Landuse and climate change Eugenia Kalnay, Ming Cai, Hong Li, YoungKwon Lim University of Maryland a - PowerPoint PPT Presentation

Loading...

PPT – Landuse and climate change Eugenia Kalnay, Ming Cai, Hong Li, YoungKwon Lim University of Maryland a PowerPoint presentation | free to view - id: 210ae5-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Landuse and climate change Eugenia Kalnay, Ming Cai, Hong Li, YoungKwon Lim University of Maryland a

Description:

University of Maryland and Florida State University. Mario Nunez and C. Ciappesoni, Argentina ... Results from China (Zhou, Dickinson, et al, PNAS, 2004) ... – PowerPoint PPT presentation

Number of Views:36
Avg rating:3.0/5.0
Slides: 34
Provided by: euge123
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Landuse and climate change Eugenia Kalnay, Ming Cai, Hong Li, YoungKwon Lim University of Maryland a


1
Land-use and climate change Eugenia Kalnay,
Ming Cai, Hong Li, Young-Kwon LimUniversity of
Maryland and Florida State UniversityMario Nunez
and C. Ciappesoni, Argentina(Nature, 2003,J.
of Climate, 2005, JGR, 2006, in press,Lim et
al, GRL 2005,Nunez et al, in preparation,Lim,
Cai and Kalnay, 2006)
  • Also Zhou, Dickinson et al, 2004, PNAS

2
Reanalysis and Observations Example Baltimore
City, MD
In many stations the observations (in red) become
warmer than the reanalysis (in blue). We
attribute the difference OMR in the trends, at
least partially, to land properties and changes
not included in the Reanalysis.
3
OUTLINE
  • OMR Observation minus Reanalysis method to
    estimate surface effects on climate change
  • Eastern US (Kalnay and Cai, Nature, 2003, Jan
    2006 JGR)
  • Land-surface impact is strongly regional ( -)
  • Cai Kalnay, J of Clim 2005 reanalysis w/frozen
    model can estimate well greenhouse trends from
    observations
  • Results from China (Zhou, Dickinson, et al, PNAS,
    2004)
  • Results from Argentina (Nunez et al, in
    preparation)
  • Comparison of NCEP and ERA-40 (Lim et al, GRL,
    2005)
  • Vegetation reduces warming (Lim et al, 2006, in
    prep.)
  • Conclusions
  • www.atmos.umd.edu/ekalnay

4
Highlights
  • A method to estimate the impact of land-use
    change (including urbanization and agriculture)
    on surface temperatures
  • Compare the surface observations over the US with
    the NCEP 50 year Reanalysis, which is insensitive
    to land-use change.
  • OMR Observation minus Reanalysis trends
    (Nature, 2003)
  • The impact of land-use changes is comparable to
    that of greenhouse gases in decreasing the
    diurnal temperature range.
  • Land-surface impact is strongly regional and
    vegetation dependent (Lim et al 2006)
  • Use of corrected observations does not change the
    geographical distribution of the results and can
    be added a posteriori.
  • (A reanalysis made with a frozen model can detect
    an anthropogenic trend present in observations,
    Cai Kalnay, 2005)
  • Similar results in China and in Argentina (Zhou,
    Dickinson, et al, Nunez et al)
  • Results with NCEP and ERA-40 very similar (Lim et
    al, 2005)

5
Observations minus Reanalysis
  • Reanalysis Atmospheric observations for the last
    55 years are assimilated into a global
    atmospheric model (like in operational numerical
    weather prediction) but keeping the model and
    method of assimilation frozen. This avoids
    spurious climate jumps in operational analyses
    when the system is changed.
  • The NCEP-NCAR Reanalysis (NNR) used a 1994 system
    and did not use surface observations over land,
    so that it is insensitive to surface or land
    changes.
  • The ERA40 used surface data (CRU) indirectly, to
    initialize the soil temperature and moisture, so
    that it is somewhat sensitive to land changes.
  • We compare the surface observations over the US
    with the NCEP 50 year Reanalysis. OMR
    Observation minus Reanalysis trends (Nature,
    2003).
  • Land-surface impact is strongly regional and
    vegetation dependent (Lim et al 2006). Similar
    results in China and in Argentina (Zhou,
    Dickinson, et al, Nunez et al)
  • Results with NCEP and ERA-40 are qualitatively
    very similar (Lim et al, 2005, 2006) and show
    strong vegetation dependence

6
Previous methods estimated only urbanization
they classify stations into rural and urban. They
depend on the method of classification, do not
account for agriculture, irrigation, land-type or
aerosols.
Urban adjustments comparing urban and rural
observations
(degrees/100-years)
Population data (Easterling et al 1996)
Satellite nightlight (Hansen et al, 2001)
7
Note that there are areas of urban warming and
urban cooling
Urban adjustment (degrees/100-years)
-0.015C/decade small mean temperature impact, but
regionally it can be /-0.1C/decade What about
Tmax, Tmin and DTRTmax-Tmin?
Population data (Easterling et al 1996)
Satellite nightlight (Hansen et al, 2001)
8
Data for US study
  • Daily surface Tmax and Tmin from NCDC
    Cooperative Summary of the Day dataset over 48
    conterminous United States for 1950-1999. No
    adjustments are included
  • Global daily surface air Tmax and Tmin from
    NCEP/NCAR Reanalysis on 2.5 by 2.5 grids for
    1950-1999.
  • Use of corrected observations does not change
    geographical distribution and can be added

9
Analysis Procedures
  • Interpolate linearly the gridded reanalysis data
    to observational sites.
  • Only consider the sites that at least have total
    of 480 (whole) months of observations.
  • Only consider the sites with elevation lower than
    500 meters (total 1982 stations).
  • Obtain monthly data by averaging daily data.
  • Remove annual cycles from each dataset at each
    observational site (minimizes systematic errors).
  • Eliminate impact of introducing satellite
    observing system by computing trends before and
    after 1979.

10
Example Baltimore City, MD
In some stations the observations (in red) become
warmer than the reanalysis (in blue). We
attribute the difference OMR in the trends, at
least partially, to land properties and changes
not included in the Reanalysis.
11
50 year correlation of monthly mean anomalies
Over 80 correlation except over the Rockies and
in the West Coast
We only consider stations below 500m
Because of the poor correlation in the West Coast
(model orography), we exclude it from the
computations (originally got strong OMR warming)
12
40-year trends (degree per decade)
13
There is similarity in the geographical
distribution of the night-light urban trend
estimate (Hansen et al, 2001) and the NNR
land-use trend estimate(sign of delta T changed
to compare with Hansens correction)
14
Nonclimatic corrections of NCDC (unlike OMR) have
no clear geographical distribution and can be
added afterwards. The OMR method could be used
to estimate corrections.
Non-urban adjustment trends
OMR trends (Tmean)
15
Tmin Summer and Winter
WINTER
SUMMER
Uniform in summer
Regional in winter
16
Weekend effect (Forster and Solomon, 2003)
Something is different about the Midwest?
OMR in Winter
DTR
Tmin
Random DTR
17
Use of a frozen model (fixed CO2)(Cai and
Kalnay, accepted in J. of Climate)
  • Intuition suggests Reanalysis would show only a
    watered down trend of greenhouse warming, but it
    is wrong.
  • Cai and Kalnay (2004, J of Cl) show analytically
    that essentially the full trend present in the
    observations appears in the Reanalysis after a
    short transient
  • (e.g., Reanalyses show volcano impacts even
    without volcanoes)

18
Ratio of the trend between two consecutive
analysis cycles to the trend in the observation
as a function of N, the analysis step starting
from N 1, and (1 a), the weight assigned to
the observation in the data assimilation
procedure for dt/tau 0.01. (Cai and Kalnay, J.
of Clim., 2004)
1-a
19
Summary of trend statistics, 40 years
20
Results in China (Zhou, Dickinson et al, 2004,
PNAS)
Tmax
Tmin
DTR
21
(No Transcript)
22
Comparison of Ob-R2 over SE China and greenness
index
23
Observed minus Reanalysis annual trends
Small increase in Tmax
Large increase in Tmin
Decrease in DTR
24
Correlations at a provincial level
DTR correlated with increase urban
DTR correlated with decrease in greenness index
25
Nunez et al 2006Surface temperature trends
estimated by Hansen et al. anomalous behavior
over southern South America (Argentina).
26
Results from Argentina (Nunez et al, 2004)There
is a good agreement between raobs and NNR in the
interannual variability with a correlation
varying from 0.7 to over 0.9
27
In the observations, strong cooling of Tmax, weak
warming of Tmin, except in Patagonia contrast
between north and south
28
OMR Mean temperature
OMR DTR
-0.0940
OMR shows land changes not included in the
reanalysis result in a warmer Tmean, and a
reduction in DTR, especially in the center of the
country, where precipitation has been increasing.
29
Tendencies of precipitation, DTR and Ha planted
with soy
OMR DTR trend
-0.940
Precipitation in Argentina has increased
especially where DTR has strongly decreased and
where soybean cultivation has increased
30
Results in Argentina (Nunez et al, 2004)
  • Observations show a strong decrease in mean
    temperature and a strong reduction of the diurnal
    temperature range.
  • OMR indicates that surface effects increased the
    temperature and strongly reduced the DTR.
  • The areas of maximum reduction of DTR coincide
    with those where precipitation has increased, and
    where cultivation of soy has grown exponentially

31
Land cover type (MODIS) and estimated trend from
NCEP and ERA reanalyses (Lim, Cai, Kalnay, Zhou,
GRL 2005)
32
Land cover type (MODIS) and estimated trend from
NCEP and ERA reanalyses (Lim, Cai, Kalnay, Zhou,
2005)
33
Both ERA40 and NNR trends underestimate the
observed trend. They both represent well the
regional interannual changes. ERA40 uses
surface obs. indirectly to initialize soil so its
OMR is similar but smaller than for NNR.
OMR
Observed and Rean trends
From Lim, Cai, Kalnay, 2006
34
Observed trends show no dependence on NDVI
NNR and ERA40 show a positive correlation with
NDVI OMR shows a negative correlation
missing from the Reanalyses Vegetation reduces
the surface temperature trends, bare soils
increase it
OMR
Dependence of trends on NDVI (20N-50N), Lim, Cai,
Kalnay, 2006
GHCN
35
Difference in surface trends depending on the
type of land surface deserts and urban areas
warm strongly, broad leaf forests and agriculture
warm less (Lim et al, 2005)
desert
urban (big, small)
agriculture
broad leaf
36
CONCLUSIONS ON THE USE OF OMR
  • Results over eastern US, China and Argentina
    similar.
  • Impacts are regional, depend on the land-type
  • NCEP and ERA-40 give very similar OMR trends for
    different types of land.
  • Vegetation decreases the surface temperature
    trend
  • Urban and desert areas have warmer Tmean
    broad-leaf forests and agriculture areas warm
    less (but probably reduce DTR)
  • Non-climatic NCDC adjustments are geographically
    uniform. OMR could be used as a proxy to validate
    them.
About PowerShow.com