The Lagrangian theory developed for fountains in a stationary fluid is extended to predict the path and breadth of a fountain in a one-and two-layer fluid with a moderate crossflow. The predictions compare well with the results of laboratory experiments of fountains in a one-layer fluid. The empirical spreading parameter determined from the one-layer experiments is used in the theory for fountains in a twolayer crossflow. Though qualitatively correct, the theory underpredicts the height and radius of the fountains. Similar to the behaviour of fountains in two-layer stationary ambients, the fountain in a two-layer crossflow is observed to exhibit three regimes of flow: it may penetrate the interface, eventually returning to the level of the source where it spreads as a propagating gravity current; upon descent, it may be trapped at the interface where it spreads as a propagating intrusion; it may do both, partially descending to the source and partially being trapped at the interface. These regimes are classified theoretically and empirically. The theoretical classification compared the buoyancy excess of the descending flow to the density difference between the two layers. The regimes are also classified using empirically determined regime parameters which govern the relative initial momentum of the fountain and the relative density difference of the fountain and the ambient fluid.

<p>Tornado warnings for more extreme near-storm environments generally have higher skill, with better probability of detection and false alarm ratio both associated with tornadic environments featuring higher CAPE, shear, storm-relative helicity, and lower LCL heights. Tornadoes do, however, often occur in association with more marginal near-storm environments, suggesting several possibilities that may occur in combination, including: (1) the proximity sounding used to define the environment may not be representative of the environment in which the tornadic storm actually developed (e.g., via convective contamination or positioning of the sounding to the wrong side of an environmental gradient), (2) the presence or absence of a tornado may display an extreme sensitivity to minor fluctuations in the environment (e.g., if CAPE is limited in an environment, a very small amount of CAPE may tip the balance), or (3) the presence or absence of a tornado may display an extreme lack of sensitivity to environmental parameters (e.g., the driving force for a given tornado is described by small-scale stochastic processes rather than any bulk features of the near-storm environment).</p> <p>In order to gauge the appropriateness and limitations of traditional near-storm environmental parameters in the cases for which near-storm environments are marginal, this work digs into US tornado events that are clustered via self-organizing maps around a node in which the 480 km x 480 km squares of significant tornado parameter (STP) surrounding the recorded tornado events are characterized by universally low values (the examination of a broad area surrounding the tornado helps ensure that representativeness does not hinge on a single proximity sounding). This marginal cluster is then further split via a nested SOM to tease out any subtle variations and patterns in the environment that may be present. Discussion focuses on regional, seasonal, and diurnal variability of these patterns across tornado events, with an eye to establishing some of the situations in which tornadogenesis occurs in spite of marginal-to-unfavorable conditions in the near-storm environment, opening the door to idealized numerical modeling studies.</p>

No abstract