Keeping Us Honest: Examining Climate States and Transition Probabilities of Precipitation Projections in General Circulation Models

Panaou, Toni; Asefa, Tirusew; Nachabe, Mahmood

doi:10.1061/(asce)wr.1943-5452.0000910

Cited by 6 publications

(1 citation statement)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, the fidelity of GCMs in representing precipitation is vital to ensuring that climate impact assessments and projections are built on a reliable foundation (Chang et al, 2018; Hwang and Graham, 2013). A higher degree of accuracy in precipitation simulation is essential for reliable climate impact assessments and the development of effective water management strategies to mitigate potential impacts on regional and local water resources (Panaou et al, 2018). Moreover, precipitation simulations are essential for projecting potential changes in extreme events such as droughts and floods, which have significant societal and economic consequences (Bao et al, 2017; Srivastava et al, 2020; Wehner et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Enhanced performance of CMIP6 climate models in simulating historical precipitation in the Florida Peninsula

Wang,

Asefa

2024

Intl Journal of Climatology

Self Cite

View full text Add to dashboard Cite

Assessing the performance of General Circulation Models (GCMs) in simulating historical regional precipitation is essential for climate assessments. This study analysed 18 GCMs from the Coupled Model Intercomparison Project Phase 5 (CMIP5) and 27 GCMs from Phase 6 (CMIP6) for the Florida Peninsula. Naïve combinations formed the CMIP5‐Ensemble and CMIP6‐Ensemble. The reference dataset consisted of global gridded precipitation data from National Oceanic and Atmospheric Administration. GCM‐simulated precipitation data were compared to the reference dataset across multiple time scales between 1951 and 2005. Most CMIP5 and CMIP6 GCMs exhibited a negative bias at the daily time scale, with an absolute value ranging between 0 and 2 mm day−1 at the median level, except for INM‐CM4 and MRI‐CGCM3 (CMIP5) and CNRM‐ESM2‐1 (CMIP6). Two performance metrics, that is, Pearson correlation and mean absolute error (MAE), were used to evaluate the GCMs. A correlation value above 0.3 is statistically significant at the significance level (α = 0.05). For the Pearson correlation between simulated and observed monthly precipitation, only 15% of CMIP5 GCMs (3 out of 19) had a median value above 0.3, whilst this increased to approximately 39% (11 out of 28) for CMIP6 GCMs. For monthly climatological precipitation, only 21% (4 out of 21) of GCMs in CMIP5 showed a correlation with a median value of 0.8 or higher. This percentage rose to 50% (14 out of 28) for CMIP6. A notable difference was observed in the MAE between CMIP5 and CMIP6 climate models. Lastly, GCM raw outputs were evaluated based on rainy season characteristics (onset, demise and duration). Significant improvements from the CMIP5 to the CMIP6 models were observed in capturing the rainy season in the Florida Peninsula. This study thoroughly compares CMIP5 and CMIP6 climate models in simulating historical precipitation at various time scales for the study region. Although the results are specific to the study area, the methods can be applied more broadly.

show abstract