Confidence and Uncertainty in Simulating Tropical Cyclone Long-Term Variability Using the CMIP6-HighResMIP

Abstract

Abstract Confidence and uncertainty issues of simulations were seldom evaluated in previous studies although the climate models are widely used. This study evaluates the performance of the CMIP6-HighResMIP simulations in presenting long-term variability of tropical cyclone (TC) genesis frequency (TCGF) and track density (TCTD) and quantifies the relative contributions of internal and external forcing to TC activities during the 1950–2014. There is overall poor model performance in simulating long-term changes in TC activities over the Northern Hemisphere, including interdecadal variabilities and long-term linear trends. The simulated long-term changes in TCGF and TCTD over the eastern North Pacific (ENP) in six high-resolution models show opposite characteristics to the observations. Moreover, most models cannot capture the variabilities of TCGF and TCTD over the western part of the western North Pacific (WNP) and northern part of the North Atlantic (NA). However, these models show a high degree of confidence in reproducing the interdecadal variabilities and linear trends of TCGF and TCTD over the eastern part of the WNP and the tropical NA. Quantitative evaluations further show that there are the opposite relative contributions of long-term climate variabilities to TCGF and TCTD changes over the ENP between the observations and the multimodel ensemble mean, followed by large model biases over the western WNP and the northern NA, but relatively consistent contributions over the southern NA and the Caribbean. These results help us cope with contrasting and consistent future TC changes among the model projections. Significance Statement While climate models have been widely used to project future changes in tropical cyclone (TC) activity, few studies have examined to what extent we can trust these model projections. We used the CMIP6-HighResMIP simulations to quantify the model biases in presenting TC activity, and evaluate the relative contributions of internal and external forcing to TC activities. In general, the HighResMIP has large discrepancies in representing longer-term climate variability of TC activity. However, the models can capture well TC activity over the eastern part of the western North Pacific and tropical Atlantic, which is attributed to good performance of models in reproducing the relationship between long-term climate variabilities beyond interannual scale and TC activity. These results highlight confidence and uncertainty in future TC changes among the model projections.