580 likes | 742 Vues
Relationships Between Patterns of Atmospheric Circulation and U.S. Drought over the Past Several Centuries. Zhang, Zhihua Department of Environmental Sciences University of Virginia. Committee: Professor Michael Mann ( adviser ), Department of Environmental Sciences
E N D
Relationships Between Patterns of Atmospheric Circulation and U.S. Drought over the Past Several Centuries Zhang, Zhihua Department of Environmental Sciences University of Virginia Committee: Professor Michael Mann (adviser), Department of Environmental Sciences Professor Jose Fuentes, Department of Environmental Sciences Professor Bruce Hayden, Department of Environmental Sciences Professor Henry Shugart, Department of Environmental Sciences Professor Ted Chang, Department of Statistics
“And it never failed that during the dry years the people forgot about the rich years, and during the wet years they lost all memory of the dry years. It was always that way.” —John Steinbeck East of Eden
Is it going to be dry or wet this year? We need to understand the past history of drought to better assess future prospects for drought.
The goal of my research is to address such questions as: • In what ways do the temporal and spatial patterns of US drought change over time? • To what degree are those drought patterns linked with larger-scale atmospheric circulation changes? • What is the relative importance of climate variability in various regions of the tropics and extratropics in determining patterns of conterminous U.S. drought?
OUTLINE To place modern climate changes in a longer-term context and explore the fuller range of potential variability, I have: • Extended the drought record father back in time with dendroclimatic reconstructions of summer drought (PDSI) patterns over the conterminous U.S back to 1700 • Extended the atmospheric circulation record back in time through proxy-based reconstructions of boreal cold- and warm-season global SLP patterns back through the 17th century
OUTLINE To more fully assess the potential relationships between U.S. drought and larger-scale influences by atmospheric circulation patterns and dynamical modes of climate variability, I have 3. Analyzed the evidence for coherent modes of variability in the joint U.S. drought/seasonal SLP field over the modern instrumental period 4. Investigated the longer-term relationship between U.S. summer drought and atmospheric circulation anomaly, making use of proxy-based pre-reconstructions of past centuries
Reconstructions of U.S. summer (JJA) drought (PDSI) patterns back to 1700
U.S. drought reconstructions Proxy network: 483 tree ring chronologies
U.S. drought reconstructions PDSI grids Rocky Mts Appalachian Mts This grid spacing is 2º lat. × 3º long.
U.S. drought reconstructions Method (RegEM): • The method is based on a regularized expectation maximization algorithm (RegEM), which offers some theoretical advantages over previous methods of CFR. • This approach calibrates the proxy data set against the instrumental record by treating the reconstruction as initially missing data in the combined proxy/instrumental data matrix. • With optimally estimating the mean and covariance of the combined data matrix through an iterative procedure, RegEM can produce a reconstruction of climate field with minimal error variance (Schneider, T., 2001; Rutherford et al, 2003; Mann et al, 2002).
RegEM CFR approach Mann, M.E., Rutherford, S., Wahl, E., Ammann, C., Testing the Fidelity of Methods Used in Proxy-Based Reconstructions of Past Climate, Journal of Climate, 18, 4097-4107, 2005.
U.S. drought reconstructions Matrix of dataset 1978yr • To calculate the reconstruction scores, we only used part of the available instrumental data for calibration (1928-1978) and keep some instrumental data (1895-1927) free for verification. • For final reconstruction, we employed all available instrumental data. • Code was from • http://www.math.nyn.edu/~tapio/imputation/. PDSI dataset 1927yr 1895yr missing data need to be recon. Present years past years Proxy dataset 1700yr PDSI gridpoints Tree-ring chronologies
RE d i s t r i b u t i o n f o r v e r i f i c a t i o n i n t e r v a l ( g l o b a l p r o x y d a t a r e c o n . r e g U.S. drought reconstructions i o n a l P D S I ) 5 0 1 . 0 0 3 . . 0 0 3 . 0 0 . 1 0 . 6 5 3 0 0 5 1 0 0 . . 0 . 1 5 5 1 4 4 5 . 5 0 . 4 0 . 5 0 0 3 0 . 4 5 . 0 0 0 0 0 0 3 . 6 0 . . 3 0 0 . 5 3 0 0 . 4 - 5 0 . 1 . 0 0 0 . 0 1 5 . 1 5 . 0 0 0 1 5 6 3 3 . . 5 . 0 0 0 1 0 . . 0 0 4 0 5 4 . . 0 5 4 5 0 0 . . 1 3 0 . 0 6 5 0 0 0 6 . . 0 3 0 . 0 . 1 5 0 6 0 0 3 . 0 5 0 Mean=.3614 4 . 0 0 . 0
Time series of regional and domain mean drought back to 1700 1930’s Dust Bowl RegEM Cook et al.
The spatial patterns of reconstructed U.S. drought based on RegEM 1736 1708 1864 1800
The spatial patterns of reconstructed U.S. drought based on RegEM 1745 1726 1833 1793
Reconstructions of cold-season (Oct-Mar) and warm-season (Apr-Sep) global SLP patterns back to 1601
Global SLP reconstructions • Hybrid frequency-domain RegEM • Different types of proxy data exhibit fundamentally different frequency-domain fidelity characteristics. • Some variables such as sediments, ice core and historical records are only decadal/low-frequency resolved proxy indicators. • Stepwise RegEM • Proxy data do not share a common length, stepwise procedure can better use climate information in the calibration process. (Rutherford et al, 2005; Mann et al, 2005)
Global SLP reconstructions Spatial distribution of full proxy database (high-frequency) Year (before 2000 AD)
Global SLP reconstructions Spatial distribution of full proxy database (low-frequency) Year (before 2000 AD)
Procedures of reconstructing global SLP Climate Low-frequency band Proxy PCs(dense tree-ring) High-frequency band Full proxies Full proxies (including lag+1,0,-1) Screened proxies (95%) with local climate Reconstructing low-frequency climate Reconstructing high-frequency climate Summing reconstructed low/high-frequency climate
Spatial verification scores Boreal cold-season Boreal warm-season
Verification using long-term European SLP data(Luterbacher et al.,2002) Boreal cold-season Boreal warm-season Nodal area No real data
1982/83 ENSO
Comparison with other reconstructions Mann: 0.41 Stahle: 0.42 Jones: 0.83 Luterbacher: 0.43 Cook: 0.37 Vinther: 0.31
Analysis of Modern Relationship between Patterns of SLP and U.S. Drought (1895-1995)
The MTM-SVD method • The MTM-SVD method [Mann and Park, 1994; 1999] has been widely used in the detection of spatiotemporal oscillatory signals in one or several simultaneous climate data fields. • The MTM-SVD method identifies distinct frequency bands within which there is a pattern of spatially-coherent variance in the data that is greater in amplitude than would be expected under the null hypothesis of spatiotemporal colored noise. • This method differs from conventional EOF-based approaches in that both phase and amplitude information are retained in the data decomposition.
MTM-SVD spectra Cold-season SLP/U.S. summer drought Warm-season SLP/U.S. summer drought ENSO signal ENSO signal Bi-decadal signal 99% sign. 99% sign.
Spatial reconstructions of peak ENSO signal (5-yr) coincident with peak positive ENSO (TNH) extratropical teleconnection pattern (Livezey and Mo 1987)
Spatial reconstructions of peak ENSO signal (5-yr) Cold-season Warm-season
Comparison with standard composites (cold-season) obs. sign. recon.
Comparison with standard composites (warm-season) obs. sign. recon.
Spatial reconstructions of warm-season bidecadal (22 yr) signal coincident with peak domain wet
Spatial reconstructions of warm-season bidecadal (22 yr) signal Time-domain recon. vs. raw data Domain mean Great plains South west Schubert et al. 2004
Analysis of Past Relationship between Patterns of SLP and U.S. Drought with proxy-based data (1700-1870)
MTM-SVD spectra (recon. data) ENSO signal Quasi-decadal signal 99% sign. Weak ENSO Weak ENSO Mann 2000 Bi-decadal signal ENSO signal 99% sign.
Spatial reconstructions of peak ENSO signal (3.5 yr) coincident with peak positive ENSO (TNH) extratropical teleconnection pattern (Livezey and Mo 1987)
Spatial reconstructions of peak ENSO signal (3.5 yr) Cold-season Warm-season
Time-domain reconstructions associated with 3.5 yr period ENSO signal Cold-season Warm-season
Spatial reconstructions of cold-season quasidecadal (11 year) signal coincident with peak domain wet
Spatial reconstructions of cold-season quasidecadal (11 year) signal Time-domain reconstructions Tourre et al. 2001
Spatial reconstructions of warm-season bidecadal (24 year) signal coincident with peak domain wet
Spatial reconstructions of warm-season bidecadal (24 year) signal Time-domain reconstructions Schubert et al. 2004