Boosting Thailand’s palm oil yield with advanced seasonal predictions

Parichart Promchote, Binod Pokharel2,∗, Liping Deng3,∗, Shih-Yu SimonWang, Jin-Ho Yoon and Piya Kittipadakul 1 Department of Agronomy, Kasetsart University, Bangkok 10900, Thailand 2 Central Department of Hydrology and Meteorology, Tribhuvan University, Kirtipur 44600, Nepal 3 College of Ocean and Meteorology, Guangdong Ocean University, Zhanjiang 524088, People’s Republic of China 4 Department of Plants, Soils, and Climate, Utah State University, Logan, UT 84322-4820, United States of America 5 School of Earth Sciences and Environmental Engineering, Gwangju Institute of Science and Technology, Gwangju 61005, Republic of Korea ∗ Authors to whom any correspondence should be addressed.


Introduction
Palm oil, an indispensable global commodity, plays a crucial role in Thailand's economy (Maluin et al 2020). This versatile product, found in everything from food to biodiesel and industrial applications (Kamil andOmar 2017, Chiarawipa et al 2020), cements Thailand's position as the third-largest producer in the world. The industry is particularly vital in the southern regions, where most plantations are located (Dallinger 2011), providing employment opportunities in rural areas and bolstering the nation's GDP and foreign exchange earnings through exports. However, oil palm productivity hinges on climatic conditions, with precipitation patterns being especially influential. Both local climate and remote forces like El Niño-Southern Oscillation (ENSO) significantly impact production. In neighboring Malaysia, research by Kamil and Omar (2017) has uncovered the consequential role of ENSO events on precipitation and, subsequently, palm oil productivity.
Despite progress in ENSO prediction and seasonal forecasts, a persistent gap remains between our understanding of climatic influences on oil palm productivity and reliable prediction methods that leverage these influences. While there is a wealth of literature on seasonal and sub-seasonal prediction, it is important to recognize that there are still technical challenges and considerable uncertainty in these predictions. This perspective article highlights the urgency to bridge this gap, examining existing research and advocating for the development of dynamic seasonal prediction models for palm oil production. Addressing this challenge could substantially enhance the industry's resilience and sustainability in the face of climate variability and change. Our preliminary investigation into the relationship between palm oil yield and climate underscores the immense potential of seasonal prediction as an invaluable tool for securing the future of southern Thailand's palm oil industry.

Data
For our analysis, we obtained Thailand's palm oil yield (kg rai −1 ) data from the Office of Agricultural Economics, Ministry of Agriculture and Cooperatives and the Department of Internal Trade, Ministry of Commerce. These data provided insights into production trends and seasonality. We focused on the annual palm oil yield time series from 2003 to 2021, with a strong emphasis on southern Thailand. The unique seasonality of palm oil production in this region, characterized by two peak harvests per year, highlights the importance of understanding climate factors for ensuring stability and economic significance. Chiarawipa et al (2020) found that the observed increase in precipitation in southern Thailand may have contributed to higher yields in recent years. Therefore, we utilized monthly mean precipitation data spanning 1979-2021 from NOAA's PREC/L gridded precipitation data (Chen et al 2002).
Our choice to highlight precipitation over other climatic factors in our study does not exclude the potential impact of other meteorological variables such as temperature and solar radiation. As you rightly noted, these factors often influence crop yield in general. In fact, a study by Abubakar et al (2023) explored the correlation between these variables and palm oil production but found no significant relationship. We concur with their recommendation for further research in this area. Antidotal evidencebased on our interview with palm tree growers-and Okoro et al (2017) indicated varying effects of climate change on palm oil yield and emphasized the need to identify influential climate factors for accurate prediction.
Regarding large-scale climate variability, sea surface temperature (SST) data is vital for comprehending the link between large-scale climatic variability and regional precipitation affecting palm oil production. We utilized NOAA's Optimal Interpolated SST dataset to explore these patterns and their potential impact on Thailand's palm oil yields. Additionally, the ECMWF's ERA5 reanalysis dataset offers meteorological variables like cloud cover, precipitable water, and sea level pressure (SLP), enabling the assessment of global and regional climate factors influencing palm oil yield in southern Thailand.

Preliminary analysis
Our analysis commenced by examining the seasonality relationship between palm oil yield and regional precipitation. Figure 1(a) depicts the crosscorrelation of palm oil yield with 5 months moving averages of local precipitation south of 12 • N. Notably, the lag of −9 months (NDJFM) exhibits the highest correlation coefficient, with significant correlations ranging from 13 to 7 months before the harvest season. The hot weather and occasional highprecipitation events during this period play a critical role in the growth cycle of palm trees, setting the stage for fruit production in the following season. This specific focus on NDJFM precipitation stems from the unique physiology of palm trees.
While early-season precipitation serves as an empirical predictor, in the context of Thailand, there are several key challenges associated with seasonal prediction. These challenges involve capturing the complex interactions between regional climate systems, such as the monsoon dynamics, and largescale climate patterns like ENSO. The diverse topography and geographical features of Thailand add further intricacies to accurately predict local climate variables.
Given the strong dependency of palm oil yield on precipitation, it is essential to examine long-term climate variability. Thailand's monsoon rainfall and extreme events have links to decadal climate variability (Faikrua et al 2020, Pumijumnong et al 2020. Figure 1(b) illustrates annual precipitation in southern Thailand from 1950, revealing a stepwise change. Negative phases of the Pacific Decadal Oscillation coincide with increased annual precipitation in the region. Further investigation is required to anticipate potential climate regime changes that could reverse precipitation variability and palm oil yield. The reported increase in precipitation and palm oil production (Chiarawipa et al 2020) may be temporary or less stable than initially assumed.
We generated Pearson's correlation maps to illustrate connections between Thailand's palm oil productivity and large-scale SST and climatic variations. These maps display November-March means of SST, cloud cover, and precipitable water content anomalies correlated with 2003-2021 palm oil yield data. One-point correlation maps represent the statistical relationship between a variable like palm oil yield and spatially distributed meteorological fields, such as SST or atmospheric pressure. By calculating correlation coefficients between the target variable and each grid point, these maps identify regions with strong associations, providing valuable insights into potential climatic drivers influencing palm oil yield. Figure 2(a) displays a distinct La Niña condition, associated with increased palm oil yield in southern Thailand up to 9 months before the primary production season. This aligns with previous research linking La Niña to above-average rainfall in Southeast Asia, including Thailand and Malaysia. Figure 2(b) indicates significant correlations between cloud cover over the Maritime Continent and Southeast Asia and increased palm oil yield, while figure 2(c) confirms this observation with heightened precipitable water content. These maps highlight the unusually moist conditions preceding the subsequent year's rise in palm oil yield, attributed to La Niña during the NDJFM season. These findings underscore the influence of large-scale climatic modes, such as La Niña, on local meteorological factors like precipitation and cloud cover, impacting palm oil yield. Figure 2(d) exhibits the characteristic La Niña pattern with an intensified tropical Pacific east-west dipole, enhancing the Walker circulation. This finding aligns with Dong et al (2022), indicating that the atmospheric Rossby wave response in the troposphere, driven by tropical heating over the Tropical Western Pacific, contributes to the anomalous lowpressure system. The influence of tropical oceans and a reinforced Walker circulation, rather than extra tropical circulations, affects winter rainfall in southern Thailand. These insights are valuable for developing dynamic seasonal prediction models for palm oil yield and total production, enabling the industry to enhance its resilience and sustainability by preparing for climate variability and change.  Only values with the correlation coefficient that is significant at p < 0.1 are plotted. The area of Thailand locates in red circles and its southern region is a major oil palm plantation.

Discussions
In Thailand, palm oil production is closely tied to the seasonal monsoon period from May to October, which provides the necessary water for fruit bunch development. While year-round production practices and abundant rainfall during the monsoon season reduce sensitivity, the NDJFM period holds greater significance for palm tree growth due to less frequent but impactful precipitation events. These events align with peak ENSO development phases and overall climate variability in southern Thailand (Singhrattna et al 2005, Dong et al 2022. Further north, ENSO influences have been observed in the increased variability of the Chao Phraya River's peak season flow (Xu et al 2019). Hensawang et al (2021) also demonstrated the substantial impact of ENSO on rice yields in Thailand over an 8 months timeframe, highlighting the broad influence of ENSO on agricultural productivity.
Our analysis reveals ENSO as a consistent climate influencer for palm oil yield, as shown in supplemental figure S1, displaying the 11 years sliding correlations between palm oil yield and wintertime ENSO (Nino-3.4 index). This steady association, unlike palm oil yield's fluctuating relationship with the Pacific Meridional Mode (Amaya 2019), validates ENSO's potential as a stable predictor for palm oil yield. Recent advancements in dynamic seasonal prediction models (Kim et al 2018, White et al 2021 hold promise for palm oil yield prediction, enabling stakeholders to make informed decisions on crop management, optimizing production and bolstering the industry's resilience and sustainability amid changing climate conditions and growing demand. ENSO events significantly influence palm oil yield in Thailand, but a gap remains between known climatic influences and reliable seasonal prediction methods. Advancements in ENSO prediction, such as Zhou and Zhang (2023), offer opportunities to enhance palm oil production management in Southern Thailand. Future work should develop dynamic models for direct prediction of ENSO and Thailand's climate, incorporating local and largescale climate variables. Studies like Dong et al (2022) demonstrate the skillfulness of these models in predicting wintertime rainfall and seasonal precipitation, while decadal climate prediction in Southeast Asia shows promise (Kim et al 2018). Enhancing the resolution and accuracy of models, incorporating comprehensive observational data, and advancing our understanding of the underlying mechanisms driving seasonal climate variability are critical for improving the skill and reliability of seasonal prediction in Thailand. Integrating local meteorological factors with large-scale climate drivers poses a foreseeable challenge, as it requires capturing complex interactions that affect palm oil yield. By implementing such models, stakeholders can anticipate palm oil production, refine management strategies, and contribute to the long-term success and sustainability of Thailand's palm oil industry.

Moving forward
As global demand for palm oil grows, its economic importance and climate impacts on Southern Thailand's palm oil yields become more pronounced. Addressing these challenges requires dynamic seasonal climate prediction systems providing reliable information for stakeholders. Thailand's focus on research and development, including ENSO events in forecasting models, enhances accuracy in predicting rainfall and temperature patterns, benefiting oil palm-growing regions. This empowers farmers and plantation owners to make informed decisions on crop management, boosting palm oil production despite climate challenges, and leveraging innovations like hybrid palms that bear fruit 24 months after planting (Sequiño and Abocejo 2019).
Future research should focus on the development of dynamic models for direct prediction of ENSO and Thailand's climate. These models should integrate local meteorological factors, such as precipitation, temperature, and solar radiation, with large-scale climate variables like ENSO and monsoon patterns. Recent studies have demonstrated the effectiveness of these integrated models in predicting wintertime rainfall and seasonal precipitation patterns in Southeast Asia. Implementing such models will enable stakeholders to anticipate palm oil yield and production more accurately, improve management strategies, and contribute to the longterm success and sustainability of Thailand's palm oil industry.
In future work, it would be valuable to explore potential measures that can be adopted in advance to safeguard palm oil yield based on skillful seasonal or sub-seasonal prediction of precipitation and other climate factors. These measures could include implementing irrigation strategies, adjusting planting schedules, optimizing fertilization and nutrient management practices, and enhancing pest and disease management. By incorporating these countermeasures, the palm oil industry can proactively adapt to forecasted climate conditions and optimize yield outcomes.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).