energy, water and phenology controls on the annual...
TRANSCRIPT
ENERGY, WATER AND PHENOLOGY CONTROLS ON THE ANNUAL CARBON AND WATER CYCLES USING REMOTE SENSING TO UNDERSTAND CLIMATE VARIABILITY NATURE GEOSCIENCE
Julia Green– Columbia University, Pierre Gentine– Columbia University, Joe Berry– Carnegie Institute, Jung-Eun Lee– Brown University, Jana Kolassa– Columbia University
2
Introduction and Motivation
¨ Discrepancies exist between GCM results and observations ¨ Certain processes in the carbon cycle are not well understood
L.G.G. de Gonc alves et al. / Agricultural and Forest Meteorology 182– 183 (2013) 111– 127 113
Fig. 1. Local-scale (A) evapotranspiration (ET) and (B) GPP as originally modeled by IBIS (brown line, Botta et al., 2002) and NCAR GCM + Community Land Model (blue line,Lee et al., 2005), and as observed (±SD across years 2002-2004, shaded areas) from eddy tower in Tapajos National Forest (km 67 site). MODIS EVI (average 2000–2004, blacksquares) is plotted with GPP in (B). Models show dry-season declines, in contrast to observations from both satellite and eddy towers. (C) GPP and EVI in a pasture/agriculturalarea (km 77 site) that has opposite seasonality from nearby (12 km distant) forest site in (B). (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of the article.)
The experiments consisted of uncoupled land surface modelsimulations forced by standardized atmospheric variables mea-sured at eight sites across the Amazon region as shown in Fig. 2,where each site represents a different biome or a variant of it result-ing from disturbance. Section 2 describes the eight sites used in theproject. Section 3 describes the gap-filled and quality controlleddata, which span eight years of observations. Section 4 presents adetailed description of the 23 different models and their variantsthat participated in the LBA-DMIP experiments. A detailed com-mon protocol for driving models and reporting simulation resultshas been produced and its full version is given in Appendix A.
2. Site descriptions
The sites include four evergreen broadleaf forests, a deciduousbroadleaf forest, a savanna site, and two pastures. Seven of the eight
sites are in Brazilian Amazonia, while a savanna site in the state ofSão Paulo was also included (Fig. 2).
The evergreen broadleaf forests sites are in the Reserva Biológ-ica do Cuieiras (Cueiras Biological Reserve) near Manaus, Amazonas(the K34 tower); the Floresta Nacional do Tapajós (Tapajós NationalForest) near Santarém, Pará (the K67 and K83 towers); the ReservaJarú (Jarú Reserve) (the RJA tower) located near Ji-Paraná, Rondô-nia. Manaus K34 is the most western of the central Amazoniansites and is located 60 km north of the city of Manaus (Araújoet al., 2002); it has a dry season from July through September.The Santarém (K67 and K83) forest experimental sites are nearthe confluence of the Tapajós and Amazon rivers (Hutyra et al.,2007; Miller et al., 2004; Saleska et al., 2003). K67 is in an undis-turbed primary forest, but at the K83 site about 15% of the treeswith diameter at breast height greater than 35 cm were selectivelylogged over a 700-ha area during three months starting September2001 (Figueira et al., 2008; Miller et al., 2007). The Jaru Biological
Fig. 2. Flux sites spatial distribution across different ecosystems of Amazonia.
Gonçalves, L., et al. "Overview of the Large-Scale Biosphere–Atmosphere Experiment in Amazonia Data Model Intercomparison Project (LBA-DMIP)" (2013). NASA Publications. Paper 133. http://digitalcommons.unl.edu/nasapub/133
Introduction Cont.
¨ Potential options for improvement: In-situ point leaf level measurements Flux tower measurements (canopy/ecosystem
scale)
3
M. Jung, M. Reichstein, A. Bondeau, Towards global empirical upscaling of FLUXNET eddy covariance observations: Validation of a model tree ensemble approach using a biosphere model. Biogeosciences 6,2001 (2009).
REMOTE SENSING!!!
Research Goals
¨ To improve our understanding of the variability of land and atmospheric variables related to the carbon and water cycles Temporally (interannual, seasonal) Spatially (climatic conditions, ecosystems)
¨ To define spatially the control on the annual carbon and water cycles Energy Water Phenology
4
Importance
¨ Advance our understanding of how vegetation responds to increases in atmospheric CO2
¨ Show us the effect of water stress on the CO2 cycle
¨ Improve the performance of General Circulation, and Land Surface, and Vegetation Models
¨ Allow us to more accurately make climate change predictions and weather forecasts
5
Remote Sensing Datasets 6
Parameter Source
Net Radiation Clouds and the Earth's Radiant Energy System (CERES)
Precipitation Global Precipitation Climatology Project (GPCP)
Solar Induced Fluorescence (SIF)
Global Ozone Monitoring Experiment–2 (GOME-2)
EVI Moderate Resolution Imaging Spectroradiometer (MODIS)/ Multiangle
Temperature GHCN_CAMS Gridded 2m Temperature
Solar Induced Fluorescence (SIF) 7
¨ During photosynthesis a plant absorbs energy through its chlorophyll % used for ecosystem gross primary production (GPP) % lost as heat % re-emitted (SIF)
¨ Relationship between GPP and SIF is linear
Guanter, L., et al. " Global monitoring of terrestrial sun-induced chlorophyll fluorescence from space." (2013). International Conference: Towards a Global Carbon Observing System:Progresses and Challenges.
Climate Regimes
Mediterranean Climate
8
Tropical Climate Mid-Latitude Climate Monsoonal Climate
Mediterranean Climate 9
¨ In Mediterranean climates radiation and precipitation are out of phase. Limited by light in winter and water in summer. Interannual variability in GPP is due to the variability in precip.
Monsoonal Climate 10
¨ Monsoonal climates have peak in radiation that drives the precip., which then drives the GPP.
¨ Variability in GPP due to interannual variability in precip.
Tropical Climate 11
¨ Tropical climates average annual cycles vary greatly between regions
Mid-Latitude Climate 12
¨ Mid-latitude climates have radiation, GPP and EVI in phase(GPP peaks slightly before radiation and decreases at conclusion of phenological cycle). Largest interannual variability in precip.
Correlation between Radiation and SIF
¨ Less strong in tropical and desert regions than in mid-latitudes where radiation is driving GPP.
13
Correlation between Temperature and SIF
14
¨ Similar to radiation but less highly correlated– opposite in monsoonal regions (temperature drops during monsoon season) and transitional zones.
Correlation between EVI and SIF
¨ Greenness typically has high correlations with SIF– but not in the very wet tropical regions (EVI is constant year round) and some desert regions (SIF is very minimal)
15
Correlation between Precipitation and SIF
16
¨ Precipitation highly correlated with SIF in transition zones. Regions with the most rain have lower correlation due to the cloud coverage and constant EVI
Combining Correlations to RGB Plot
R G B
Corr(SIF, EVI)
Corr(SIF, Net Radiation)
Corr(SIF, Precipitation)
Controls on Carbon and Water Cycles
18
Corr(SIF, Net Radiation) Corr(SIF, EVI) Corr(SIF, Precipitation)
Conclusions
¨ Globally defined each climatic regime in terms of GPP as light, water or phenology controlled.
¨ Learned wet tropical forests behave differently (eg. Amazon vs. Congo Rainforest) in the following ways: The Amazon is more light limited than the Congo
Rainforest In the Amazon less precipitation (to a point) is
beneficial to photosynthesis ¨ Changes in the water cycle will therefore affect
distinct regions differently
19