Using Climate Modeling to Analyze Asian Summer Monsoon Change

By Xiaoqiong Li, Climate and Society ’13

A “Monsoon” is traditionally defined as a seasonal reversing surface wind and corresponding change in precipitation. Monsoon regions experience a sharp contrast between a wet summer and a dry winter. The Asian monsoon, including regional monsoons over South Asia, East Asia and Northwest Pacific, is one of the major monsoon systems of the world. Since summer monsoon precipitation dominates the total annual precipitation in these regions, analyzing changes to the monsoon has important implications for water resources, agriculture, food security, etc.

My internship, following research work completed through the academic year, is focused on analyzing Asian summer monsoon variability and change, both 20th century historical period and projections for 21stcentury. I work with Dr. Mingfang Ting at the Division of Ocean and Climate Physics, Lamont-Doherty Earth Observatory of Columbia University.

We use observational datasets and the most recent climate model simulations in Coupled Model Intercomparison Project Phase 5 (CMIP5) performed for the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5). Based on statistical methods used abundantly in climate science such as Empirical orthogonal functions (EOF) analysis and linear regression, CMIP5 multi-model simulations show a clear drying trend in summer monsoon precipitation (June – August) over large area of east China and India in the 20th century and general wetting trend in 21th century projections under RCP8.5 emission scenario (Figure 1). The reversal of the past trend in future models is quite interesting and needs more work to try to analyze the physical processes controlling these changes, especially their links with anthropogenic climate change.

August precipitation linear trend in Asian monsoon region in models simulations, averaged across 23 CMIP5 models for the 20th century. Units are mm/day per century.

August precipitation linear trend in Asian monsoon region in models simulations, averaged across 23 CMIP5 models for the 21st century. Units are mm/day per century.

In climate science research, climate modeling allows scientists to simulate past climate and make future projections under different scenarios. However, limitations in climate models also add more uncertainties in simulated results, especially when it comes to making future projections. For example, model results have quite significant discrepancies compared to observations (mostly mixed signals) over Asian monsoon region for 20^th century precipitation change. This raises the question of how much we can trust the models. Therefore model skill is an important factor to consider in modeling research. In this study, we used 23 selected climate models to compute the multi-model mean to lower the biases by using individual models. Performing evaluation of the models could also help to eliminate models that are less accurate and thereby improve the quality of future climate change projections.

In this internship, most of my work is associated with manipulating large datasets and model outputs completed through computer programming in Matlab and Ingrid (the language which IRI/LDEO Climate Data Library is built on). This is where I benefited a lot through the coursework in statistics, modeling as well as training in programming during the past year. The IRI/LDEO Climate Data Library has a great abundance of climate data conveniently for use, which is essential in conducting research work.