01 S, 55 W)station: Óbidos (01 S, 55 W)
Year
mon
th
Large interannual variability in the hydrological cycle in Amazonia
Lon
Lat
-70 -60 -50 -40
-20
-10
010
-70 -60 -50 -40
-20
-10
010
Mean Precipitation (mm)
1050 1250 1450 1650 1850 2050 2250 2450 2650 2850 3050 3250 3450
Projections of regions converted to savanna for 10 and 25% reduction in precip. Note that this approach captures the effects of rainfall patterns (e.g. sea breeze front).
-25% -10%current
Courtesy: S . Wofsy
Conclusions IIThe future of biome distribution in Amazonia
in face of land cover and climate changes
• Natural ecosystems in Amazonia have been under increasing land use change pressure.
• These large-scale land cover changes could cause warming and a reduction of rainfall by themselves in Amazonia.
• The synergistic combination of regional climate changes caused by global warming and by land cover change over the next several decades could tip the biome-climate state to a new stable equilibrium with ‘savannazation’ of parts of Amazonia (and ‘desertification’ in Northeast Brazil) and catastrophic species losses.