Figure: Comparison of the SPEI drought index values for the observed 1993–2022, 3-month cumulative D values: Panel (a) SPEI based on 1981–2010 observations, Panel (b) SPEI based on LOCA2 projected 2031–2060 conditions, and Panel (c) SPEI based on WG-projected 2031–2060 conditions. An SPEI ≤−1.5 represents severe drought conditions. According to the observed 1981–2010 climate description in Panel (a), severe drought conditions occurred 17 times from 1993 to 2022 with a minimum calculated SPEI of −2.1. Using LOCA2 2031–2060 conditions, severe drought occurred four times between 1993 and 2022, with a minimum SPEI of −1.9. For the 1993–2022 observations analyzed using WG 2031–2060 conditions, no severe droughts occurred, and the minimum SPEI was −1.4. This comparison demonstrates that the WG 2031–2060 climate description is significantly warmer and drier on average than the LOCA2 2031–2060 description.
Climate change is reshaping our world in ways we are just beginning to understand. One of the most pressing challenges is predicting how these changes will impact water resources, crucial for agriculture, industry, and daily life. The concept of weather attribution, which examines the likelihood of specific weather events occurring under different climate conditions, has emerged as a vital tool in this effort. By understanding the role of human-induced climate change in altering weather patterns, scientists can better predict and prepare for the future. This approach is particularly relevant for managing water resources, as it helps forecast conditions like droughts that can have severe economic and environmental impacts.
Weather attribution has become increasingly vital as the effects of climate change intensify. Recent research led by Nick Martin, formerly of Southwest Research Institute in San Antonio, Texas, explores how incorporating weather attribution into water budget projections can enhance our understanding of future drought conditions. This work, published in the journal Hydrology, compares expectations for future severe drought among historical observations, LOCA2-downscaled CMIP6 future climate simulation results, and weather attribution-guided statistically projected future climate. Stochastic weather generators (WG) are the statistical simulation tool used to predict weather attribution constrained future climate.
Weather attribution estimates the likelihood of observed weather events occurring under different climate scenarios and thus the change in likelihood for occurrence of severe drought under human induced climate change. The weather attribution study employed to guide statistical projection of future climate for this work suggests that severe three-month drought is five times more likely to occur given human induced climate change. Conceptually, five times more likely means that a 1 in 25-year drought in 2000 is now a 1 in 5-year drought in 2020s. The WG produced synthetic future climate is constrained, or “calibrated”, to produce five times more likely severe drought during 2031–2060. This method simulates future weather patterns, including droughts, in a way that reflects historical data, recently observed weather, and anticipated future climate changes.
“Weather attribution provides observed change in likelihood for extreme events, including drought, that is required to assess, plan, and prepare mitigation for future risk to water resources from human-induced climate change. Once the change in likelihood is attributed, synthetic statistical projections of future climate, embodying the new extreme event likelihood, provide a framework for water resources planning and risk assessment,” said Martin, highlighting the potential of this approach to provide water budget forecasts that describe inherent uncertainty in, and risk related to, future conditions.
The implementation site was the Frio River basin in south-central Texas, an area crucial for water resource management due to its direct communication between surface water and the Edwards Aquifer. A WG was calibrated to synthetically produce stochastic weather across 2031–2060 that provides a climate description where severe three-month drought is five times more likely to occur relative to historical observations. This enhanced drought likelihood is based on expectations for significantly higher temperatures and reduced soil moisture in the future compared to historical norms. Expectations for increased temperature and decreased soil moisture are supported by CMIP6 future climate simulation results and weather attribution studies based on recently observed weather.
In this study, magnitude and likelihood of three-month drought is described using the Standardized Precipitation Evapotranspiration Index (SPEI). SPEI is based on precipitation and temperature data and provides a climatic drought index that is sensitive to global warming. The observed three-month water deficit (D), calculated as precipitation depth less potential evapotranspiration depth, is the drought measurement that is transformed, standardized, and normalized to generate the SPEI. The “standardization” part provides the likelihood, or probability, for three-month drought magnitudes based on the precipitation and temperature data set used to calculate the SPEI. Drought categories by SPEI value range and cumulative probabilities for select SPEI values are shown in the table below.
The significant increase in the probability of drought conditions observed for recent extreme events is the critical factor guiding water resource planning. The difference in likelihood for observed January 2000 three-month drought, shown on the figure above, identifies diverging expectations among historical conditions, LOCA2-downscaled CMIP6 future climate simulation results, and weather attribution-guided WG projected future climate. The observed three-month water deficit (D) for January 2000 is -217 mm. When calculated from 1981–2010 observations, SPEI for -217 mm is -1.9 with a cumulative probability of 0.03 corresponding to severe drought. When determined using weather attribution constrained WG projections for 2031–2060, the SPEI for D of -217 mm is -0.9 with a cumulative probability of 0.17 corresponding to mild drought. This identifies that a three-month D of -217 mm for November, December, and January is 5.7 (0.17 / 0.03 = 5.67) times more likely to occur in the WG projected climate for 2031–2060 than in observed climate during 1981–2010. When calculated using LOCA2-downscaled CMIP6 climate simulation results across 2031–2060, the SPEI for D of -217 mm is -1.6 with a cumulative probability of 0.05 denoting severe drought. Historically observed severe drought (November, December, and January D of -217 mm) is 3.4 (0.17 / 0.05 = 3.4) times more likely to occur in WG projected climate than in LOCA2-downscaled CMIP6 climate simulation results from 2031–2060.
The significance of these findings lies in their potential applications for water resource management and planning. By providing an enhanced description of likelihood of future extreme events, the study’s methodology can inform strategies for water conservation and allocation, helping to mitigate the impacts of severe droughts. This approach can be extended to other regions and water systems, offering a valuable tool for addressing the challenges and risks posed by climate change.
In summary, this study demonstrates the critical role of weather attribution in enhancing the characterization of uncertainty in future water budget projections. The findings underscore the need for innovative approaches in water resource management, particularly as climate change continues to alter the frequency and intensity of extreme weather events. As Martin concluded, the ability to predict and prepare for severe droughts is essential for sustainable water management and the resilience of communities dependent on these vital resources.
Journal Reference
Martin, Nick. “Incorporating Weather Attribution to Future Water Budget Projections.” Hydrology, 2023, 10, 219. DOI: https://doi.org/10.3390/hydrology10120219
About the Author
Nick Martin is a water scientist with vodanube llc and RESPEC based in Fort Collins, CO. His background is as a surface water and groundwater hydrologist and software developer. He focuses on risk assessment, risk mitigation, reliability, resiliency, and sustainability analyses related to climate change and legacy infrastructure on natural and engineered systems. Nick specializes in probabilistic analysis and modeling to quantify uncertainty and define environmental and economic risk. His technical interests include uncertainty analysis for decision support and data assimilation as part of water movement, transport modeling, machine learning, and deep learning studies.
