GRL_Supplementary Materials

Supplementary Materials:
Yen-Ting Hwang1*, Dargan M. W. Frierson1, Sarah M. Kang2
Department of Atmospheric Sciences, University of Washington, Seattle.
School of Urban and Environmental Engineering, Ulsan National Institute of Science and
Technology, Ulsan, Korea
*Correspondence to: [email protected], Department of Atmospheric Sciences,
University of Washington, Box 351640, Seattle, WA 98195-1640
1. Validation and Analysis of the 20th Century Reanalysis Precipitation Data
The 20th century reanalysis (20CR) project only assimilates surface pressure, monthly
sea surface temperature, and sea ice distribution [Compo et al.,2011]. Its precipitation data
requires careful examination. To validate the precipitation in 20CR, we use Maximum
Covariance Analysis [Bretherton et al., 1992; Wallace et al., 1992] to evaluate its ability to
capture year-to-year variability observed in the Global Precipitation Climatology Project
(GPCP) [Adler et al 2003; Xie et al., 2003] during 1979~2010.
The Maximum Covariance Analysis is applied to the annual mean precipitation field in
the 20CR and the GPCP. First of all, the normalized root mean squared covariance is 0.47,
which indicates significant correlation between the two fields. The correlations between the
expansion coefficients in the two fields are above 0.96 for the first five modes. The
heterogeneous maps are shown in Fig. S1. We repeat the analysis on annual mean zonal mean
precipitation field. The normalized root mean square covariance is 0.37. The correlations
between the expansion coefficients in the two fields are above 0.78 for the first five modes. The
heterogeneous maps are shown in Fig. S2.
These results indicate that the 20CR captures most of the year-to-year precipitation
variability during 1979~2010, without assimilating precipitation observations. Only assimilating
a few surface variables not only makes the 20CR independent from the GHCN rain gauge data
but also avoids the artificial trends induced by data from the introduction of newer
measurements like radiosondes or satellite data, a problem that plagues other reanalyses
[Bengtsson et al., 2004; Kinter et al., 2004] .
2. Analysis of Two Additional Precipitation Datasets: Global Precipitation Climatology
Centre (GPCC) and Climate Research Unit (CRU)
A time series of zonal mean precipitation anomalies in three different datasets (GPCC, CRU and
GHCN) is plotted in Fig. S3. Detailed algorisms of GPCC and CRU data are described in
Schneider et al. [2013], Becker et al. [2013], and Mitchell and Jones [2005]. All three datasets
show drying of the northern tropics and wetting of the southern tropics starting from late 1960s
and lasting until mid 1980s, although the magnitudes differ. Note that GHCN has coarser
resolution and presents smoother latitudinal structure.
Fig. S4 demonstrates the spatial structure of the southerward precipitation shift. For GHCN
(Fig. S4(C)), we only include the grids that have data during both time periods we consider
(1931~1950 to 1971~1990), whereas GPCC and CRU interpolate data over the entire time
period over all land area. Note that regions with sufficient GHCN have consistent trend across
the three datasets, for example, all datasets agree with the drying rend in Sahel, Venezuela over
northern edge of South America, and the moistening trend over eastern Brazil. Discrepancies
between GPCC and CRU occur over the regions that GHCN does not have data during
1931~1950. For example, GPCC reports a significant decrease in rainfall over Indonesia and
western Brazil, which are less apparent in CRU, whereas CRU reports a significant increase in
rainfall over northern Australia, which is less apparent in GPCC. These regional discrepancies
project on to zonal mean in Fig. S4(D) and the magnitudes of southern tropics wetting in Fig.
3. Attribution Analysis
The results of the attribution technique described in the main text are plotted in Fig. S5.
Positive values for the y-axis on the right imply this particular term requires an increase in
northward energy transport at the equator and thus may shift the ITCZ southward (y-axis on the
left). The y-axis on the left is calculated from the linear relationship in Fig. 3(A). Models with
more increase in the northward cross-equatorial energy transport are in red, and models with
less increase in northward cross-equatorial energy transport are in blue (last column in Fig. S5).
Even without including the indirect effect from the cloud SW term (Cs), scattering
aerosols are still the most dominant term in multi-model mean. This term also has a wide spread
because GCMs do not have standard prescriptions of scattering aerosol forcing strength and
distribution. The cloud LW effect (Cl) and water vapor greenhouse effect (WV) are positive
correlated with the shift (stacked blue to red). Their positive feedbacks on precipitation shifts
are described in previous studies [Yoshimori and Broccoli, 2009; Frierson and Hwang, 2012].
Supporting References:
Adler, R. F. et al. (2003) The Version 2 Global Precipitation Climatology Project (GPCP)
Monthly Precipitation Analysis (1979-Present). J. Hydrometeor. 4, 1147-1167.
Becker A., P. Finger, A. Meyer-Christoffer, B. Rudolf, K. Schamm, U. Schneider, and M. Ziese
(2013) A description of the global land-surface precipitation data products of the Global
Precipitation Climatology Centre with sample applications including centennial trend
analysis from 1901 – present. Earth System Science Data 5, 71-99
Bengtsson, L., S. Hagemann, and K. I. Hodges (2004) Can climate trends be calculated from
reanalysis data? J. Geophys. Res. 109, D11111, 8PP.
Bretherton, C. S., C. Smith, and J. M. Wallace (1992) An intercomparison of methods for
finding coupled patterns in climate data sets. J. Climate 5, 541-560.
Compo, G. P. et al. (2011) The Twentieth Century Reanalysis Project. Quarterly J. Roy.
Meteorol. Soc. 137, 1-28.
Frierson, D. M. W., and Y.-T. Hwang (2012) Extratropical influence on ITCZ shifts in slab
ocean simulation of global warming. J Climate 25, 720-733.
Kinter, J. L. III, M. J. Fennessy, V. Krishnamurthy, and L. Marx (2004) An evaluation of the
apparent interdecadal shift in the tropical divergent circulation in the NCEP-NCAR
reanalysis. J. Climate 17, 349-361.
Mitchell T. D., and P. D. Jones (2005) An inproved method of constructing a database of
monthly climate observations and associated high-resolution grids. International J. of
Climatology 6, 693-712
Schneider, U., A. Becker, P. Finger, A. Meyer-Christoffer, M. Ziese, and B. Rudolf (2013)
GPCC’s new land surface precipitation climatology based on quality-controlled in situ data
and its role in quantifying the global water cycle. Theoretical and Applied Climatology, 1-26
Wallace, J. M., C. Smith, and C. S. Bretherton (1992) Singular value decomposition of
wintertime sea-surface-temperature and 500 mb height anomalies. J. Climate 5, 561-576.
Yoshimori, M. and A. J. Broccoli (2009) On the link between Hadley circulation changes and
radiative feedback processes. Geophys. Res. Lett. 36 L20703, 5PP.
Xie, P. et al. (2003) GPCP Pentad Precipitation Analyses: An Experimental Dataset Based on
Gauge Observations and Satellite Estimates. J. Climate 16, 2197-2214.
Figure Legend
Fig. S1. Heterogeneous regression maps from MCA of GPCP and 20CR precipitation fields.
(A), (C) Covariance from the 20CR precipitation field regressed upon the first and second
normalized GPCP precipitation expansion coefficients, respectively; (B), (D) covariance from
the GPCP field regressed upon the first and second normalized 20CR expansion coefficients,
respectively. Spatial correlation coefficients are 0.86 between (A) and (B) and 0.78 between (C)
and (D).
Fig. S2. Heterogeneous regression maps from MCA of GPCP and 20CR zonal mean
precipitation fields. Spatial correlation coefficients are 0.90 and 0.85 between the red and the
blue lines in (A) and (B), respectively.
Fig. S3. Zonal mean precipitation anomaly (relative to the 20th century mean) based on (A) the
Global Precipitation Climatology Centre (GPCC) and (B) the Climate Research Unit (CRU) and
(C) Global Historical Climatology Network (GHCN) gridded products. Values are smoothed
with the 13-point filter to remove fluctuations of less than decadal time scales (as in Fig. 2).
Fig. S4. Changes in precipitation from 1931~1950 to 1971~1990 based on (A) the Global
Precipitation Climatology Centre (GPCC) and (B) the Climate Research Unit (CRU) and (C)
Global Historical Climatology Network (GHCN) gridded products (D) the zonal mean of (A),
(B), and (C).
Fig.S5. Attribution of precipitation shifts in (A)CMIP3 and (B)CMIP5 models. The columns are
scattering aerosols (As), the shortwave cloud effect (Cs), the surface albedo effect (I), aerosol,
water vapor, and ozone absorption (Aa), the longwave cloud effect (Cl), the lapse rate effect
(LR), the water vapor greenhouse effect (WV), the surface temperature effect (Ts), the
longwave residual term (LWr), surface flux change (O), and the sum of all terms above (All).
Each circle represents one GCM. Open circles are GCMs with no indirect aerosol effect
parameterization. Close circles are GCMs with indirect aerosol effect parameterization. The X
symbols denote the multi-model mean in each column.