AN 1100YEAR RECONSTRUCTION

1

• AN 1100-YEAR RECONSTRUCTION
• OF
• WINTER PRECIPITATION
• FROM
• JUNIPER GROWTH
• OF
• NORTHEASTERN QINGHAI PROVINCE,
• CHINA

2
Justification

• Past climatic to understand current, future
• China (Plateau) affect, reflect regional to
  sub-hemispheric climate
• Short instrumental records, begin in 1950s
• Some dendroclimatology in the area, but could be
  much more

3
(No Transcript)
4
(No Transcript)
5
Usual Lab Procedures

• Crossdated, two different people
• Measured widths
• COFECHA checks and data QC
• ARSTAN chronology building
• Conservative detrending
• Biweight robust mean estimation

6

• Notice dips in sample depth at moments of low
  growth

7
Response Functions or Correlations

• Blasing et al. (1984) T-RB 441.
• Response function could lead to false
  conclusions about dendroclimatic relationships
• publish correlation function instead of or in
  addition to a response function

8
Response Functions or Correlations

• Fritts and Wu (1986) T-RB 4631.
• little justification in using correlation
  function due to multicollinearities
• The purpose of RF is to describe, not to predict
  or reconstruct, a dendroclimatic relationship

9
Response Functions or Correlations

• Example RF Huachuca Mts.
• Width f(pMay,-pJun,cAug)tmpf(pMay,-pJun,pS
  ep-cFeb,cJul)pptf(previous year growth)
• Total R2 59

10
Response Functions or Correlations

• 11 predictors of current tree growth
• Could be useful in modern environmental study
• Factor out climate effects
• Evaluate residuals for other factor
• For dendroclimatic reconstruction
• Must switch the matrix, but to what?

11
What We Did

• Evaluate correlations of dendrochronology with
  seasons of climate (tmp and ppt) that might
  reasonably control tree growth
• Started with pJul through cDec
• Tried various 1-month, 2-month, etc. seasons
• Rank correlations to find pattern of seasons

12
(No Transcript)
13
(No Transcript)
14
Preliminary Conclusion

• Juniper growth here responds to growing season
  moisture availability
• Antecedent winter precipitation probably most
  important, directly
• Current temperature also important, indirectly by
  evapotranspiration
• Single water balance variable not strong
• Chose winter-spring (Dec.-Jun.) precipitation

15

• Log transformed tree-ring data
• Tree-ring series used at year t and t-1
• AR(1) of residuals 0.20

16

• Modern period not extraordinary in mean condition
• Tree record in sync with historical documents
• Change of phase with Dunde Ice record
• Modern period not extraordinary in 21-yr
  variability

17

• Skew
• Almost always some to left
• Kurtosis
• Occasional periods of leptokurtosis (peakedness)
• Clearly related

18
(No Transcript)
19
Conclusions

• Excellent tree-rings in western China
• Multi-proxy opportunities great
• Moisture signal
• Perhaps not monsoonal
• Agricultural significance not as high as
  elsewhere
• So far, 20th C not extraordinary
• Interesting ppt-tmp relationships