Upper-level Features Research Articles

Progressive cross-level fusion network for RGB-D salient object detection

Depth maps can provide supplementary information for salient object detection (SOD) and perform better in handling complex scenes. Most existing RGB-D methods only utilize deep cues at the same level, and few methods focus on the information linkage between cross-level features. In this study, we propose a Progressive Cross-level Fusion Network (PCF-Net). It ensures the cross-flow of cross-level features by gradually exploring deeper features, which promotes the interaction and fusion of information between different-level features. First, we designed a Cross-Level Guide Cross-Modal Fusion Module (CGCF) that utilizes the spatial information of upper-level features to suppress modal feature noise and to guide lower-level features for cross-modal feature fusion. Next, the proposed Semantic Enhancement Module (SEM) and Local Enhancement Module (LEM) are used to further introduce deeper features, enhance the high-level semantic information and low-level structural information of cross-modal features, and use self-modality attention refinement to improve the enhancement effect. Finally, the multi-scale aggregation decoder mines enhanced feature information in multi-scale spaces and effectively integrates cross-scale features. In this study, we conducted numerous experiments to demonstrate that the proposed PCF-Net outperforms 16 of the most advanced methods on six popular RGB-D SOD datasets.

Persistent warm and cold spells in the Northern Hemisphere extratropics: regionalisation, synoptic-scale dynamics and temperature budget

Abstract. Persistent warm and cold spells are often high-impact events that may lead to significant increases in mortality and crop damage and can put substantial pressure on the power grid. Taking their spatial dependence into account is critical to understand the associated risks, whether in present-day or future climates. Here, we present a novel regionalisation approach of 3-week warm and cold spells in winter and summer across the Northern Hemisphere extratropics based on the association of the warm and cold spells with large-scale circulation. We identify spatially coherent but not necessarily connected regions where spells tend to co-occur over 3-week timescales and are associated with similar large-scale circulation patterns. We discuss the physical drivers responsible for persistent extreme temperature anomalies. Cold spells systematically result from northerly cold advection, whereas warm spells are caused by either adiabatic warming (in summer) or warm advection (in winter). We also discuss some key mechanisms contributing to the persistence of temperature extremes. Blocks are important upper-level features associated with such events – co-localised blocks for persistent summer warm spells in the northern latitudes; downstream blocks for winter cold spells in the eastern edges of continental landmasses; and upstream blocks for winter cold spells in Europe, northwestern North America and east Asia. Recurrent Rossby wave patterns are also relevant for cold and warm spell persistence in many mid-latitude regions, in particular in central and southern Europe. Additionally, summer warm spells are often accompanied by negative precipitation anomalies that likely play an important role through land–atmosphere feedbacks.

Displacement Error Characteristics of 500-hPa Cutoff Lows in Operational GFS Forecasts

Abstract Cutoff lows are often associated with high-impact weather; therefore, it is critical that operational numerical weather prediction systems accurately represent the evolution of these features. However, medium-range forecasts of upper-level features using the Global Forecast System (GFS) are often subjectively characterized by excessive synoptic progressiveness, i.e., a tendency to advance troughs and cutoff lows too quickly downstream. To better understand synoptic progressiveness errors, this research quantifies seven years of 500-hPa cutoff low position errors over the globe, with the goal of objectively identifying regions where synoptic progressiveness errors are common and how frequently these errors occur. Specifically, 500-hPa features are identified and tracked in 0–240-h 0.25° GFS forecasts during April 2015–March 2022 using an objective cutoff low and trough identification scheme and compared to corresponding 500-hPa GFS analyses. In the Northern Hemisphere, cutoff lows are generally underrepresented in forecasts compared to verifying analyses, particularly over continental midlatitude regions. Features identified in short- to long-range forecasts are generally associated with eastward zonal position errors over the conterminous United States and northern Asia, particularly during the spring and autumn. Similarly, cutoff lows over the Southern Hemisphere midlatitudes are characterized by an eastward displacement bias during all seasons. Significance Statement Cutoff lows are often associated with high-impact weather, including excessive rainfall, winter storms, and severe weather. GFS forecasts of cutoff lows over the United States are often subjectively noted to advance cutoff lows too quickly downstream, and thus limit forecast skill in potentially impactful scenarios. Therefore, this study quantifies the position error characteristics of cutoff lows in recent GFS forecasts. Consistent with typically anecdotal impressions of cutoff low position errors, this analysis demonstrates that cutoff lows over North America and central Asia are generally associated with an eastward position bias in medium- to long-range GFS forecasts. These results suggest that additional research to identify both environmental conditions and potential model deficiencies that may exacerbate this eastward bias would be beneficial.

Bidirectional Parallel Feature Pyramid Network for Object Detection

State-of-the-art Feature Pyramid Networks (FPNs) often focus on extracting features across different levels. In this paper, we propose a novel architecture, Bidirectional Parallel Feature Pyramid Network (BPFPN), to capture multi-scale spatial information from each level of FPN effectively. BPFPN consists of two blocks: Cross-level Channel Attention-Refinement (ClCSAR) Block and Weighted Parallel Feature Aggregation (WPFA) Block. ClCSAR block uses a channel attention mechanism to strengthen the context information of lower-level feature with aid from the upper-level feature. WPFA block exploits discriminating information from variable receptive fields via integrating multi-branch by employing dilated convolutions and using attention mechanisms to capture the salient dependencies over branches. Considering the incremental computation, we also give a lightweight version of BPFPN, namely BPFPN-Lite, integrated with an Efficient WPFA (E-WPFA) to improve detection accuracy while maintaining efficiency. Our proposed network can be easily plugged into existing object detection models and outperforms different feature pyramids methods by 0.2 ~ 2.1 on the COCO test-dev benchmark without bells and whistles.

A study on fault diagnosis of bearing pitting under different speed condition based on an improved inception capsule network

The bearings fault diagnosis is essential for the maintenance and reliability of rotating machinery. Bearings pitting is one of the most common fault types of rotating machinery. However, due to the complex working conditions of bearings, it is challenging to diagnose the pitting faults in bearings inner and outer rings at different speeds. In this paper, an improved one-dimensional inception capsule network (IICN) is proposed to solve the problem of bearing pitting fault diagnosis under complex working conditions. Firstly, the raw bearing vibration signal is processed using the improved Inception network. The function of the stage is to approximate an optimal local sparse structure with a simple dense substructure for bearing healthy state feature extraction. And then inputs concatenated features to the primary capsule layer and the routing capsule layer. The inputs are mapped to feature vector space and weighted by the dynamic routing algorithm. The dynamic routing algorithm encodes the significant spatial relationship between low-level features and upper-level features. The Euclidean length of each capsule vector is the probability of belonging to this bearing healthy condition. In order to validate the effectiveness of the IICN method, bearings pitting experiments at different speeds were designed. The raw bearings vibration signal data under six different health conditions are collected, and the effectiveness of the IICN method is verified. Experimental results show that the IICN method can effectively distinguish different degrees of bearing pitting fault at different speeds, and its diagnostic accuracy is superior to other advanced deep learning methods.

Research on Partial Least Squares Method Based on Deep Confidence Network in Traditional Chinese Medicine

Partial least squares method has many advantages in multivariate linear regression modeling, but its internal cross-checking method will lead to a sharp reduction of the principal component, thereby reducing the accuracy of the regression equation, and the selection of principal components about the traditional Chinese medicine data is particularly sensitive. This paper proposes a kind of partial least squares method based on deep belief nets. This method mainly uses the deep learning model to extract the upper-level features of the original data, putting the extracted features into the partial least squares model for multiple linear regression and evading the problem that selects the number of principal components, continuously adjusting the model parameters until satisfied well-pleased accuracy condition. Using Dachengqitang experimental data and data sets in the UCI Machine Learning Repository, the experimental results show that the partial least squares analysis method based on deep belief nets has good adaptability to TCM data.

Transfer Hierarchical Attention Network for Generative Dialog System

In generative dialog systems, learning representations for the dialog context is a crucial step in generating high quality responses. The dialog systems are required to capture useful and compact information from mutually dependent sentences such that the generation process can effectively attend to the central semantics. Unfortunately, existing methods may not effectively identify importance distributions for each lower position when computing an upper level feature, which may lead to the loss of information critical to the constitution of the final context representations. To address this issue, we propose a transfer learning based method named transfer hierarchical attention network (THAN). The THAN model can leverage useful prior knowledge from two related auxiliary tasks, i.e., keyword extraction and sentence entailment, to facilitate the dialog representation learning for the main dialog generation task. During the transfer process, the syntactic structure and semantic relationship from the auxiliary tasks are distilled to enhance both the word-level and sentence-level attention mechanisms for the dialog system. Empirically, extensive experiments on the Twitter Dialog Corpus and the PERSONA-CHAT dataset demonstrate the effectiveness of the proposed THAN model compared with the state-of-the-art methods.

Sahel Rainfall–Tropical Easterly Jet Relationship on Synoptic to Intraseasonal Time Scales

Abstract The tropical easterly jet (TEJ) is a characteristic upper-level feature of the West African monsoon (WAM) circulation. Moreover, the TEJ over West Africa is significantly correlated with summer Sahel rainfall on interannual and decadal time scales. In contrast, the relationship between Sahel rainfall and the regional TEJ on synoptic to intraseasonal time scales is unclear. Therefore, this relationship is investigated by means of multiple statistical analyses using temporally highly resolved measurement and reanalysis data. It is shown that average correlations between convective activity and regional TEJ intensity remain below 0.3 for all synoptic to intraseasonal time scales. Especially on the synoptic time scale, the TEJ significantly lags anomalies in convective activity by one or two days, which indicates that convection anomalies are more likely to drive changes in the regional TEJ than vice versa. To further shed light on the role of the TEJ for rainfall over West Africa, a previously proposed effect of TEJ-induced upper-level divergence on the development of mesoscale convective systems (MCSs) is examined more closely. An analysis of nearly 300 Sahelian MCSs shows that their initiation is generally not associated with significant TEJ anomalies or jet-induced upper-level divergence. Furthermore, no statistically significant evidence is found that preexisting TEJ-related upper-level divergence anomalies affect intensity, size, and lifetime of MCSs. A limiting factor of this study is the focus on TEJ-induced upper-level divergence. Therefore, a possible effect of the TEJ on Sahel rainfall via other mechanisms cannot be ruled out and should be subject to future studies.

Face recognition using position-dictionaries and region covariance feature

In this paper, a new sparsity formulation called position-dictionary based sparse representation is developed for frontal face recognition. Different from the sparse representation based classification (SRC) method and the Gabor-feature based SRC (GSRC) method which both employ a global dictionary to decompose image patches, the proposed method constructs a position-dictionary for each location using training patches in the corresponding location since they resemble each other and are more likely to favor the same atoms. Sparse coefficients of each position-patch can be obtained by solving an \(l_{1}\)-norm minimization problem. For each face image, sparse coefficients of position-patches are pooled to construct a discriminative upper level feature to represent face image. PCA is used to perform dimension reduction. Each testing sample is represented as a sparse linear combination of all training samples, and recognition is accomplished by evaluating which class of training samples leads to the minimum reconstruction error. We compared the proposed method with SRC and GSRC method on three benchmark face databases. Experimental results show that the proposed method achieves higher recognition rates and is robust to a certain degree of occlusions.

Perturbing the potential vorticity field in mesoscale forecasts of two Mediterranean heavy precipitation events

ABSTRACTIn order to improve the quality of the Mediterranean high-impact weather (HIW) numerical predictions, this study proposes to modify the potential vorticity (PV) field of the model initial state, taking advantage of information provided by the water vapour (WV) channel of the METEOSAT-7 satellite. The implemented PV field modifications aim to reduce the mismatch between the upper-level PV features and the WV brightness temperatures guided by the known relation between these two fields (PV-WV technique). The PV-WV technique effectiveness is evaluated on two HIW events, and is also compared with two additional PV modification techniques from an earlier study. The chosen episodes occurred on 9–10 June 2000 and 9–10 October 2002 and produced heavy precipitation over both Spain and France. The main difference between these two episodes is found in the driving mechanism, a mesoscale cyclone for the June 2000 event and a larger low-pressure centre for the October 2002 case. The two additional PV modification techniques introduce perturbations along the zones highlighted by the MM5 adjoint model calculated sensitivity zones (PV-adjoint) and along the three-dimensional PV structure presenting the locally most intense values and gradients of the field (PV-gradient). A close examination of both case studies of the forecast rainfall fields and several objective verification indices show that the PV-WV technique performance exceeds the control (or non-perturbed) forecast skill while remaining inside the distribution obtained by both PV-gradient and PV-adjoint techniques. This PV-WV technique could be used to increase the ensemble spread introducing higher amplitude modifications. Thus, a more skilled ensemble prediction system could be built by taking advantage of the subjectivity inherent to this method (manual perturbations) and also of the uncertainty present in the initial state.

The Influence of Upper-Tropospheric Potential Vorticity on Convective Morphology

Abstract Extensive research has illuminated the roles that buoyancy and vertical wind shear have in determining convective mode. The goal of this study is to examine synoptic forcing, as measured by the form of the waves on the dynamic tropopause and the strength of the resultant temperature advection, together with the environmental variables that are currently used to differentiate storm type. Logistic regression is used to make this discrimination and provides the preliminary results. First, multicellular lines and isolated rotating cells are associated with weaker synoptic forcing. Second, where an upper-level feature is present, multicellular convection (lines and clusters combined) is favored relative to isolated storms when there is stronger synoptic forcing, the shape of the upper-level potential vorticity (PV) feature is elongated, and there is a more southerly component in the deep-layer (0–6 km) mean wind. Third, when an upper-level feature is present and the convection is multicellular, clusters are favored relative to lines when the upper wave is not cut off (i.e., a wave structure promoting advection), provided that the deep-layer shear is not oriented at an optimal angle relative to a preexisting meridional boundary (as shown by prior research). Finally, if the convection is isolated, rotating storms may be favored for a broad area of PV advection, as measured by the relative equality of the zonal and meridional axes of the advection area, suggesting the importance of broader-scale destabilization. These findings are used to formulate testable hypotheses. Future model-based experiments using PV surgery or PV regression methods are proposed to clearly elucidate the dynamics behind these relationships.

Comparison between the Large-Scale Environments of Moderate and Intense Precipitating Systems in the Mediterranean Region

Abstract A characterization of the large-scale environment associated with precipitating systems in the Mediterranean region, based mainly on NOAA-16 Advanced Microwave Sounding Unit (AMSU) observations from 2001 to 2007, is presented. Channels 5, 7, and 8 of AMSU-A are used to identify upper-level features, while a simple and tractable method, based on combinations of channels 3–5 of AMSU-B and insensitive to land–sea contrast, was used to identify precipitation. Rain occurrence is widespread over the Mediterranean in wintertime while reduced or short lived in the eastern part of the basin in summer. The location of convective precipitation shifts from mostly over land from April to August, to mostly over the sea from September to December. A composite analysis depicting large-scale conditions, for cases of either rain alone or extensive areas of deep convection, is performed for selected locations where the occurrence of intense rainfall was found to be important. In both cases, an upper-level trough is seen to the west of the target area, but for extreme rainfall the trough is narrower and has larger amplitude in all seasons. In general, these troughs are also deeper for extreme rainfall. Based on the European Centre for Medium-Range Weather Forecasts operational analyses, it was found that sea surface temperature anomalies composites for extreme rainfall are often about 1 K warmer, compared to nonconvective precipitation conditions, in the vicinity of the affected area, and the wind speed at 850 hPa is also stronger and usually coming from the sea.

Synoptic- and Frontal-Scale Influences on Tropical Transition Events in the Atlantic Basin. Part II: Tropical Transition of Hurricane Karen

Abstract A finescale simulation of the tropical transition of Atlantic Hurricane Karen in October 2001 is examined to determine the processes leading to the development of upshear convection and its effects on the process of tropical transition. The analysis shows that, as in marine extratropical cyclones, the area upshear of the pretransition cyclone is characterized by reduced stability. Lower-tropospheric frontogenesis leads to an intense burst of convection there and instigates three important processes that combine to produce a full-fledged tropical cyclone. First, the convection generates intense low-level vorticity on the western half of the cyclone, which quickly dominates the cyclone’s vorticity field eventually organizing the circulation into a small-scale, intense vortex. Second, the diabatically enhanced circulation hastens the isolation of the cyclone’s developing warm core by intensifying cold air advection on the northern and western sides of the storm and by placing evaporatively cooled air into the boundary layer to the south of the cyclone. Third, upshear convection vertically redistributes potential vorticity (PV) from the tropopause to the surface and introduces a component to the upper-level winds, which advects strong, shear-inducing PV gradients away from the column above the cyclone. These three processes transform the initial extratropical cyclone into a frontless vortex with tropical storm–force winds and a warm core in a low-shear environment. These features are sufficient, given a warm enough ocean surface, to allow self-amplification of the storm as a tropical cyclone. The results further blur the distinction between tropical and extratropical cyclones as many of the processes identified as important to transition are similar to those that characterize ordinary marine cyclones and the extratropical occlusion process with the key distinctions being that here the convection is stronger and the initial upper-level feature is weaker. Thus, tropical transition of strong extratropical precursors follows the canonical midlatitude cyclone life cycle with upshear convection serving as the catalyst that both induces and organizes processes that favor tropical cyclogenesis in the postmature phase.

Hurricane Katrina (2005). Part II: Evolution and Hemispheric Impacts of a Diabatically Generated Warm Pool

Abstract The landfall of Hurricane Katrina (2005) near New Orleans, Louisiana, on 29 August 2005 will be remembered as one of the worst natural disasters in the history of the United States. By comparison, the extratropical transition (ET) of the system as it accelerates poleward over the following days is innocuous and the system weakens until its eventual demise off the coast of Greenland. The extent of Katrina’s perturbation of the midlatitude flow would appear to be limited given the lack of reintensification or downstream development during ET. However, the slow progression of a strong upper-tropospheric warm pool across the North Atlantic Ocean in the week following Katrina’s landfall prompts the question of whether even a nonreintensifying ET event can lead to significant modification of the midlatitude flow. Analysis of Hurricane Katrina’s outflow layer after landfall suggests that it does not itself make up the long-lived midlatitude warm pool. However, the interaction between Katrina’s anticyclonic outflow and an approaching baroclinic trough is shown to establish an anomalous southwesterly conduit or “freeway” that injects a preexisting tropospheric warm pool over the southwestern United States into the midlatitudes. This warm pool reduces predictability in medium-range forecasts over the North Atlantic and Europe while simultaneously aiding in the development of Hurricanes Maria and Nate. The origin of the warm pool is shown to be the combination of anticyclonic upper-level features generated by eastern Pacific Hurricane Hilary and the south Asian anticyclone (SAA). The hemispheric nature of the connections involved with the development of the warm pool and its injection into the extratropics has an impact on forecasting, since the predictability issue associated with ET in this case involves far more than the potential reintensification of the transitioning system itself.

Potential of Advanced Microwave Sounding Unit to identify precipitating systems and associated upper‐level features in the Mediterranean region: Case studies

The potential of the Advanced Microwave Sounding Unit (AMSU) observations to identify and characterize precipitating systems in the Mediterranean region is explored. Single channels or combination channels from AMSU‐A are used to detect and locate upper level potential vorticity anomalies that are often associated with intensification of surface low systems and occurrence of extreme events, while AMSU‐B data is used to detect precipitating areas. The motivation for the approach presented here is the direct use of satellite data as an alternative for reanalysis data sets for climatological studies of Mediterranean lows, without relying on retrieval algorithms. AMSU‐A channel 8 was found to be more suitable to identify upper level southward intrusions of stratospheric air than the difference of channels 7 and 5, which detects only vertically deep intrusions. A combination of AMSU‐B channels 3 and 5 is able to discriminate moderate to strongly precipitating areas with good agreement with Tropical Rainfall Measuring Mission (TRMM) derived products and independent ground‐based data. A more stringent condition based on differences of channels 3 to 5 was found to be useful to detect deep convective clouds. We demonstrate the applicability of AMSU to detect upper level features and precipitating systems for selected case studies of extreme precipitation in the Mediterranean region. These tools will allow us to form a climatology of moderate to strongly precipitating systems, and to investigate their relationship with upper level features that may be precursors of extreme events, and to establish a typology of the precipitating systems in the Mediterranean region.

Dynamical influences of precipitation assimilation on regional downscaling

[1] Precipitation assimilation has previously been found to be useful for improving regional climate model simulations of the surface water and energy cycles. Here we show that upper-level dynamical features are also better simulated. In particular, noticeable improvements in the intensity and location of the subtropical upper-level westerly jet and associated transverse circulations were found in a regional climate model simulation of the 1993 heavy summertime precipitation over the midwestern United States. We suggest that the increased rainfall directly over the upper Mississippi River Basin during the 1993 summer was linked with the location of the upper-level jet, and its associated transverse circulation, in addition to increased low-level summertime meridional moisture transport from the Gulf of Mexico.

Two case studies of severe storms in the Mediterranean using AMSU

Abstract. Mediterranean storms and their associated upper level features are diagnosed here using Advanced Microwave Sounding Unit (AMSU) observations. AMSU-A channel 8 is used to identify upper-level intrusions of stratospheric air, which are often present upstream of heavy precipitating areas, while a combination of AMSU-B channels 3 and 5 is chosen to discriminate moderate to heavily precipitating areas. This precipitation detection method provides results that are in good agreement with TRMM rainfall product and independent ground-based precipitation data. These tools allow us to follow the concomitant evolution of two severe rainfall events in the Mediterranean region and associated upper-level features.

Weather And Death On Mount Everest: An Analysis Of The Into Thin Air Storm

Scientific interest in Mount Everest has been largely focused on the physiology of hypoxia caused by the summit's low barometric pressure. Although weather is recognized as a significant risk for climbers on the mountain, it has not been extensively studied. In this paper, we reconstruct the meteorological conditions associated with the deadly outbreak of high-impact weather on Mount Everest that occurred in May 1996 and was the subject of the best-selling book Into Thin Air. The authors show that during this event, two jet streaks—an upper-level shortwave trough and an intrusion of stratospheric air into the upper troposphere—were present in the vicinity of Mount Everest. Meanwhile, in the lower troposphere, there was convergence of water vapor transport from both the Arabian Sea and the Bay of Bengal into the region to the south of Mount Everest. The authors propose that the ageostrophic circulation associated with the upper-level features resulted in a region of large-scale ascent near Mount Everest that, in combination with the anomalous availability of moisture in the region, triggered convective activity. The resulting high-impact weather trapped over 20 climbers on Mount Everest's exposed upper slopes leading to the deaths of 8. These synoptic-scale characteristics provide some expectation of predicting life-threatening high-altitude storms in the Himalayas. In addition, the authors argue that the falling barometric pressure and the presence of ozone-rich stratospheric air that occurred near the summit of Mount Everest during this event could have shifted a coping climber from a state of brittle tolerance to physiological distress.

Influence of Atmospheric Asymmetries on the Intensification of GFDL Model Forecast Hurricanes

Abstract Hurricanes Bertha of 1996 and Erin of 2001 both intensified rapidly during part of their time over the North Atlantic. A piecewise potential vorticity (PV) inversion is applied to model output from GFDL hurricane model forecasts to determine the contributions of atmospheric features in the hurricanes’ environment to their intensification. The diagnosis indicates that Hurricane Bertha’s rapid intensification was directly augmented by an upper-level trough to the north. The significant positive impact of the trough provides quantitative confirmation of the inference of other authors. By contrast environmental interactions associated with troughs to the east and west of Hurricane Erin did not contribute directly to its rapid intensification. The implication of this result is that factors other than the troughs, including sea surface temperature, were sufficient to effect Hurricane Erin’s strengthening. Enhanced upper-level outflow concentrated northeast of the hurricane’s center that was associated with upper-level PV features to the north of Erin, including those ahead of the long-wave trough to its west, could have had some indirect contribution to its intensification. The present authors’ previous piecewise inversion applied to a model forecast of Hurricane Opal of 1995 indicated that an approaching upper-level trough did not significantly contribute to the hurricane’s lower-tropospheric intensification. The conclusions of this paper demonstrate that this result is neither an exception nor the rule.

Surface and upper-level features associated with wintertime cold surge outbreaks in South Korea

The surface and upper-level features associated with a sharp drop of wintertime daily temperature over South Korea is investigated in this study. This sharp drop in daily temperature is called a cold surge and is one of the most hazardous weather phenomena in East Asian winters. An upper-level baroclinic wave of 60° wavelength propagating eastward at a phase speed of 12° longitude per day across the continent of northern China from the west of Lake Baikal toward the eastern coast of China causes the outbreak of cold air over South Korea. The cooling associated with the upper-level baroclinic wave is found at all altitudes under the geopotential height-fall center near the tropopause. The development in the ridge seems to derive the early evolution of the eastward-propagating sinusoidal wave, whereas the trough is connected directly with the tropospheric temperature-drop. An enhancement of the wintertime East Asian jet stream after the outbreak of a cold surge is a response to the steep temperature gradient associated with the developing baroclinic wave.