A PDF file should load here. If you do not see its contents
the file may be temporarily unavailable at the journal website
or you do not have a PDF plug-in installed and enabled in your browser.
Alternatively, you can download the file locally and open with any standalone PDF reader:
https://link.springer.com/content/pdf/10.1007%2Fs00382-018-4177-4.pdf
South Asian monsoon precipitation in CMIP5: a link between inter-model spread and the representations of tropical convection
South Asian monsoon precipitation in CMIP5: a link between inter- model spread and the representations of tropical convection
Samson Hagos 0 1
L. Ruby Leung 0 1
Moetasim Ashfaq 0 1
Karthik Balaguru 0 1
0 Oak Ridge National Laboratory , Oak Ridge, TN , USA
1 Atmospheric Sciences and Global Change Division, Pacific Northwest National Laboratory , 902 Battelle Boulevard, Richland, WA 99352 , USA
2 Samson Hagos
CMIP5 models exhibit a mean dry bias and a large inter-model spread in simulating South Asian monsoon precipitation but the origins of the bias and spread are not well understood. Using moisture and energy budget analysis that exploits the weak temperature gradients in the tropics, we derived a non-linear relationship between the normalized precipitation and normalized precipitable water that is similar to the non-linear relationship between precipitation and precipitable water found in previous observational studies. About half of the 21 models analyzed fall in the steep gradient of the non-linear relationship where small differences in the normalized precipitable water in the equatorial Indian Ocean (EIO) manifest in large differences in normalized precipitation in the region. Models with larger normalized precipitable water in the EIO during spring contribute disproportionately to the large inter-model spread and multi-model mean dry bias in monsoon precipitation through perturbations of the large-scale winds. Thus the intermodel spread in precipitable water over EIO leads to the dry bias in the multi-model mean South Asian monsoon precipitation. The models with high normalized precipitable water over EIO also project larger response to warming and dominate the inter-model spread in the multi-model projections of monsoon rainfall. Conversely, models on the flat side of the relationship between normalized precipitation and precipitable water are in better agreement with each other and with observations. On average these models project a smaller increase in the projected monsoon precipitation than that from multi-model mean. This study identified the normalized precipitable water over EIO, which is determined by the relationship between the profiles of convergence and moisture and therefore is an essential outcome of the treatment of convection, as a key metric for understanding model biases and differentiating model skill in simulating South Asian monsoon precipitation.
1 Introduction
The South Asian monsoon is a prominent large-scale
circulation feature. It influences a significant fraction of the
world population that depends on the monsoon rainfall for
food, energy production and many other economic
activities. Understanding the physical processes that control the
monsoon and its response to natural and anthropogenic
forcings is of high societal and scientific value. Because of the
complex processes involved, representing the monsoon and
projecting its future changes has been a major challenge
in climate modeling. Many global climate models in the
Coupled Model Intercomparison Project (CMIP) display a
dry bias in their simulation of the present day South Asian
monsoon precipitation. While the multi-model mean of the
latest set of models in CMIP5 shows some improvements in
the spatial distribution of precipitation over that of CMIP3,
the inter-model spread remains large and the delayed as well
as weak monsoon rainfall persisted in many models
(Sperber
et al. 2013)
.
The 10-year mean (1996–2005) seasonal cycle of pre
cipitation in the models and the projected summer monsoon
rainfall under the RCP8.5 scenario are displayed in Fig. 1.
Figure 1a shows the seasonal cycle of the 10-year mean all
India precipitation (AIP, defined as the average precipita
tion over the land region from 5°N to 30°N and 70°E to
90°E) from the historical simulations of 21 CMIP5 models
and the observed precipitation from TRMM-3B42
(Huffman
et al. 2007)
, GPCP
(Huffman et al. 2001)
and rain gauge data
from the India Meteorological Division (IMD). The list of
CMIP5 models and the relative strength of their monsoon
14
12
10
1
− y 8
a
d
m6
m
4
2
0
All India Monsoon precipitation
Strong
Strong
Strong
Weak
Strong
Strong
Strong
Weak
Strong
Strong
Strong
Strong
Strong
Weak
Weak
Weak
Weak
Weak
Weak
Weak
Weak
2 4 6 8 10
Present day precipitation (mm/day)
12
precipitation compared to the observations are shown
in Table 1. Besides the considerably drier multi-model
mean summer rainfall than observed, the large inter-model
spread is remarkable. Atmospheric models with prescribed
observed sea surface temperatures (SSTs) reproduce most
of the bias and spread, which indicates that the modeling
issues originate, to first order, from the atmosphere
(Ashfaq
et al. 2016; Bollasima and Ming 2012; Meehl et al. 2006)
,
and more specifically relate to the treatment of convection
(Turner and Slingo 2009; Bush et al. 2015)
. The summer dry
bias is also associated with stronger easterly surface winds
in spring (...truncated)