Changes in Global Land Surface Moisture Conditions During 1950-2004

1
Changes in Global Land Surface Moisture
Conditions During 1950-2004

• Aiguo Dai
• National Center for Atmospheric Research,
  Boulder, CO, USA
• Workshop on Vulnerability of the Carbon Cycle to
  Drought and Fire
• 5-9 June 2006, Canberra, Australia
• Collaborators Taotao Qian and Kevin
  Trenberth

2
Outline

• Observed changes in surface air temperature,
  precipitation, humidity, and soil moisture
• Land surface model-simulated changes in soil
  moisture content and PDSI changes
• Global climate model-simulated changes
• Summary

3
Sfc. T Trend 1950-2005 CRU data
4
EOF 1 of Land Precip is Associated with ENSO
ENSO SST
PC 1
Dai et al. 1997
5
Sahel Rainfall 1948-2004
T
P
Sahel 10o-20oN, 18oW-20oE
6
Eastern (gt135oE) Australian Rainfall 1948-2005
P
Slope ?0.034 mm/day/decade
ENSO
P data PRECL (NCEP) (1948-1996) GPCP
(1997-2005) ENSO Index http//www.cdc.noaa.gov/pe
ople/klaus.wolter/MEI/
7
Trends in Surface Humidity 1976-2004
Dai 2006, J. Climate
8
Surface specific humidity q from Reanalyses
Global (60S-75N) Land
Observations
Northern Hemisphere Land
ERA-40 Reanalysis
NCEP/NCAR Re.
Southern Hemisphere Land
9
Trends in Soil Moisture Sparse Observations
(Robock et al.2000 2005)
10
How have the observed changes in P,T and other
fields affected global soil moisture?

• Because of a lack of observations, we employed
  three different approaches to examine historical
  changes in global soil moisture content
• Analysis of the Palmer Drought Severity Index
  (PDSI) calculated from observed T and P based on
  the Palmer model (a bucket-type model) (see Dai
  et al. 2004)
• Analysis of the soil moisture content simulated
  by a comprehensive land surface model (namely
  CLM3) forced by observation-based atmospheric
  forcing
• Analysis of soil moisture changes in global
  climate models.

11
Palmer Drought Severity Index (PDSI)

• The PDSI is computed using a bucket-type land
  surface model using observed P (Chen et al.02)
  and T (Jones Moberg03)
• It is a normalized measure of the cumulative
  departure in atmospheric moisture supply (P) and
  demand (E) at the surface
• E is based on Thornthwaite (1948)
• PDSI is correlated with observed soil moisture
  content
• Caveats No vegetation, no snow processes, crude
  estimate of E, not always comparable spatially,
  etc.

12
PDSI vs. Observed Soil Moisture Content
(from Dai et al.04, JHM)
Illinois
d
(Soil moisture data from Robock et al.00 and
Hollinger Isard94)
13
PDSI vs. Streamflow
Paraná
Streamflow
PDSI
Amazon
Lena
Congo
Columbia
Orinoco
Changjing
Mississippi
14
CLM3 Simulations 1948-2004

• CLM3 Community Land Model Version 3, a
  comprehensive land surface model designed for
  coupled climate simulations. It simulates most
  land surface processes, including surface fluxes,
  land hydrology, and stomatal physiology and
  photosynthesis.
• The CLM3 was run at T42 (2.8o) with the spatial
  heterogeneity of land surface represented as a
  nested sub-grid hierarchy in which grid cells are
  composed of multiple land units, snow/soil
  columns, and plant functional types. (Oleson et
  al. 2004).
• The CLM3 forcing data combine intra-month
  variations from the NCEP/NCAR or ERA40 reanalysis
  with longer-term variations from observations
  (Qian et al. 2006). Variables T, P, q, V, Ps and
  S?.
• For example Precipitation P (Pmo / Pmr) Pr,
  where Pmo is observed monthly precip. from Chen
  et al.(2002), Pmr is monthly precip. from
  Reanalysis, Pr is 6-hourly precipitation from the
  reanalysis.

15
NCAR CLM3 Structure
Oleson et al. 2004 ( http//www.ccsm.ucar.edu/mode
ls/ccsm3.0/ )
16
CLM3-simulated vs. Observed 1m Soil Moisture
Obs.
CLM
Illinois
r 0.87
Year
Obs.
CLM
E. China
r 0.63
Year
17
River outflow for Worlds 200 Largest Rivers
18
Streamflow CLM3 vs. Obs.
Amazon
Paraná
Obs.
CLM
Lena
Congo
Orinoco
Columbia
Changjing
Mississippi
Yenisey
19
(No Transcript)
20
Trends from 1948-2004
PDSI (change/50yrs)
Soil Water from CLM3 (mm/50yrs)
21
mm/50yr
All Forcing
Trends in CLM3- smulated top 1m soil water from
3 runs
dP only
Red Drying
dT only
22
Evap.
CLM3-simulated E and Soil Water from 3 Runs,
60oS-75oN Averages
All Forcing Run
Precip.-only Run
Temp.-only Run
Soil Water
23
Precipitation, PDSI and CLM3-simualted Soil Water
over Eastern Australia
Precip.
PDSI
Soil Water
24
Global Percentage Dry Areas
Based on PDSI
Based on CLM3
Dry cases Bottom 20 percentiles
25
Sensitivity of the Dry Area to Precip. and Temp.
Changes
26
Total Soil Moisture Simuluated by Coupled
GCMs From IPCC AR4 Historical Runs
Year
27
Soil Moisture Trends in Coupled GCM
Simulations IPCC 20th Century All Forcing Run,
1948-1999, ANN
CCSM3
HadCM3
GFDL CM2.1
GISS
Red Drying
28
Projected Soil Moisture Changes () by Coupled
GCMs IPCC SRES A1B, 2080-2099 minus 2000-2019,
JJA
GFDL
CCSM3
GISS
HadCM3

Red Drying in 2080-2099
29
Summary

• Both the PDSI and CLM3 simulations show a general
  drying trend over global land areas since the
  1970s
• Large warming during recent decades over Eurasia
  and northern N. America enhanced evaporation and
  contributed to the drying in the N. Hemisphere
• Precipitation decreases over Africa, East Asia
  and East Australia are the main cause of the
  drying in these regions
• Both the PDSI and CLM3 simulations show 50 or
  more increases in global dry areas since the
  1970s, with a large jump in the early 1980s due
  to 1982/83 El Niño
• Coupled GCM simulations also suggest a general
  drying over global land areas since the 1970s,
  although the magnitude and regional patterns
  differ among the models and
• Actual trends in soil moisture may be influenced
  by irrigation and other human activities, which
  are not considered here.

30
Visit www.cgd.ucar.edu/cas/adai for related
papers and the PDSI and other data sets.
Related publications Dai, A., K. E. Trenberth,
and T. Qian, 2004 A global data set of Palmer
Drought Severity Index for 1870-2002
Relationship with soil moisture and effects of
surface warming. J. Hydrometeorology,  5,
1117-1130. Qian, T., A. Dai, K. E. Trenberth,
and K. W. Oleson, 2006 Simulation of global land
surface conditions from 1948-2004. Part I
Forcing data and evaluation.J. Hydrometeorol., in
press.
31
All forcing run 60S-75N
32
Effects of T and P on PDSI
PDSI Trends 1950-2002
Red Drying
dTdP case
Most red areas are statisitically significant
dP only case
33
CLM Simulations 1948-2002

• CLM3 a comprehensive land surface model
  designed for coupled climate simulations. It
  simulates most land surface processes, including
  surface fluxes, soil and snow processes and land
  hydrology (Oleson et al. 2004).
• The CLM3 forcing data combine intra-month
  variations from the NCEP/NCAR atmospheric
  reanalysis with longer-term variations from
  station records (Qian et al. 2005). Variables T,
  P, q, V, Ps and downward solar radiation.
• CLM simulations All-forcing run, dT and dP runs.

34
Palmer Drought Severity Index (PDSI)

• The PDSI is computed using a bucket-type land
  surface model using observed P (Chen et al.02)
  and T (Jones Moberg03)
• It is a normalized measure of the cumulative
  departure in atmospheric moisture supply (P) and
  demand (PE) at the surface
• PE is based on Thornthwaite (1948)
• PDSI is correlated with observed soil moisture
  content
• Caveats No vegetation, no snow processes, crude
  estimate of E, not always comparable spatially,
  etc.