Figure 3 - available via license: Creative Commons Attribution 3.0 Unported
Content may be subject to copyright.
An example two-dimensional k-d tree (k = 2) built from nodes a through h. Dividing planes are constructed by cycling through each coordinate and determining the median node (left). This gives rise to a tree structure (right) that, in conjunction with an input node, can then be searched recursively for a corresponding rectangular domain in physical space. The last leaf node is labeled as the best candidate for nearest neighbor and the tree is " unwound " to test other potential candidates. The number of nodes that need to be examined is limited to domains that overlap a hypersphere with origin at the input node and with distance to the current candidate.
Source publication
This paper describes a new open-source software framework for automated
pointwise feature tracking that is applicable to a wide array of climate
datasets using either structured or unstructured grids. Common climatological
pointwise features include tropical cyclones, extratropical cyclones and
tropical easterly waves. To enable support for a wide...
Similar publications
The rainfall prediction accuracy of seven state-of-the-art operational models during landfall (LF) of tropical cyclones formed in the North Indian Ocean was assessed. The rainfall predicted by these models 1-3 days (24-72 h) before the LF day of tropical cyclones was compared with TRMM_3B42 rainfall data. Cyclone induced rainfall occurs along its t...
In order to understand the current and potential use of ensemble forecasts in operational tropical cyclone (TC) forecasting, a questionnaire on the use of dynamic ensembles was conducted at operational TC forecast centers across the world, in association with the World Meteorological Organisation (WMO) High-Impact Weather Project (HIWeather). The r...
Over the past few decades, the world has seen substantial tropical cyclone (TC) damages, with the 2017 Hurricanes Harvey, Irma and Maria entering the top-5 costliest Atlantic hurricanes ever. Calculating TC risk at a global scale, however, has proven difficult given the limited temporal and spatial information on TCs across much of the global coast...
Forecasting the trajectory and intensity of tropical cyclones (TCs) is important in disaster mitigation as TC usually causes huge damages. However, it remains a substantial challenge due to the limited understanding of TC complexity. Still, TCs have been observed and recorded for several decades, and so can be predicted if viewed as a spatial‐tempo...
Tropical cyclones (TCs) constitute one of the major types of meteorological disasters in China, causing extensive damage. Based on the latest quality‐controlled provincial‐level TC disaster losses data over China from 2001 to 2020, the spatiotemporal variation in TC disaster and its link with TC activity in the form of prevailing track, precipitati...
Citations
... The distribution of MTA occurrence across latitude is clearly bimodal with one narrow peak around 10 • latitude as well as a wider peak centered around 45 • latitude ( Fig. 4a-d). The minimum in frequency of detections occurs around 20 • latitude, close to the 15 • and 23.25 • -cutoffs used in the atmospheric river detection schemes of Ullrich and Zarzycki (2017) and Shearer et al. (2020), respectively. ...
... Most detection algorithms for atmospheric rivers use IVT as one of their criteria to define the feature. Typical thresholds used are 250, 500 and 700 kg s −1 m −1 (e.g., CONNECT by Sellars et al. (2015), TempestExtremes by Ullrich and Zarzycki (2017), and the Rutz et al. (2014) algorithm). In comparison, the most common IVT value along moisture transport axes is around 350 kg s −1 m −1 (Fig. 4b). ...
The water vapor transport in the extratropics is mainly organized in narrow elongated filaments. These filaments are referred to with a variety of names depending on the contexts, for example atmospheric river, warm moist intrusion, warm conveyor belt, and feeder air stream. Despite the various names, these features share essential properties, such as their narrow elongated structure. Here, we propose an algorithm that detects these various lines of moisture transport in instantaneous maps of the vertically integrated water vapor transport. The detection algorithm extracts well-defined maxima in the water vapor transport and connects them to lines that we refer to as moisture transport axes. By only requiring a well-defined maximum in the vapor transport, we avoid imposing a threshold in the absolute magnitude of this transport or the total column water vapor. Consequently, the algorithm is able to pick up moisture transport axes at all latitudes without requiring region-specific tuning or normalization. We demonstrate that the algorithm can detect both atmospheric rivers and warm moist intrusions, but also prominent monsoon air streams as well as low-level jets with moisture transport. Atmospheric rivers sometimes consist of several distinct moisture transport axes, indicating the merging of several moisture filaments into one atmospheric river. We showcase the synoptic situations and precipitation patterns associated with the occurrence of the identified moisture transport axes in example regions in the low, mid, and high latitudes.
... To detect the TC track in AGCM, we used TempestExtremes (Ullrich and Zarzycki 2017) and the criteria of warm core (temperature anomaly at 300 hPa showed increase of 0.4 °C within 8° of the candidate node) and sea level pressure (SLP) minimum (2 hPa within 4° of the candidate node), following a previous study (Sect. 3.1 of Ullrich and Zarzycki 2017). ...
In this study, we examine the long-term variability in the western North Pacific (Philippine Sea) summer monsoon (WNPSM) from the 1880s to the present. To determine the sea surface temperature (SST)-induced atmospheric variability, atmospheric general circulation model (AGCM) ensemble simulations are conducted with the observed SST. For the AGCM experiments, we use two convection schemes and SST datasets to examine the uncertainty in the convective parameterization and robustness of the simulated atmospheric response. In the AGCM simulation (T42; approximately 300-km horizontal resolution, 3-member ensemble in each), interdecadal variability can be seen as a cyclonic anomaly in the 1890s–1930s. This cyclonic anomaly is robust to different convection schemes and SST datasets. The observed rainfall over the land demonstrates consistent features. In our AGCM ensemble simulations, sensitivity experiments with several SST patterns show that the interbasin effect (i.e., contribution of tropical Indian Ocean and Atlantic) played a key role in the WNPSM cyclonic anomaly during the 1890s–1930s. Further, idealized SST experiments show that reduction in SST over the tropical Indian Ocean and Atlantic causes the WNPSM cyclonic anomaly. Additionally, using high-resolution AGCM experiments (T119; approximately 100-km horizontal resolution, 4-member ensemble in each), we investigate the increase in tropical cyclone (TC) activity over the western North Pacific during the 1890s–1930s. However, variability of the TC activity is sensitive to the convective parameterization.
... For a majority of the ENSO performance metrics, model error and inter-model spread are substantially larger than observational uncertainty ( Fig. 3an). This highlights the systematic biases, like the double Intertropical Convergence Zone (ITCZ) (Fig. 3a), that are persisting through CMIP phases (Tian and Dong, 2020). Similarly, ENSO process metrics (Fig. 3t-w) indicate large errors in the feedback loops generating SST anomalies, indicating a different balance of processes in the model and in the reference and possibly compensating errors (Bayr et al., 2019;Guilyardi et al., 2020). ...
... Another poten- tial avenue for the PMP involves leveraging machine learning (ML) techniques and other state-of-the-art data science techniques being used for process-oriented ESM evaluation works (e.g., Nowack et al., 2020;Labe and Barnes, 2022;Dalelane et al., 2023). Applications of ML detection, such as for storms, using TempestExtremes (Ullrich and Zarzycki 2017;Ullrich et al., 2021), and fronts (e.g., Biard and Kunkel, 2019), can enable additional specialized storm metrics for high-resolution simulations. For convection-permitting mod-els, yet more storm metrics can be applied such as mesoscale convective systems. ...
Systematic, routine, and comprehensive evaluation of Earth system models (ESMs) facilitates benchmarking improvement across model generations and identifying the strengths and weaknesses of different model configurations. By gauging the consistency between models and observations, this endeavor is becoming increasingly necessary to objectively synthesize the thousands of simulations contributed to the Coupled Model Intercomparison Project (CMIP) to date. The Program for Climate Model Diagnosis and Intercomparison (PCMDI) Metrics Package (PMP) is an open-source Python software package that provides quick-look objective comparisons of ESMs with one another and with observations. The comparisons include metrics of large-to global-scale climatologies, tropical inter-annual and intra-seasonal variability modes such as the El Niño-Southern Oscillation (ENSO) and Madden-Julian Oscillation (MJO), extratropical modes of variability, regional monsoons, cloud radiative feedbacks, and high-frequency characteristics of simulated precipitation, including its extremes. The PMP comparison results are produced using all model simulations contributed to CMIP6 and earlier CMIP phases. An important objective of the PMP is to document the performance of ESMs participating in the recent phases of CMIP, together with providing version-controlled information for all datasets, software packages, and analysis codes being used in the evaluation process. Among other purposes, this also enables modeling groups to assess performance changes during the ESM development cycle in the context of the error distribution of the multi-model ensemble. Quantitative model evaluation provided by the PMP can assist modelers in their development priorities. In this paper, we provide an overview of the PMP, including its latest capabilities, and discuss its future direction.
... Our research goal was to identify short-term-drought and heatstress events, as defined below. To this end, we used the objectoriented analysis program Tempest Extremes (TE; Ullrich and Zarzycki, 2017). As implied by its name, TE was developed to track features such as tropical and extratropical cyclones, as well as tropical easterly waves. ...
... TE is sufficiently efficient ) that we can extract a large collection of events from data sets covering multiple decades in a sizeable region, thereby allowing us to develop a climatology of the targeted events. Details of TE's formulation and coding appear in Ullrich and Zarzycki (2017). An advantage of the approach is that it allows one to diagnose events based on regional climate characteristics that potentially can be defined by stakeholder-relevant impactful events (Sillmann et al., 2021;Fisel et al., 2023) rather than more standard hydroclimate metrics, such as return periods for extreme events or percentile exceedances. ...
Introduction
Crops are vulnerable to precipitation and heat extremes during late spring through summer.
Methods
We analyzed for a north-central U.S. region short-term drought and agricultural heat stress during April-May-June-July. We used the 4-km Parameter Elevation Regression on Independent Slopes Model (PRISM) for observations, aggregated to a 25-km grid, and two 25-km Regional Climate Model version 4 (RegCM4) simulns used either GFDL- or MPI-GCM boundary conditions. We chose 1981-2000 as our contemporary time period, and 2041-2060 as our scenario time period, which used the Representative Concentration Pathway 8.5 emissions scenario. We used object-oriented analysis to identify events of interest in observations and simulations by identifying objects in a space-time domain that meet specified criteria, such as exceeding a heat-stress temperature threshold. The event diagnosis allowed analysis of compound events, occurring when temperature and drought objects overlap.
Results
Identified objects yielded events that can undermine agricultural productivity and which are thus relevant to decision makers, making them building blocks for possible climate storylines. The observations and simulations showed similar spatial distributions of event frequencies across the analysis region. However, the simulations attained this distribution by having fewer events that tend to cover larger areas compared to observed events, suggesting that the effective resolution of the simulations was coarser than their 25-km grids. Short-term drought frequency increased and heat-stress frequency decreased in transitioning to the scenario climate. When compounding occurred heat-stress events generally preceded the short-term drought events. The overlapping, compound events tended to be more extreme compared to non-overlapping events of either type.
Discussion
The information yielded projected changes in these agriculturally motivated events. One prominent conditional behavior emerging from the work was that a heat-stress event should be a warning to watch for potential drought, as both could compound each other to more intense levels.
... Tracks from the "hist-1950" simulations covering the 36-year period 1979-2014, and tracks from the "highres-future" simulations covering the 36-year period 2015-2050 are used. The TC tracks from HighResMIP that we use in this study are based on TempestExtremes (Ullrich & Zarzycki, 2017), a scale-aware feature tracking software that operates on the model's native grid. Projections of TC activity using TempestExtremes, which tracks TC vortices based on sea-level pressure anomalies, are broadly consistent with those using a tracker that operates on vorticity anomalies , suggesting that our results may not be overly sensitive to the choice of the TC tracking algorithm. ...
Tropical Cyclones (TCs) inflict substantial coastal damages, making it pertinent to understand changing storm characteristics in the important nearshore region. Past work examined several aspects of TCs relevant for impacts in coastal regions. However, few studies explored nearshore storm intensification and its response to climate change at the global scale. Here, we address this using a suite of observations and numerical model simulations. Over the historical period 1979–2020, observations reveal a global mean TC intensification rate increase of about 3 kt per 24‐hr in regions close to the coast. Analysis of the observed large‐scale environment shows that stronger decreases in vertical wind shear and larger increases in relative humidity relative to the open oceans are responsible. Further, high‐resolution climate model simulations suggest that nearshore TC intensification will continue to rise under global warming. Idealized numerical experiments with an intermediate complexity model reveal that decreasing shear near coastlines, driven by amplified warming in the upper troposphere and changes in heating patterns, is the major pathway for these projected increases in nearshore TC intensification.
... The number of ARs days over the NEUS region is thereby projected to increase as atmospheric moisture increases with warming climate, leading to more frequent pure AR-EP days. It is worth noting that alternative AR detection algorithms, such as Tempest (Ullrich & Zarzycki, 2017), which uses time-dependent IVT thresholds, have been employed in previous studies. These alternative methods have highlighted that changes in AR frequency can be predominantly influenced by a meridional shift in storm tracks. ...
The Northeast United States (NEUS) has faced the most rapidly increasing occurrences of extreme precipitation within the US in the past few decades. Understanding the physics leading to long‐term trends in regional extreme precipitation is essential but the progress is limited partially by the horizontal resolution of climate models. The latest fully coupled 25‐km GFDL (Geophysical Fluid Dynamics Laboratory) SPEAR (Seamless system for Prediction and EArth system Research) simulations provide a good opportunity to study changes in regional extreme precipitation and the relevant physical processes. Here, we focus on the contributions of changes in synoptic‐scale events, including atmospheric rivers (AR) and tropical cyclone (TC)‐related events, to the trend of extreme precipitation in the fall season over the Northeast US in both the recent past and future projections using the 25‐km GFDL‐SPEAR. In observations, increasing extreme precipitation over the NEUS since the 1990s is mainly linked to TC‐related events, especially those undergoing extratropical transitions. In the future, both AR‐related and TC‐related extreme precipitation over the NEUS are projected to increase, even though the numbers of TCs in the North Atlantic are projected to decrease in the SPEAR simulations using the SSP5‐8.5 projection of future radiative forcing. Factors such as enhancing TC intensity, strengthening TC‐related precipitation, and/or westward shift in Atlantic TC tracks may offset the influence of declining Atlantic TC numbers in the model projections, leading to more frequent TC‐related extreme precipitation over the NEUS.
... For instance, TC trackers operationally used at ECMWF, NCEP and Met Office have been respectively described in Van der Grijn (2002), Marchok (2002Marchok ( , 2021 and Heming (2017). Other trackers are reviewed in Ullrich and Zarzycki (2017) and Bourdin et al. (2022) . ...
Detection and tracking of Tropical Cyclones (TCs) in numerical weather prediction model outputs is essential for many applications, such as forecast guidance and real-time monitoring of events. While this task has been automated in the 90s with heuristic models, relying on a set of empirical rules and thresholds, the recent success of machine learning methods to detect objects in images opens new perspectives. This paper introduces and evaluates the capacity of a convolutional neural network based on the U-Net architecture to detect the TC wind structure, including maximum wind speed area and hurricane-force wind speed area, in the outputs of the convective-scale Arome model. A dataset of 400 Arome forecasts over the West Indies domain has been entirely hand labeled by experts, following a rigorous process to reduce heterogeneities. The U-Net performs well on a wide variety of TC intensities and shapes, with an average intersection-over-union metric around 0.8. Its performances however strongly depend on the TC strength, the detection of weak cyclones being more challenging since their structure is less well defined. The U-Net also significantly outperforms an operational heuristic detection model, with a significant gain for weak TCs, while running much faster. In a last part, the capacity of the U-Net to generalize on slightly different data is demonstrated in the context of a domain change and a resolution increase. In both cases, the pre-trained U-Net achieves similar performances as on the original dataset.
... due to the seasonality of precipitation (Wehner, 2004(Wehner, , 2013Wehner et al., 2021). Furthermore, extreme precipitation differs by storm type (Kunkel et al., 2012) and an extreme value sample should be further sub-sampled using storm trackers (Prabhat et al., 2012;Ullrich and Zarzycki, 2017) or some other technology. We recognize that this is not a standard practice but recommend it as an aspiration. ...
Methods for calculating return values of extreme precipitation and their uncertainty are compared using daily precipitation rates over the Western U.S. and Southwestern Canada from a large ensemble of climate model simulations. The roles of return-value estimation procedures and sample size in uncertainty are evaluated for various return periods. We compare two different generalized extreme value (GEV) parameter estimation techniques, namely L-moments and maximum likelihood (MLE), as well as empirical techniques. Even for very large datasets, confidence intervals calculated using GEV techniques are narrower than those calculated using empirical methods. Furthermore, the more efficient L-moments parameter estimation techniques result in narrower confidence intervals than MLE parameter estimation techniques at small sample sizes, but similar best estimates. It should be noted that we do not claim that either parameter fitting technique is better calibrated than the other to estimate long period return values. While a non-stationary MLE methodology is readily available to estimate GEV parameters, it is not for the L-moments method. Comparison of uncertainty quantification methods are found to yield significantly different estimates for small sample sizes but converge to similar results as sample size increases. Finally, practical recommendations about the length and size of climate model ensemble simulations and the choice of statistical methods to robustly estimate long period return values of extreme daily precipitation statistics and quantify their uncertainty.
... This could provide insight as whether the model is producing physically plausible outcomes or predictions. Other examples could include evaluating performance during extreme events such as cyclones or atmospheric rivers, and could involve using atmospheric phenomena-based tracking algorithms on GCM or RCM outputs (e.g., Ullrich & Zarzycki, 2017). Additionally, if an RCM emulator is capable of emulating variables such as specific humidity, winds, and mean sea level pressure, it may be possible to apply tracking algorithms to its outputs directly to assess cyclones and atmospheric rivers. ...
Despite the sophistication of global climate models (GCMs), their coarse spatial resolution limits their ability to resolve important aspects of climate variability and change at the local scale. Both dynamical and empirical methods are used for enhancing the resolution of climate projections through downscaling, each with distinct advantages and challenges. Dynamical downscaling is physics based but comes with a large computational cost, posing a barrier for downscaling an ensemble of GCMs large enough for reliable uncertainty quantification of climate risks. In contrast, empirical downscaling, which encompasses statistical and machine learning techniques, provides a computationally efficient alternative to downscaling GCMs. Empirical downscaling algorithms can be developed to emulate the behavior of dynamical models directly, or through frameworks such as perfect prognosis in which relationships are established between large-scale atmospheric conditions and local weather variables using observational data. However, the ability of empirical downscaling algorithms to apply their learned relationships out of distribution into future climates remains uncertain, as is their ability to represent certain types of extreme events. This review covers the growing potential of machine learning methods to address these challenges, offering a thorough exploration of the current applications and training strategies that can circumvent certain issues. Additionally, we propose an evaluation framework for machine learning algorithms specific to the problem of climate downscaling as needed to improve transparency and foster trust in climate projections.
Significance Statement
This review offers a significant contribution to our understanding of how machine learning can offer a transformative change in climate downscaling. It serves as a guide to navigate recent advances in machine learning and how these advances can be better aligned toward inherent challenges in climate downscaling. In this review, we provide an overview of these recent advances with a critical discussion of their advantages and limitations. We also discuss opportunities to refine existing machine learning methods alongside new approaches for the generation of large ensembles of high-resolution climate projections.
... Figures 3c-3f provides support of this hypothesis. Tracking the moisture contained within the area defined by the TCLVs using TempestExtremes (Ullrich & Zarzycki, 2017;Ullrich et al., 2021) reveals a large poleward increase of the PW within the TCLVs (Figure 3c). Increased PW at high latitudes is not due to increased evaporation or moisture advection by a mean meridional circulation. ...
This work explores the impact of rotation on tropical convection and climate. As our starting point, we use the RCEMIP experiments as control simulations and run additional simulations with rotation. Compared to radiative convective equilibrium (RCE) experiments, rotating RCE (RRCE) experiments have a more stable and humid atmosphere with higher precipitation rates. The intensity of the overturning circulation decreases, water vapor is cycled through the troposphere at a slower rate, the subsidence fraction decreases, and the climate sensitivity increases. Several of these changes can be attributed to an increased flux of latent and sensible heat that results from an increase of near‐surface wind speed with rotation shortly after model initialization. The increased climate sensitivity results from changes of both the longwave cloud radiative effect and the longwave clear‐sky radiative fluxes. This work demonstrates the sensitivity of atmospheric humidity and surface fluxes of moisture and temperature to rotation.
