Explore temporal variability of the summer droughts in the SE US and its SST forcing Rong Fu, Wenhong Li 2, Lei Huang Jackson School of Geosciences, The University of Texas at Austin 2 EAS/Georgia Tech In collaboration with Kingtse Mo at NCEP NOAA CPPA PI Meeting, September 29-October, 2008
Two droughts of the century since 2000: In Oct. 2007: there was less than 3 months of water left in the Atlanta city reservoir. (USA Today). Agriculture lost due to drought: $.3 billion in SE US during 2007 (National Drought Mitigation Center). Normalized May July mean precipitation anomalies (by σ) averaged over the Southeast United States (25N 36.5N, 76 9W)
Two measurements of drought in SE US used in this study: Early Summer Drought index: The Standardized Precipitation Index (SPI, 6-month) averaged for May- July season for the spatial domain of 25 36.5 N, 76 9 W). SPI is influenced by integrated rainfall anomalies from winter to summer. Early Summer Precipitation Anomaly Index (PI): Mean precipitation anomalies in the SE US for May-July.
Droughts intensity and summer rainfall anomalies in the SE US appear to increase since late 970s: How do external SST changes and internal land surface feedback contribute to this increase in magnitude of summer rainfall anomalies and drought intensity in the SE US? SPI value summer mean SPI value CPC Summer 6 mon-spi (MJJ) 3 2 0 950-955 960 965 970 975 980 985 990 995 2000 2005-2 -3 year droughts Summer rainfall anomalies (MJJ) Data Source: CPC-Chen data
Previous works: Fewer studies on droughts in the SE US: Stahle & Cleaveland 992: decadelong spring rainfall extremes have been a prominent feature of SE US over the past 000 yrs. Mo & Schemn 2008: Seasonally varying SST forcing, e.g., La Nina leads to positive P in winter but negative P in summer the SE US; Seager et al. 2008: weak connection with SSTA, thus low predictability
However, We can learn from many works on droughts over the great plain: Tropical and extratropical Pacific influence: Namias 955,82; Chang and Wallace 987; Trenberth et al. 988; Mo et a. 997; Ting and Wang 997; Livezey and Smith 999; Barlow 200: Combined tropical Pacific and Atlantic influences, esp. for summer droughts: Hu & Feng 200a&b; Schubert et al. 2004a, b, 2007 Soil moisture feedbacks: Namias 99; Findell and Eltahir 997; Koster et al. 2000; Schubert et al. 2004, 07
Influence of Tropical Pacific and Atlantic SSTA as represented by the idealized SSTA runs by the US Drought Working Group: Linear Trend Pattern (LT) Pacific Pattern (Pac) Atlantic Pattern (Atl) Source: Schubert SSTA forcing based on leading EOFs of annual mean SST - 90-2004)
Modeled SSTA influence on summer drought in the SE US: Models disagree on the combination of the tropical Pacific and Atlantic SST forcings. GFDL and NSIPP: warmer Atlantic and/or warmer Pacific CAM3.0: Warmer Pacific and colder Atlantic 25 20 5 0 GFDL AM2.: 0~- -~-2-2~-3-3~-4 25 20 5 0 5 NSIPP: 0~- -~-2-2~-3-3~-4 PcAw PwAc PwACLw UW 5 0-0.5-0.6-0.7-0.8-0.9 - mean Clim--- Clim--- cpca--- cpna--- cpw A--- npca--- cpca--- cpna--- cpwa--- npca--- npwa--- wpca--- wpna--- wpwa--- npwa--- wpca--- wpna--- wpwa--- nltcpwa nltwpca plt-exp nltcpwa Tw PcA plt-exp pltcpwa pltwpca unif-ex pl TcPwA pltwpca un if-e x ctpwtanlt-exp wtpcta- wtpwta- ctpwtanlt-exp wtpcta- wtpwta- 0-0.6-0.7-0.8-0.9 warm Pac cold Pac warm ATL cold ATL cntrl AMIP - 0 2 3 4 5 6 7 ave negative SPI 35 30 25 20 5 0 5 0 0~- -~-2-2~-3-3~-4 wp wa runs npwa wpwa nltwpca plt-exp PnAn PcAc PcAn PcAw PnAc PnAw PwAc PwAn PwAw cwlc cwlw nnlc nnlw wclc wclw Uw AMIP PnAn PcAc PcAn PcAw PnAc PnAw PwAc PwAn PwAw cwlc cwlw nnlc nnlw wclc wclw Uw AMIP -0.6-0.65-0.7-0.75-0.8-0.85-0.9-0.95 CAM3.0
Question: How well can global climate models simulate the variability of the early summer precipitation anomalies over the SE SU and its relationship with the SSTA of the Pacific and Atlantic?
Issue: The relative roles between the Pacific and Atlantic SSTA may vary on decadal scales: Hu & Peng 200a&b: Relative influence of the two completing factors, ENSO and southerly flow, on drought in the great plain varies at multi-decadal scales. Giannini et al. 200: NAO influences on the inter-decadal change of ENSO influences on Caribbean region. Approach: Wavelet and joint wavelet analysis (Torrence & Webster 999; Grinsted et al. 2004)
Model-data Comparison of the SPI (979-2000): CAM3.5 AMIP Models: CAM3.0 AMIP Observations: CPC-Chen -2yr NSIPP AMIP GFDL AM2. AMIP GPCP 3-5yr
Model-data Comparison of the summer PI (979-2000): Models: Observations: CAM3.5 AMIP CAM3.0 AMIP CPC-Chen -2yr NSIPP AMIP GFDL AM2. AMIP 3-5yr
What force the observed biannual variability? 5% against red noise SPI (6-8N, 20-60W) SE SU: 25-36.5N, 76-9W Both N. Atlantic SSTA and Nino 34 have significant bi-annual variations during late 980s and early 990s, coincide with spectrum peak of SPI and PI. What are the relative roles of biannual variations of Nino 34 and NAtl SSTA in determine the variabilities of SPI and PI? Ninn34:5S-5N, 70-20W N. Atl: 6-85N, 20-60W)
Coherence between PI-Nino, and PI-NAtl anomalies in models and observations-: GFDL AM2. AMIP: PI-Nino34: Observed PI-N. Atl: Too strong coherence on 3-5 yr scales, too weak biannual ENSO influences; Biannual Atlantic influence generally agree with observations. GFDL AM2., AMIP 0 o : in phase 90 o : Nino or NAtl Leads PI
Coherence between PI-Nino, and PI-NAtl anomalies in models and observations-2: NSIPP, AMIP: PI-Nino34: Observed PI-N. Atl: Better biannual ENSO influences, but missing biannual N. Atl. influence; Too strong coherence on 3-5 yr scales, NSIPP, AMIP 0 o : in phase 90 o : Nino or NAtl Leads PI
Coherence between PI-Nino, and PI-NAtl anomalies in models and observations-3: CAM3.5, AMIP PI-Nino34: Observed PI-N. Atl: Too strong biannual ENSO influences at wrong time, weaker biannual N. Atl. influence; Too strong coherence on 3-5 yr scales. CAM3.5, AMIP 0 o : in phase 90 o : Nino or NAtl Leads PI
Coherence between PI-Nino, and PI-NAtl anomalies in models and observations: PI-Nino34: Observed PI-N. Atl: CCM3.0, AMIP No biannual influence; Too strong coherence on 3-5 yr scales. CCM3.0, AMIP 0 o : in phase 90 o : Nino or NAtl Leads PI
Summary: For the 20-yr period of our analysis, droughts (SPI and PI) in SE US appear to be dominated by biannual variability, whereas droughts in the global climate model (we examined) are dominated by 4-yr variability. Biannual ENSO mode appears to contribute to the observed drought variability, whereas in models, 4-yr ENSO and perhaps N. tropical Atlantic modes appear to dominate the drought variability. Needs to exploring the dynamic configurations preferred by the biannual variability. Longer AMIP runs are needed for examine decadal variations of the drought!
How does Nino34 and N. Atlantic SSTA influence the circulation anomalies? Observations: N. Atlantic SSTA appears to project stronger influence on Z 200mb anomalies in the E. US and Caribbean Sea, whereas Nino 34 appears to project stronger influences on Z 200mb anomalies over the W. US. Regression δz 200mb -Nino 34 Regression δz 200mb -NAtl-SSTA NCEP, 979-2000
How does Nino34 and N. Atlantic SSTA influence the circulation anomalies? Regression δz 200mb -Nino 34 δz 200mb -NAtl-SSTA Regression δz 200mb -Nino 34 δz 200mb -NAtl-SSTA NCEP GFDL AM2.0 CAM3.5