Agriculture highly depends on climate conditions, as climate variabilitycan affect crop yield and product quality. As climate continues to change due to human activities, the agricultural sector becomes more vulnerable to climate risks. Therefore, adaptation strategies are needed to reduce the impacts. One potential adaptation is by using seasonal climate forecasts (SFCs), which provide information about the expected seasonal conditions in the relatively short term and can be used to make informed decisions on agricultural practices based on possible future climate conditions^[1][2].

Climate extremes and long-term warming present significant threats for agricultural productivity, forestry, and fisheries in worldwide regions^[3][4]. Severe events like droughts, floods, and heatwaves directly contribute to reduced crop production, rising food prices, and increased food insecurity for millions of people^[3]. There is an urgency to address these climate challenges since agricultural production is still inherently sensitive to climate variability, which is the leading source of interannual yield changes globally^[5]. With baseline temperatures rising and extreme weather intensifying, traditional climate risk management faces evident constraints, driving agricultural systems to transition from basic management changes to more systemic adaptation measures^[5].

Modern agriculture climate risk is conceptualized by the IPCC risk framework (Figure 1). As shown in Figure 1a, climate risk typically arises at the confluence of three core elements: climate hazards (such as heatwaves or droughts), agricultural exposure (such as location of crops and livestock), and socioeconomic vulnerability^[3]. The IPCC Sixth Assessment Report (AR6), however, extends this approach (Figure 1b) to include the complexity of adaptation and mitigation actions^[3]. In the agricultural environment, human resources, such as changes in land use , governance, and adaptive technology directly influence hazards, exposure, and vulnerability^[3]. Hence, climate risk in farming systems is not a merely biophysical occurrence, but a dynamic, interconnected system influenced by both climatic extremes and human policy actions^[5][3].

Based on international climate risk frameworks, the two basic forms of agricultural climate risks are roughly characterized as:

1. Physical Risks: Direct hazards originating from changing climatic conditions.

• Acute Physical Risks: Sudden and high impact climate events such as heavy rainfalls, flash floods, and extreme droughts resulting in direct physical damage to existing crops and causing acute (severe) yield losses of 34%-37% and immediate financial shocks to agricultural production^[6].
• Chronic Physical Risks: Long-term, gradual changes, including persistent temperature rises, changing rainfall seasonality, sea-level rise leading to soil salination, and glacier retreat^[3]. Increasing baseline temperatures can accelerate soil moisture loss and facilitate the spread of agricultural pests and diseases, resulting in gradual decreases in crop production^[7].

2. Transition Risks: Operational and financial risks indirectly caused by governmental, market and technical changes to a decarbonized economy.

• Policy and Regulatory Risks: Stricter environmental policies and regulations that restrict the excessive use of agricultural inputs (such as chemical fertilizers and pesticides) raise operational and transition costs in high-input farming systems^[8][9].
• Market Dynamics: Changing trade norms, private sustainability standards, and evolving consumer demand for low-carbon supply chains creates competitive disadvantages and risk of exclusion from the market for unsustainable farming operations^[8].

Climate change-related physical and transition concerns do not affect all agricultural systems equally. Systemic vulnerability varies greatly by geographical location, economic capital, institutional support and infrastructure^[10]. Due to high dependence on rainfed agriculture, limited access to credit, little crop insurance and lack of downscaled climate knowledge, smallholder farmers in developing regions confront disproportionate climate risk^[11][10]. The pooled risks destabilize food supply systems, increase market price volatility and worsen rural poverty on a global scale^[8].

Multi-level adaptive frameworks are required to manage agricultural climate risk from the farm to national policy:

• Agronomic Adaptations: Adjusting planting dates, adopting heat- and drought- tolerant crop varieties, expanding water-harvesting infrastructure, and using conservation tillage^[5] .
• Climate Information Services: Utilizing seasonal climate forecasts and early warning systems enables proactive decision-making, such as adjusting fertilizer application or altering irrigation schedules, before negative climate conditions occur^[11].
• Systemic Transformation: Beyond short-term field-level adjustments, long-term resilience needs systemic changes in resource allocation, including targeted diversification of production systems and livelihoods, and land-use transitions^[5].

Seasonal climate prediction estimates broad climate conditions from approximately one month to one year ahead, and although it cannot tell a farmer how much rain will fall on a particular afternoon, it can indicate whether the coming season is more likely to be wetter, drier, warmer, or cooler than usual. This distinction matters because agricultural decisions are rarely made one day at a time. Farmers purchase seed, prepare land, organise labour, and commit limited money weeks or months before the outcome is known. Seasonal forecasts therefore support anticipatory risk management by presenting possible climate outcomes, rather than promising a single certain future^[11].

Seasonal predictions combine observations with numerical models that represent interactions among the atmosphere, oceans, land, and other parts of the Earth system. Satellites, weather stations, ocean buoys, and reanalysis products help scientists establish the initial climate state, while datasets such as ERA5 provide a consistent record that supports monitoring and model development^[12]. Because tiny differences in starting conditions can produce different outcomes, forecasting centres ensembles, in which many simulations explore a range of plausible futures.  

Forecast centres often combine outputs from several models into a multi-model ensemble because each model represents atmospheric and oceanic processes differently. Agreement across models can strengthen confidence, while disagreement indicates that decision-makers should consider a wider range of outcomes. Before release, the results are commonly compared with historical observations and converted into probabilities, which helps transform raw model output into climate information that is more meaningful for a particular region, season, and agricultural decision^[11].

Much seasonal predictability comes from climate components that change more slowly than daily weather. Sea-surface temperatures influence atmospheric circulation for months, while El Niño and La Niña, the Indian Ocean Dipole, the Southern Annular Mode, soil moisture, and long-term warming can shift regional rainfall and temperature probabilities. However, a climate driver does not create the same agricultural effect everywhere, because topography, coastlines, soils, and seasonal timing modify its influence. A strong El Niño may increase drought risk in one farming region, while another region becomes wetter, so global signals must be interpreted through local climate knowledge and production conditions.

Seasonal outlooks are commonly expressed probabilistically, such as a 60 percent chance of below-normal rainfall. Probabilistic forecasting is valuable because it quantifies uncertainty instead of hiding it behind a single best guess^[13]. A 60 percent probability does not guarantee drought, but indicates that below-normal rainfall is more likely than under usual climatological conditions.

A forecast becomes useful only when it arrives before a real decision and can be translated into feasible choices. Before planting, farmers may adjust sowing dates, select crop varieties, prepare irrigation, or delay costly inputs, and during crop growth they may revise water allocation, fertiliser use, and pest surveillance. Research shows that seasonal information can support key agricultural decisions across the production cycle^[15], while downscaled forecasts can also strengthen advance planning for climate-sensitive pest populations^[16]. In Australia, forecast-informed management has produced measurable economic benefits, although the value varies across crops, regions, and farm systems^[14].

Even so, a seasonal forecast should never be treated as an instruction that overrides a farmer's experience. Soil condition, market prices, labour, credit, household responsibilities, and tolerance for loss all shape what action is realistic. Forecasts are therefore most effective when scientists, extension officers, and producers interpret them together, because trust grows through dialogue and locally relevant decisions, not through a coloured map alone. As Darbyshire et al. (2020)^[17] argue, forecast value is context-dependent, so successful adaptation requires both technical skill and a human process that respects how farmers live with risk.

Even though SCF has advanced in recent years and in many parts of the world, some challenges and gaps remain that may hinder farmers' engagement in using SCF.

SCFs are valued when they deliver economic benefit by increasing profit or avoiding loss. Forecast value increases with reliability^[17]. The accuracy of seasonal climate forecasts varies for different regions and weather parameters. For example, the European Centre for Medium-Range Weather Forecasts (ECMWF) showed inadequate skill for predicting precipitation in some parts of the world. This is worrying because rainfall is arguably the most important parameter and information for farmers^[18].

It is essential to improve forecast reliability. When farmers utilized SCFs for farm decision-making, inaccurate forecasts can lead to failed adaptation measures. Consequently, farmers may decrease their confidence in SCFs. Forecast accuracy can be improved by increasing model resolution, better physical processes, improved initial conditions, representation of model uncertainty and international collaboration^[19].

Often meteorology institutes provide general information regarding forecasts. For example, the Indonesian Agency for Meteorology, Climatology, and Geophysics (BMKG) provides season onset, rainfall probabilistic compared to the average, and peak of season information^[20]. However, farmers are expecting more detailed and practical information, such as rainfall amounts, planting guidance (type of plant and varieties), onset and duration of drought, and sun severity, which all support farm-level decisions^[21].

The type of information desired by farmers may vary depending on regions; thus, to understand farmers’ needs, two-way communication between scholars and farmers is ideal^[22]. Collaboration also has to develop up to government institution levels to establish standard and continuous information for such detailed information. For example, information such as planting guidance cannot be produced only by a meteorology institution; agriculture institutions that have a better understanding and data around agriculture must be involved.

Even when climate forecasts are available, the information itself may not be accessible for all. Miscommunication can occur between forecast providers and farmers. This is due to the use of scientific terminology, farmers’ expectations for accurate forecasts, and how farmers make decisions not solely based on forecasts^[11]. In other cases, expected information is only available in a specific form of media. For example, in East Africa, detailed forecasts are available in newspapers; thus, only literate people can utilize the information, while farmers who are less literate depend on radio^[21]. Increasing farmers' climate literacy, including introducing forecast uncertainty and widespread SCF information, can help farmers to better utilise SCFs for agricultural adaptation.

There is also inhomogeneous SCF information, reliability, and literacy across the world. Typically, developed countries have better climate forecast information, reliability, and literacy compared to developing countries. For example, in Australia, the Bureau of Meteorology (BoM) provides forecasts with finer resolution and continuously improves its accuracy^[23]. In developing regions, due to limited resources, forecasts are less advanced, resulting in lower reliability (Figure 4). For example, IGAD Climate Prediction and Application Centre provides climate services for 11 countries in East Africa with coarse resolution. Farmers in these countries opt for indigenous knowledge as they perceive climate forecasts to be unreliable^[21]. Although useful, indigenous knowledge may become less accurate due to increased rainfall variability as the planet warms^[24]. It is essential to improve seasonal climate forecasts in developing countries as these regions typically have lower adaptive capacity compared to farmers in developed countries. Thus, farmers in developing countries may suffer severe losses due to misguided decisions.

a. Australia seasonal climate forecast	b. East Africa seasonal climate forecast
Figure 4. Comparison between developed (Australia) (a) and developing (East Africa) (b) Seasonal Climate Forecast. (source: https://www.bom.gov.au/ and https://www.icpac.net/about-us/)

Australia became one of the first countries to implement SCFs in 1989^[25], and the development of SCFs was mainly driven by Australia’s high rainfall variability compared to other regions with similar climate conditions^[26]. These conditions emphasise the importance of seasonal forecasts to protect agricultural sectors that depend on rainfall (Parton et al., 2019).

The first SCFs were based on statistical relationships between the Southern Oscillation Index (SOI) and rainfall data^[27], before being extended to include SOI phase and sea surface temperature^[25]. Then, in 2013, Australia implemented a dynamical model using the Predictive Ocean Atmosphere Model for Australia (POAMA), which can enhance rainfall and maximum temperature accuracy^[25]. POAMA was subsequently replaced by the Australian Community Climate and Earth-System Simulator–Seasonal (ACCESS-S), which has a higher model resolution (60 km horizontal with 85 vertical levels), resulting in reduced climate bias^[28].

SCFs information in Australia is provided by the Bureau of Meteorology (BOM) through probabilistic forecasts of rainfall and temperatures, showing regions that have a likelihood of wetter, drier, warmer, or cooler conditions from weeks to months across different regions. To make this scientific climate information more understandable and accessible for farm decision-making, Agriculture Victoria collaborated with Grain Research & Development Corporation (GDRC), South Australian Research and Development Institute (SARDI), and Federation University through the “Using Seasonal Forecast Information and Tools” project. The project produced several outputs, including “The Fast Break”, which delivers seasonal forecast commentary  through newsletters, webinars, and engaging short videos; the “Local Climate Tool”, which helps users to understand how ENSO/IOD influence local rainfall; and forecast guidance. Additionally, the project included training workshops and farmer case studies demonstrating how the use of SCF information can inform farm and risk management decisions and improve profitability, which can provide practical examples for other farmers^[29].

The use of SCF information also provides economic benefits for Australian agriculture by helping farmers adjust their management practices^[14]. Luo et al. (2024)^[30] noted that SCF can increase crop production by 68%, averaging 281 kg/ha in six locations across eastern Australia. At the national scale, the Centre of Economics (2014)^[31] estimated the economic benefit of improved seasonal forecasting to be around $110 million to $ 1,930 million.

Indonesia began operating the SCF system in 1993 under BMKG, using a statistical-analogue approach based on SOI–rainfall relationships across 102 regions and seasonal outlooks from BOM^[32]. Over time, BMKG enhanced its seasonal forecasting by introducing the statistical application Hybrid BMG in 2011 and started incorporating dynamical seasonal models from ECMWF in 2014 to generate climate predictions up to seven months^[33][34]. The latest SCF in Indonesia was based on the combination of statistical models with raw and post-processed dynamical models from ECMWF SEAS5, improving high annual rainfall cycle correlation beyond 0.8 in most Indonesian regions^[35].

Operational SCF in Indonesia provides wet and dry season predictions, rainfall variability, and potential forest fire risk, supporting agricultural decision-making. BMKG collaborated with multi-sector institutions to start developing the Food Crop Planting Calendar Atlas in 2007, which further extended into the Integrated Cropping Calendar Information System, namely “Siap Tanam 2.0”^[36]. These systems provide information regarding planting time and area, suitable crop varieties, potential risk of drought and flooding at subdistrict level^[37].       

(a) dry-season	(b) wet-season
Figure 7. Examples of SCF information in Indonesia, presenting predicted two different season: dry-season (a) wet-season (b). (source: https://iklim.bmkg.go.id/id/)

The Indonesian government has continued to support farmers’ access to SCF information through the Climate Field School (SLI), which was established in 2010. This programme was supported by the Australian Government through AusAID to help farmers use seasonal climate information in agricultural practical decisions, including planting schedules and crop management^[38]. A case study in East Nusa Tenggara reported a twofold increase in crop yields after participating in SLI, where farmers applied seasonal climate information to their farming practices^[39].

Figure 8. The application of SCFs in Indonesia for agriculture sector: the Integrated Cropping Calendar Information System (a) Climate Field School in Subang Regency, 2025 (b)[40]

To maximise the benefits of SCFs in agricultural adaptation, future efforts can focus on improving their usability, accessibility, and institutional support.

While SCFs provide valuable information about future climate conditions, their effectiveness arises when this information can result in decision-making that reduce the vulnerability and impacts of climate variability. The benefits of SCFs move beyond forecast accuracy to usability, whether the information is available at the appropriate time, whether it represents relevant climate risks for agriculture, and whether it can inform feasible management responses for farmers^[1]. Therefore, forecasts should be issued with a sufficient lead time that enables farmers to consider the information before finalising decisions for production cycle^[41].  To improve their agricultural relevance, SCFs should move beyond broad tercile categories by relating forecasts to local historical observations and providing probabilities for locally meaningful climate conditions, including rainfall onset and dry spell duration, such as the Enhancing National Climate Services (ENACTS) in Africa^[42][43]. Moreover, SCFs become more actionable when combined with feasible adaptation responses, such as climate smart agriculture (CSA) practices^[44].

The aim of improving accessibility of SCFs is to ensure that farmers can reach climate information and strengthen their capacity to understand, interpret, and apply the information in their decisions^[45]. To achieve this need, capacity building could be applied as integral component of SCFs implementation, by involving farmers, agricultural extension officers, and other intermediaries to help farmers translating climate information into practices^[45]. For example, the implementation of Participatory Integrated Climate Services for Agriculture (PICSA) in Africa, Latin America, and South Asia to combine climate information and agricultural decisions through participatory learning (Figure 9).

Moreover, accessibility also depends on how information is communicated. Climate information should be communicated using clear and locally understandable language, since technical or ambiguous terms may be interpreted differently by farmers^[46], as illustrated in the Victoria’s Fast Break and Climatedogs. Furthermore, to continuously improve SCFs with local needs, establishing two-way communication between forecast providers and farmers can allow farmers to have opportunities to provide feedback on the usefulness, timing, and relevance of climate^[43][47].

While improved forecast usability and accessibility can enhance farmers’ ability to use climate information, the long-term effectiveness of SCFs depends on supportive institutions, policies, and resources. Agricultural adaptation requires coordination  multiple actors, including meteorological agencies, agricultural departments, research institutions, extension services, local governments, and farmers, as illustrated by India’s Agrometeorological Advisory Services^[48]. In addition, institutional support is needed to integrate SCFs into existing agricultural planning and risk-management systems^[49], as illustrated by the ASPIRE project in the Sahel, which explored how seasonal forecasts could inform adaptive social-protection and early-action systems. Furthermore, farmers may understand climate risks and receive appropriate recommendations, but their ability to respond can be constrained by economic and technical barriers, including limited access to credit and agricultural support services ^[50]. Therefore, climate adaptation should be supported by enabling policies and programmes that provide farmers with the resources and opportunities needed to implement appropriate responses, such as credit supply, food crisis management, trade and agricultural insurance^[2].

• Weather forecasting

Acknowledgement: Grammarly was used in the making of this article.