Preliminary analysis of Clivar-Drought simulations from GFDL

1
Preliminary analysis of Clivar-Drought
simulations from GFDL

• Model AM2.1 (atmospheric component of CM2.1)
  with various prescribed SST scenarios
• Goal To identify regions with drought-related
  sensitivities to the imposed SST forcing patterns

2
Big picture precip differences
3
Methodology

• Calculate grid-specific soil moisture (precip,
  runoff) deciles for the control run
• Values can be 1-10
• Use those decile bounds to calculate deciles for
  each of the experiments relative to the control
  run
• Values can be 0-11
• 0 means smaller than any value observed during
  the control run in that grid cell
• 11 means larger than any value observed during
  the control run in that grid cell

4
Control run

• Time series and histogram of regional average
  deciles
• To look at frequency of extremes, histograms of
  frequency of x consecutive months below or above
  given thresholds

5
Experiment compared with control

• Time series and histogram of regional average
  deciles
• Compare significance of mean difference using
  modified t-test
• Compare significance of differences of histogram
  probabilities using a chi-squared goodness-of-fit
  test

6
Experiment compared with control

• Frequency of extremes in regional average deciles
  
• Compare significance of differences between the
  distributions for the extremes using a
  Wilcoxon/Mann-Whitney rank-sum test
• p lt 0.05 means the median of the experiments
  dist. is sig. diff. from the controls at 95
  confidence level

7
Orinoco river region soil mois deciles
8
Control run drought area

• Can also look at the percent of the region
  experiencing drought (lt2nd decile) at any one
  time
• Or the number of consecutive months with greater
  than 40 (or some other threshold value) of the
  grid cells in drought
• Conversely, pluvial periods are characterized by
  small percentage of grid cells in drought
  plotting number of consecutive months with
  greater than 10 of the grid cells in drought

9
Orinoco river region drought area
10
(No Transcript)
11
Orinoco River Region Summary

• Pacific forcing significantly
• reduces mean rainfall by 1.9 deciles, mean soil
  moisture by 2.5 deciles, and mean runoff by 3
  deciles
• increases drought frequency, duration, and area
• North Atlantic forcing significantly
• increases mean rainfall by 0.9 deciles, mean soil
  moisture by 1 decile, and mean runoff by 1.5
  deciles
• increases frequency and duration of very wet
  periods
• Trend forcing no significant impact
• Pac/NAtl combinations strongest impact when the
  anomalies are of opposite sign, particularly
  Pac/-NAtl

12
U.S. Southwest soil moisture deciles
13
Central U.S. soil moisture deciles
14
Moving forward

• Other regions and other simulations are also
  being studied
• Other drought measures and statistics are being
  added
• I welcome thoughts and contributions!