By God-Frey Jacinthe
Executive summary
Haiti is among the countries most vulnerable to climate disasters. Over the past 30 years, hurricanes and floods have caused thousands of deaths and significant economic losses, slowing down the country ' s development. Unfortunately, projections by the Intergovernmental Panel on Climate Change (IPCC) suggest that these extreme events are worsening. However, global climate models present significant inaccuracies at the scale of Haiti, limiting their direct use for adaptation planning. This study corrected the biases of nine IPCC climate models for the historical period 1920-1940 and then used six corrected models to analyze precipitation trends up to 2100. The results show a general decrease in annual precipitation (up to 48% in some regions), particularly marked during rainy seasons, with major implications for agriculture, water management and sustainable development planning in Haiti.
Introduction : Haiti, a country on the front lines of climate change
Climate change is one of the major challenges of our time, profoundly changing climate systems around the world. Successive IPCC reports have documented irrefutable evidence of global warming, changes in precipitation patterns and increased frequency of extreme events (IPCC, 2014). In the Caribbean, these changes are already reflected in rising temperatures and changes in rainfall patterns, with dramatic consequences for island populations.
Haiti occupies a particularly worrying position in this context. The Climate Risk Index ranks the country third among the nations most affected by climate events between 1996 and 2015, with an annual average of 254 deaths and economic losses accounting for nearly 1.5% of GDP (Kreft et al., 2017). Recent disasters illustrate this vulnerability: in 2004, tropical storm Jeanne killed more than 3000 people and caused damage equivalent to 7% of GDP. In 2008, four successive cyclones devastated the country, killing 793 people and destroying 14.6% of GDP. More recently, Hurricane Matthew in 2016 has claimed more than 500 lives and caused losses in excess of $1 billion.
In the face of this alarming situation, adaptation planning requires reliable climate projections. However, the IPCC's global climate models, although essential for understanding future climate change, present too coarse spatial resolutions (100 to 300 km) and systematic biases that limit their direct use on a small country scale such as Haiti (27,750 km2). The aim of this study is therefore to correct these biases in the IPCC climate scenarios (CMIP6) and to analyse the future evolution of precipitation in Haiti until 2100, thus providing essential data for decision makers and planners.
Methodology: correcting models to better predict
This research took place in two main phases. The first phase involved assessing and correcting the biases of global climate models for Haiti. We used nine CMIP6 models (the sixth phase of the IPCC coupled model comparison project) and compared them with 12 historical rainfall data sets from the Simbi database, covering the period 1920-1940 (Bathelemy et al., 2024).
Two statistical methods of bias correction were evaluated: the scaling method (scaling) and the quantile mapping method (Quantile Mapping). These approaches aim to adjust model simulations to better match historical observations. After evaluating their performance over a validation period (1935-1940), the scaling method was found to be the most effective and was used to correct future projections.
In the second phase, we applied this correction method to six global climate models to analyze the evolution of precipitation from 2025 to 2100, according to the most pessimistic scenario (SSP5-8.5, without mitigation policy). Three climate indices were calculated: total annual precipitation (PRCPTOT), number of rainy days (RR1), and mean precipitation intensity (SDII). In addition, we used the average of 6 corrected global climate models (GCMs) to estimate relative changes in monthly, annual, and seasonal precipitation for the near future (2025-2050), semi-Lintain (2050-2070), distant (2070-2100), and for the period (2025-2100) compared to the reference period (1920-1940). This was done on the three rush indices used. Finally, the linear model was used to estimate trends in precipitation climatic indices for each station on an annual and seasonal scale over the periods: (2025-2050), (2050-2070), (2070-2100) and 2025-2100.
Results: A Drier Future for Haiti Models generally underestimate Haitian precipitation
The assessment of global climate models (GCMs) reveals significant biases in their ability to reproduce Haiti's historical precipitation. The majority of models underestimate monthly precipitation from March to October, with mean differences between -2 and -4 mm per day. However, some models such as CNRM-ESM2-1 significantly overestimate precipitation, particularly during the dry season, with biases of up to 160% in some stations. These wet and dry biases also vary according to the seasons: in winter (December-January-February), several models overestimate precipitation, while they systematically underestimate it during rainy seasons.
This assessment confirms the absolute need to correct these biases before using the models for impact studies. The two methods tested significantly reduced these Biais, although some stations were not fully corrected for some models. The scaling method was found to be slightly superior, completely correcting the biases in four models against three for the Quantile Mapping method.
A general decrease in precipitation by 2100
Corrected projections indicate a worrying trend: all sites surveyed will show a decrease in total annual precipitation over the period 2025-2100. This decline will be between -4 and -11 mm per year depending on the region, with the largest decreases in the departments of West, Artibonite, Central and South. Station P_104, located in the West Department, will show the largest decrease with -11.6 mm per year.
This decrease in precipitation will be accompanied by a reduction in the number of rainy days throughout the country. Paradoxically, the average rainfall intensity will also decrease at almost all stations, with the notable exception of the Southern Department, which will experience a slight increase of 0.002 mm per day. This suggests a future rainfall regime characterized by less rain, distributed over less days, with generally lower intensities.
Contrast seasonal variations
Seasonal analysis reveals complex patterns. Rainy seasons (spring and autumn) will be particularly affected by the decrease in precipitation, with decreases of -6.27% for some stations in summer. On the other hand, the winter dry season (December-January-February) will experience an upward trend in precipitation at most stations, with the exception of two stations (P_091 and P_100) that will continue to dry.
The departments of the West, Artibonite and the South will be most affected by the fall in summer and spring precipitation. This seasonal redistribution of rains will have major implications for Haitian agriculture, traditionally dependent on the two rainy seasons for crops.
Dramatic changes from the current climate
Compared to the reference period 1920-1940, the period 2025-2100 will be marked by considerable changes. Annual rainfalls will fall by 5 to 48% in different departments, with the Centre most affected (-48%) and the South least affected (-5%). This reduction will be accompanied by a significant increase in the number of rainy days in some regions (up to 129% in West and Artibonite), but with a significant decrease in precipitation intensity (63 to 69% decrease).
On a monthly scale, the projections indicate decreases in precipitation from January to November, particularly marked in July (up to -66% for the end of the century) and April (-45% to -53%). Only in December will there be a significant increase in precipitation, with increases of 52 to 66% depending on the future periods considered.
Conclusion and perspectives: critical data for adaptation
This study is the first systematic correction of biases in the global climate models of the 6th phase of the IPCC's intercomparison of coupled models (CMIP6) project for Haiti. It provides high resolution climate projections (5 km × 5 km) essential for climate change adaptation planning. The results confirm the fears: Haiti will face a drier climate by 2100, with potentially important consequences for agriculture, water resource management, hydroelectric power generation and food security.
Projected decreases in precipitation, particularly during the rainy seasons crucial for agriculture, require a redesign of development strategies. The departments of the West, Artibonite and Centre, already facing major water challenges, will require priority investments in water conservation infrastructure and drought-friendly agricultural techniques.
However, this study has some limitations. The reference period used (1920-1940) is shorter than the generally recommended thirty years for bias correction. Moreover, the statistical methods used do not fully capture the complex physical processes influencing precipitation in the mountainous relief of Haiti.
Research perspectives include the use of dynamic disaggregation methods (regional climate models) to better represent topographic effects, the extension of analysis to other climatic variables such as temperature and evapotranspiration, and the characterization of sources of uncertainty in projections. It would also be crucial to extend this analysis to the extreme events (extended droughts, torrential rains) that constitute the most immediate threats to the Haitian people.
Meanwhile, the corrected data produced by this research are now available to the Haitian scientific community and decision makers. In a context where hydrometeorological data remain scarce and often inaccessible in Haiti, these projections are a valuable tool for developing adaptation policies based on robust scientific data. In the face of the climate emergency, investing in understanding and anticipation of future change is no longer an option, but a vital necessity for Haiti's resilience.
This work was carried out within the framework of the project CLIMEXHA of the Research Unit in Geosciences (URGéo) of the State University of Haiti, thanks to funding from the Research Fund for Development of the Bank of the Republic of Haiti (FRD-BRH).
References
Bathelemy, R., Brigode, P., Andréasian, V., Perrin, C., Moron, V., Gaucherel, C., Tric, E., & Boisson, D. (2024). Simbi:historical hydro-meteorological time series and signatures for 24 attacks in Haiti. Earth System Science Data, 16(4), 2073-2098.
IPCC (2014). Climate change 2014: Synthesis report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.
Kreft, S., Eckstein, D., & Melchior, I. (2017). Global Climate Risk Index 2017. Who Suffers Most From Extreme Weather Events? Germanwatch.
Moron, V., Frelat, R., Jean-Jeune, P. K., & Gaucherel, C. (2015). Interannual and intra-annual variability of rainfall in Haiti (1905–2005). Climate Dynamics, 45(3-4), 915-932.
Taylor, M.A., Centella, A., Charlery, J., Bezanilla, A., & Campbell, J.D. (2015). Haiti: Climate change – historical data and future projections. Capacity-building programme for climate change in the Caribbean.
Terrier, M., Rançon, J.-P., Bertil, D., Oak, F., Desprats, J.-F., Lecacheux, S., Le Roy, S., Stollsteiner, P., & Bouc, O. (2017). Atlas of natural threats in Haiti. Bureau of Geological and Mining Research (BRGM).
God-Frey JACINTHE
Geosciences Research Unit (URGeo)
LMI-CARIBACT
State University of Haiti
jacinthegod@gmail.com
