Climate Modelling in Southeast Asia is Critical for Food Security

Southeast Asia’s farmers are dependent on seasonal rainfall and weather patterns to cultivate crops. Climate change is upending this. As a result, governments and countries need to adapt agricultural practices to protect food security. This includes upping their game in the realm of climate modelling.

The warnings are stark. The ASEAN State of Climate Change Report 2021 has identified agriculture as a highly vulnerable sector. Current Nationally Determined Contributions (NDCs) submitted before COP26 is projected to lead to 2.8 degree Celsius of warming by 2100.

ASEAN governments have ramped up near-term weather forecasting (next days, weeks or months) and communications to improve farmer and business decision-making to reduce crop losses. For example, the Lao Climate Service for Agriculture (LaCSA) provides agrometeorology services with 3-hour, 7-day and 3-6 month weather forecasts useful for decision making on irrigation, planting and harvesting. The forecasts are disseminated through channels such as bulletins, television, WhatsApp, Facebook and YouTube. International organisations have sought to strengthen forecasting capabilities through capacity building. This includes the World Meteorological Organisation (WMO), the ASEAN Specialised Meteorological Centre (ASCM), as well as those offered by ASEAN dialogue partners.

However, to be food secure, ASEAN countries need to understand how climates will change in 20- or even 50-year time frames, as climate change’s impact on agriculture is projected to lead to a 6 per cent decline in the region’s GDP if countries do not adapt. Climate modelling is needed to facilitate land-use planning, develop infrastructure, or divert funding for crop resilience research. To do so requires access to local- and national-level climate scenarios for impact assessment and decision-making. This can only be developed through the use of high-resolution climate models.

For a more food-secure future, countries and governments need to follow the science available in climate modelling to avoid being blindsided.

As described by the US’ National Centre for Atmospheric Science, a climate model is a computer simulation of Earth’s systems, which takes into account the atmosphere, ocean, land and ice, to recreate past climates and predict future climates. This requires access to vast amounts of data, massive supercomputing power, and highly skilled scientists to set and analyse them; this is not something every country can do. There are approximately 100 global climate models, created by well-resourced international scientific institutions. These models have contributed to the Coupled Model Intercomparison Projects (CMIP) referenced by the Intergovernmental Panel for Climate Change (IPCC) in its Assessment Reports (most recently referenced is CMIP6).

However, CMIP6 global models have a coarse spatial resolution of up to 150 km. To use climate models for meaningful risk assessment and decision-making on agriculture, a downscaled (or zoomed-in) model is required (see Figure 1 for illustration) — preferably with higher spatial resolutions of 10 km or less. Resolutions of this scale can provide insights into which areas might experience varying conditions such as higher temperatures, changes in precipitation and inundation from sea level rise. This is crucial for climate impact assessment for the agriculture sector.

Figure 1: A Simplified Illustration of Coarse versus Higher Spatial Resolution in Climate Modelling

Most nations in Southeast Asia, except for Vietnam, the Philippines and Singapore, have yet to create downscaled climate models, though many are in varying stages of development (Table 1). Commonly sourced models are from the Coordinated Regional Climate Downscaling Experiment – Southeast Asia (CORDEX-SEA) and/or Singapore’s Third National Climate Change Study (V3). CORDEX-SEA is a downscaled model with spatial resolution of 5 km based on CMIP5 (CMIP6 downscaling is ongoing), whereas V3 has an 8 km resolution model based on CMIP6 (and a 2 km resolution for Singapore and Peninsula Malaysia). Others have produced agriculture impact assessments based on other projections, such as hydrological projections, which do not provide a complete picture of climate impacts. In the absence of national downscaled models, the impact of climate change on agriculture and food security cannot yet be fully grasped.

The Need to Zoom In

Table 1: Climate Modelling Agencies and Climate Impact Modelling Capabilities

Country	National agency in charge of climate modelling and/or projection	Capacities in climate impact modelling	Climate impact models created for agriculture?
Brunei	Brunei Darussalam Meteorological Department (BDMD)	Preliminary results from downscaled CMIP6, CORDEX-SEA and V3	Unknown
Cambodia	Ministry of Water Resources and Meteorology	Internal capacity for climate modelling yet to be developed	Not using climate models.
Indonesia	Badan Meteorologi, Klimatologi, dan Geofisika (BMKG)	CORDEX-SEA Projections period, GCM CSIROMK3.6 and Hadgem Resolution 25 x 25 km)	Yes
Lao PDR	Climate, Agro-Meteorology and Remote Sensing Division, Department of Meteorology and Hydrology,	Internal capacity for climate modelling yet to be developed	Not using climate models.
Malaysia	MetMalaysia, Universiti Kebangsaan Malaysia (UKM), National Water Research Institute of Malaysia (NAHRIM)	In the process of development, based on downscaled CMIP6 and CORDEX-SEA, 4km resolution	Unknown
Philippines	Philippine Atmospheric Geophysical and Astronomical Services Administration (PAGASA)	Downscaled CMIP6, 1km resolution	Yes
Singapore	Singapore Centre for Climate Research (CCRS), Meteorological Services Singapore (MSS)	Dynamic downscaled CMIP6 to create a regional model known as the Third National Climate Change Study (v3), 2km resolution.	Climate impact modelling completed but not specific to agriculture as sector is minor.
Thailand	Climate Center, Thai Meteorological Department	In the process of developing CMIP6-based model	Climate impact modelling on agriculture completed by other organisations
Vietnam	Viet Nam Meteorological and Hydrological Administration (VNMHA), Department of Climate Change, Vietnam Institute of Meteorology, Hydrology and Climate Change (IMHEN):	Published climate change scenarios based on CMIP5/AR5 models and CORDEX SEA in 2020, running experimental 4.4km model	Scenario projections for temperature, rainfall, and extreme events, sea level rise can be utilised for agriculture planning.

Downscaled climate modelling can highlight climate trends that may not be apparent through forecasting. For example, Singapore’s V3 model identified days with high wet bulb globe temperature (WBGT) — a composite measure that takes into account air temperature, humidity, wind and solar radiation — will be the dominating concern in Singapore. The V3 model forecast that the number of such days will increase from the historical 24 days per year to an average of 270 days. This is equivalent to three in four days by the end of the century (under the worst-case scenario, Table 2). This implies that Singapore farmers should invest in indoor farming and cooling systems. Similarly, the Philippines’ climate modelling has identified areas where the largest decrease in very wet days by 2100 will be in Northern Mindanao and Visayas; it implies a need to ramp up drought resilience measures in those regions.

Blistering

Table 2: Number of days annually with daily maximum wet bulb globe temperature (WBGT) equal to or exceeding 33^oC, in Singapore.

Scenario	Number of days with daily max WBGT equal to or exceeding 33oC annually
Observed (based on historical average)	24 days
Future	Mid-century (by 2050)	End Century (by 2100)
SSP1-2.6	75 (53 to 112) days	81(54 to 135) days
SSP2-4.5	87 (61 to 131) days	142 (107 to 205) days
SSP5-8.5	113 (86 to 155) days	270 (207 to 326) days

To secure Southeast Asia’s food future, Southeast Asian nations need to prioritise climate modelling and impact modelling for agriculture. This ensures that leaders and policymakers can start early preparations and allocate sufficient resources for adaptation measures.

Nations that have yet to develop their own climate models should consider accessing already developed regional models, such as CORDEX-SEA or V3. V3 data was made freely available for download in April 2025. Sea level projections can also be developed using NASA’s SLR Projection tool. These models’ accuracy can be improved if governments make more weather station and tide gauge data publicly available.

These models can then be used to create climate impact assessments for sectors including agriculture. Relevant agencies, research organisations and other stakeholders can then develop risk maps and adaptation plans. This can hasten much-needed preparation for agricultural resilience.

Climate change is propelling nations and farmers into uncertainty. Climate modelling, however, can provide some illumination. For a more food-secure future, countries and governments need to follow the science available in climate modelling to avoid being blindsided.

2025/196

The author would like to acknowledge the organisers and participants of the Fourth Workshop on ASEAN Regional Climate Data, Analysis and Projections (ARCDAP-4), 21-24 April 2025, Singapore, organised by the ASEAN Specialised Meteorological Centre and Centre for Climate Research Singapore.