Predictability of weather and climate

Abstract:

<p>Tropical cyclones (TCs) are some of the most dangerous natural hazards that human civilisation is exposed to. Effective adaptation for coastal regions requires reliable forecasts of risks for the season. Natural Hazard models such as the Synthetic Tropical cyclOne genRation Model (STORM) developed by Bloemendaal et al. (2020) are a common choice to assess risks without the expense of running a full forecast model. STORM has so far only been compared to observations on a basin-wide scale. However, for useful risk assessments in coastal regions, the model is also required to be skilful on much smaller spatial scales. We examine landfall statistics in some key areas such as the Gulf of Mexico.  Numerous indices for TC genesis have been developed over the past decades that aim to derive genesis locations from meteorological variables. None of the currently operational indices however is capable of realistically modelling interannual variability in genesis numbers and locations. Here, we compare the purely statistical Poisson interannual variability to that observed. Using Poisson regression between observations and driving environmental variables such as relative sea surface temperatures and wind shear, we then produce a new index for genesis location that has better predictive skill on interannual time scales.</p>

Abstract:

<p>The direct detection — or tracking — of tropical cyclones (TC) in gridded datasets outputs from reanalyses or model simulations is required to assess TC statistics. This issue has been tackled independently by many modeling centers or research groups; hence there is little homogeneity in the existing methods. The trackers – i.e., the algorithms used to perform that tracking -- generally fall into one of two categories: physics-based or dynamics-based. Physics-based trackers use sea-level pressure as their primary tracking variable, with additional warm-core and intensity criteria, whereas dynamics-based trackers use kinematic variables such as vorticity.</p><p>We compared four trackers taken from both categories and that we deem very different from one another in terms of their formulation: UZ (sometimes called TempestExtremes, Ullrich et al. 2021), OWZ (Tory et al. 2013), TRACK (Hodges et al. 2017) and CNRM (Chauvin et al. 2016). We assessed their performances by tracking TCs in ERA5 and comparing the outcome to the IBTrACS database – a collection of TC observations from several meteorological centers worldwide.</p><p>We find typical detection rates ranging from 70 to 80% and False Alarm (FA) rates ranging from 20 to 50% depending on the trackers. Based on the finding that a large proportion of these FAs are extra-tropical cyclones, we adapted an existing filtering method that relies on the relative positions of the detected tracks and the upper troposphere subtropical jet. When applied identically to the four trackers, it reduces FA rates to figures ranging from 9 to 30% while leaving detection rates unchanged.</p><p>Even though we were able to find most of the observed TCs in ERA5, we find, in agreement with several results in the recent literature, that their intensity is largely underestimated. However, and perhaps counterintuitively, there is no simple attenuation relationship between observed and reanalyzed TCs: for example, the strongest observed TCs are found in ERA5 with intensities covering almost the entire TC intensity scale.</p><p>We conclude by providing guidelines applicable when faced with the question of which tracker(s) to use depending on the research question. In particular, we show that using several trackers is not necessarily relevant for optimizing detection skills but combining them can be helpful to gain insight into different aspects of TCs in the same dataset.</p><p>Finally, we used the expertise gained above to track TCs in a set of HighResMIP simulations performed with the IPSL-CM7A model at different resolutions. In agreement with recent results, we find that the ability to simulate TCs improves significantly with resolution. Even though the intensity of simulated TCs remains too weak on average, the global statistics approach observations for simulations at a few tens of kilometers of horizontal resolution.</p>

Abstract:

Recent extreme weather across the globe highlights the need to understand the potential for more extreme events in the present-day, and how such events may change with global warming. We present a methodology for more efficiently sampling extremes in future climate projections. As a proof-of-concept, we examine the UK’s most recent set of national Climate Projections (UKCP18). UKCP18 includes a 15-member perturbed parameter ensemble (PPE) of coupled global simulations, providing a range of climate projections incorporating uncertainty in both internal variability and forced response. However, this ensemble is too small to adequately sample extremes with very high return periods, which are of interest to policy-makers and adaptation planners. To better understand the statistics of these events, we use distributed computing to run three 1000-member initial-condition ensembles with the atmosphere-only HadAM4 model at 60km resolution on volunteers’ computers, taking boundary conditions from three distinct future extreme winters within the UKCP18 ensemble. We find that the magnitude of each winter extreme is captured within our ensembles, and that two of the three ensembles are conditioned towards producing extremes by the boundary conditions. Our ensembles contain several extremes that would only be expected to be sampled by a UKCP18 PPE of over 500 members, which would be prohibitively expensive with current supercomputing resource. The most extreme winters we simulate exceed those within UKCP18 by 0.85 K and 37% of the present-day average for UK winter means of daily maximum temperature and precipitation respectively. As such, our ensembles contain a rich set of multivariate, spatio-temporally and physically coherent samples of extreme winters with wide-ranging potential applications.