1. Introduction
Extratropical cyclones (ETCs) play a critical role in regulating Earth’s energy balance because they efficiently transport heat and energy from the tropical regions poleward, as well as influencing the climate system through the clouds and precipitation that they produce. ETCs and their fronts are associated with up to 90% of rainfall in storm-track regions (Catto et al. 2012) and are an important source of freshwater for the middle and upper latitudes because they produce a majority of the precipitation observed in the planet’s temperate zones (Heideman and Fritsch 1988; Hawcroft et al. 2012).
Most of the water vapor that is converted to clouds and precipitation within extratropical fronts and cyclones is transported poleward by the warm conveyor belt (WCB; Harrold 1973; Browning et al. 1973; Carlson 1980; Browning 1986; Wernli 1997; Catto 2016). WCBs typically originate in a moist subtropical marine planetary boundary layer between 20° and 47°N and between 17° and 42°S (Wernli and Davies 1997; Eckhardt et al. 2004; Madonna et al. 2014), and they transport large quantities of sensible and latent heat poleward and upward into ETCs (Browning 1990). The uptake of moisture in a WCB is commonly associated with anomalies of latent heat flux associated with a reduction of near-surface relative humidity and enhanced wind velocity (Neiman and Shapiro 1993; Pfahl et al. 2014).
The dynamics and evolution of ETCs at the synoptic level have been well known for many decades (e.g., Bjerknes and Solberg 1922; Shapiro and Keyser 1990), but there remain unanswered questions associated with how changes in cyclone circulation and environment relate to variations to the distribution and intensity of clouds and precipitation, along with how surface processes, especially over the ocean, relate to ETC development (e.g., Crespo and Posselt 2016). Current in situ and remote sensing measurements are limited in their ability to observe surface processes in oceanic ETCs. The ocean surface around ETCs is often obscured by clouds, limiting the utility of visible and infrared remote sensing. Spaceborne passive microwave and radar instruments can penetrate clouds, but nearly all have infrequent revisit time and large data gaps and/or have their signals attenuated by precipitation (as is the case for most scatterometers), causing them to miss key details in ETC development. As such, the Cyclone Global Navigation Satellite System (CYGNSS) can observe a significant number of low-latitude extratropical fronts and cyclones and aid in our understanding of how surface processes play a role in cyclogenesis and evolution in the lower latitudes (Small et al. 2008; Hoskins and Hodges 2002, 2005; Graf et al. 2017). In this paper, we use an orbit simulator and a database of objectively identified ETC fronts and cyclone centers to demonstrate the frequency with which CYGNSS will observe marine extratropical cyclones forming in the lower midlatitudes, as well as showing how representative the CYGNSS-sampled surface winds and surface heat fluxes will be by using a reanalysis dataset.
2. Data and methods
a. Simulated CYGNSS ground tracks and specular point locations
CYGNSS observes the signal from the global positioning system (GPS) reflected from Earth’s surface using a delay Doppler mapping instrument (DDMI) with a multichannel Global Navigation Satellite Systems (GNSS)-R receiver, low-gain zenith antenna, and two high-gain nadir antennas. On every CYGNSS observatory, the DDMI selects four specular points each second within the highest-sensitivity region of its antenna view pattern. The wind-induced roughness of the ocean surface scatters the GPS signal, and the scattering signature is used to estimate the surface wind speeds at a 25-km spatial resolution per specular point (Ruf et al. 2016). Our analysis requires the locations of surface specular points for signals generated by the GPS transmitters and observed by the CYGNSS constellation.
The Spacecraft Orbital Characterization Kit (SpOCK) simulates the trajectories of the GPS and CYGNSS constellations expected over a 1-yr period, assuming all eight satellites are in science-data-collecting mode and in their final constellation configuration. From the modeled positions and velocities of each transmitting (GPS) and target (CYGNSS) spacecraft, SpOCK derives the positions of Earth’s surface reflection points. These are defined as the locations at which the angle between the local vertical and the GPS is equal to the angle between the local vertical and the CYGNSS observatory. Each CYGNSS satellite receives reflected signals from about 10 GPS transmitters at all times, but only the four signals with the highest range-corrected gains (RCGs) are selected; the other specular points are ignored. The RCG is defined as the look-angle-dependent antenna gain divided by the sum of the distance from CYGNSS to the surface specular point squared and the distance from the GPS satellite to the surface specular point squared. When all eight CYGNSS observatories are in science-data-collecting mode, the entire constellation returns 32 surface specular point observations each second.
b. Cyclone- and front-detection and compositing method
The cyclone and frontal transects are produced with the aid of an objective cyclone- and front-detection algorithm, which has previously been used to create a composite of observations from A-Train satellite data (Naud et al. 2010, 2012, 2015; Naud and Kahn 2015) and which we use in this study to create a composite of simulated CYGNSS specular point locations. Cyclone centers are identified using the algorithm described by Bauer and Del Genio (2006) and Bauer et al. (2016) and are freely available in an online database (https://gcss-dime.giss.nasa.gov/mcms/).
Identification of frontal boundaries utilizes 6-hourly gridded temperature, humidity, wind, and geopotential height information (Naud et al. 2010, 2015, 2016), which were obtained from the second release of the Modern-Era Retrospective Analysis for Research and Applications (MERRA-2; Rienecker et al. 2011; Posselt et al. 2012; Molod et al. 2015). For this, aspects of two different objective and automated front-detection algorithms are combined: a temperature gradient-based method that was described by Hewson (1998) and a wind direction and strength-based method that was used by Simmonds et al. (2012). As described in Naud et al. (2012), warm fronts are detected with the Hewson (1998) technique. Cold fronts are detected with both methods [see Schemm et al. (2015) for a discussion of each method’s merits and Naud et al. (2016) for a detailed description of how they are combined]. Frontal locations are used as anchors for compositing MERRA-2 surface winds, latent heat fluxes, and sensible heat fluxes along a line perpendicular to each front type, with the origin representing the frontal location at the surface and the x axis representing the distance from the front (e.g., Naud et al. 2012, 2015). The result is a database of front-relative composites for all fronts and cyclones detected within a given period of time.
CYGNSS observations commonly occur as continuous lines of specular points a few hundreds of kilometers in length along Earth’s surface, as a specific GPS transmitter’s orbit carries it across the CYGNSS antenna gain pattern (Ruf et al. 2013, 2016). Using one year of synthetic CYGNSS footprints and assuming no year-to-year variability in the orbits of either CYGNSS observatories or GPS transmitters, we can generate a synthetic sample of CYGNSS observations of fronts and cyclones from several years of MERRA-2 data. This second set of CYGNSS-specific warm- and cold-front-centered composites may then be compared with the full set of data sampled from all fronts and storms in MERRA-2 between 40°S and 40°N to evaluate the robustness of CYGNSS’s sampling.
3. Results
By using all extratropical cyclone features that were objectively identified in the MERRA-2 database from 2014 to 2015 and the 1-yr database of simulated CYGNSS specular points repeated over these two years, we can examine the frequency with which CYGNSS may observe ETCs and their fronts within the constellation’s latitude range. Figure 1 depicts the expected frequency of number of simulated specular points that fall within 500 km of a cyclone center (Fig. 1a) and within 25 km of a warm front (Fig. 1b) and cold front (Fig. 1c) accumulated over both years. We use these constraints because ETCs often have diameters over 1000 km whereas the fronts’ direct surface impact and influence tend to occur close to the front itself (Naud et al. 2016). Despite the nonuniform sampling of the CYGNSS constellation (Fig. 1d), a significant portion of the globe between 40°S and 40°N is observed by CYGNSS each day. As such, storms in the simulated CYGNSS sample exhibit a spatial distribution consistent with the preferred location of extratropical cyclones within the 40°S–40°N latitude band (e.g., Hoskins and Hodges 2002, 2005). The number of cyclones and fronts in the Northern Hemisphere increases with increasing distance from the equator. In the Southern Hemisphere, the number of CYGNSS simulated observations of cyclones and fronts is maximized around 25°–35°S. Sampling of cyclones and fronts in the Northern Hemisphere is nearly twice as frequent in the western Atlantic Ocean as it is in the eastern Atlantic and is nearly 5 times as frequent in the western Pacific as it is in the eastern Pacific. These patterns reflect the influence of the warm western boundary currents (Kuroshio east of Japan; Gulf Stream east of North America) on the development of ETCs. In the Southern Hemisphere, we see less zonal variability in ETC and frontal observations from CYGNSS. This is expected, because ETCs and associated WCBs are distributed more evenly throughout the Southern Hemisphere than they are in the Northern Hemisphere (Eckhardt et al. 2004). Although we do not know how many ETCs CYGNSS will observe per annum, our simulations imply that CYGNSS will be able to make many observations of ETCs developing in the lower latitudes.
Further assessment of the sampling effectiveness of the CYGNSS constellation can be assessed by examining the composite distribution of winds and surface fluxes across cold and warm fronts and comparing the CYGNSS subsample with the full sample (40°S–40°N) from the MERRA-2 dataset. Figure 2 depicts the composite distribution of winds and surface latent and sensible heat fluxes for the entire database and for the CYGNSS subset for the 2014 and 2015 calendar years. Individual years are shown, as is the composite of all transects over two years; comparison between individual years and the total composite provides an estimate of the year-to-year variability in ETC locations in comparison with CYGNSS sampling. Since the observatories are intended to maintain even spacing around Earth, their orbits will not change from year to year. If we assume consistent GPS transmitter orbits, then the differences between simulated observations in 2014 and 2015 will be due to changes in the location and frequency of ETCs in each year.
Consistent with previous studies (e.g., Sinclair 2013), wind speeds are maximized near the surface frontal boundary, with a local minimum at the location of the front itself. Wind speeds are larger in the warm sector ahead of cold fronts, with slightly weaker winds west of fronts in the cold air. This is consistent with stronger south-to-north transport ahead of the cold front associated with the WCB. Winds are stronger poleward of warm fronts in a narrow region that is approximately 100 km wide. The surface fluxes follow the distribution of winds in the cold air (west of cold fronts and poleward of warm fronts), but they are much lower in the warm air equatorward and eastward of the cyclone center. Surface fluxes depend not only on the wind speed but also on the surface-to-air temperature and water vapor contrast, and they play a crucial role in ETC and frontal development (Neiman and Shapiro 1993). Smaller fluxes in the warm sector reflect the relatively higher near-surface relative humidity in this region associated with the WCB, along with the relatively smaller air–sea temperature differences.
CYGNSS observes the distribution of wind speeds and fluxes across cold fronts (Fig. 2, left column) very effectively, with wind speed deviations in the composite of less than 0.5 m s−1 over a large range of distances from the surface front. Although latent and sensible heat fluxes are also well observed, CYGNSS-sampled fluxes exhibit a larger interannual range on the cold (west) side of the front relative to the warm (east) side. In addition, CYGNSS-sampled sensible heat fluxes appear to be systematically larger than the full dataset by a few watts per meter squared east of the frontal location. CYGNSS-sampled wind speeds and fluxes match the full sample very closely on the warm (equatorward) side of warm fronts (Fig. 2, right column) but exhibit more variability on the cold (poleward) side. CYGNSS-sampled wind speeds appear to be systematically higher by approximately 0.5–1.0 m s−1. Consistent with higher wind speeds, latent heat fluxes are also larger poleward of the front in the CYGNSS sample, as compared with the full database. Of interest is that CYGNSS-sampled sensible heat fluxes are systematically lower on the cold side of the warm fronts. At first glance, this appears to run counter to the fact that the wind speeds are biased toward being slightly too high.
These discrepancies between the CYGNSS subsample and the complete MERRA-2 sample (40°S–40°N) on the cold side of warm fronts cannot be traced to retrieval errors, because the CYGNSS-sampled data come from the same source as the full dataset and are not perturbed with any assumed retrieval noise. Instead, the differences are likely due to the CYGNSS sampling pattern. Recall that CYGNSS orbits in a tropical inclination with bounds at ±35° latitude; with a science-antenna point at a 28° angle with respect to the subsatellite point, dense specular point coverage extends to approximately 38° latitude and then rapidly decreases (Ruf et al. 2013). In practice, this means that there are relatively fewer CYGNSS samples with increasing distance poleward of the warm front. If the samples in the simulated CYGNSS dataset are biased toward having more members closer to the equator and if these samples also occur over regions with smaller air–sea temperature differences (as is more often the case at lower latitudes), then the surface sensible heat fluxes may be lower, even in the presence of slightly higher winds.
A more detailed evaluation of the CYGNSS sample versus the full MERRA-2 sample can be obtained by comparing histograms of wind speeds and surface fluxes over the full range of distances from the front, as well as for the cold and warm sides of the fronts independently (Fig. 3). Histograms of wind speeds match closely for cold fronts, and there is little difference in the histograms of wind speeds on the cold versus warm sides of the fronts. The heat fluxes also match very closely between the CYGNSS and full samples. As one might expect, given the greater air–sea temperature and water vapor contrast on the cold side of fronts, the latent and sensible heat fluxes are systematically higher in these regions.
In contrast to cold frontal regions, there are differences between the CYGNSS-sampled winds and fluxes in warm frontal regions. Close examination reveals that the histograms match much better on the warm side of fronts, relative to the cold side. On the cold side, it is clear that the CYGNSS-sampled wind speeds are systematically higher across all values of wind speed by 0.5–1.0 m s−1. In contrast, there are relatively fewer values of latent heat flux lower than ~150 W m−2, and relatively greater numbers of latent heat flux values higher than ~150 W m−2 in the CYGNSS dataset.
4. Discussion and conclusions
In this paper, we conducted an examination of the simulated CYGNSS sampling of MERRA-2 winds and surface fluxes across warm and cold fronts, using one year of simulated CYGNSS specular point locations and multiple years of cyclone center and front locations objectively identified in reanalysis data. Despite its tropical mission, we have shown that CYGNSS has the ability to observe extratropical cyclones and fronts developing in the lower latitudes. Wind speed and surface flux distribution observed by simulated CYGNSS measurements matched the full cyclone sample very effectively, with somewhat larger deviations poleward of the warm fronts where CYGNSS does not observe as frequently. Our conclusion is that there will be sufficient information in the CYGNSS measurements during its nominal mission lifetime to provide new insights into the distribution of winds and surface fluxes in oceanic low-latitude ETCs, as well as future analysis of warm conveyor belts. As Madonna et al. (2014) stated, many WCBs begin their ascents near the surface over the oceans between ~20° and 40° in both hemispheres, an area that is well observed by CYGNSS. In addition, the ability of CYGNSS to observe winds even in regions of heavy precipitation, along with the planned collocation of CYGNSS surface winds with the Global Precipitation Measurement (GPM) mission Integrated Multisatellite Retrievals (IMERG) precipitation dataset, will motivate studies of wind–precipitation relationships in frontal zones.
In closing, we point out that the sampling study we have conducted provides the most optimistic estimate of CYGNSS’s potential. In reality, the surface wind speed and surface flux retrievals will contain errors (2 m s−1 below 20 m s−1 wind speeds; 10% error above 20 m s−1 winds) that derive from a number of different sources (Ruf et al. 2013). In addition, our study assumes consistent sampling with all eight CYGNSS spacecraft. In reality, there will be occasional periods during which one or more of the spacecraft is taken out of science-data mode and placed in drag mode to maintain constellation spacing. This will lead to a small decrease in the number of samples available, but we do not believe it will fundamentally change CYGNSS’s sampling of ETCs and fronts.
Acknowledgments
The authors of this work were supported by the CYGNSS mission under NASA Science Mission Directorate Contract NNL13AQ00C, along with partial funding from NASA Grants NNX13AQ33G and NNX16AD82G. A portion of this research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. The comments given by the two anonymous reviewers were appreciated and helped to improve the clarity of analysis and discussion.
REFERENCES
Bauer, M., and A. D. Del Genio, 2006: Composite analysis of winter cyclones in a GCM: Influence of climatological humidity. J. Climate, 19, 1652–1672, doi:10.1175/JCLI3690.1.
Bauer, M., G. Tselioudis, and W. B. Rossow, 2016: A new climatology for investigating storm influences in and on the extratropics. J. Appl. Meteor. Climatol., 55, 1287–1303, doi:10.1175/JAMC-D-15-0245.1.
Bjerknes, J., and H. Solberg, 1922: Life cycle of cyclones and the polar front theory of atmospheric circulation. Geofys. Publ., 3, 1–18.
Browning, K. A., 1986: Conceptual models of precipitation systems. Wea. Forecasting, 1, 23–41, doi:10.1175/1520-0434(1986)001<0023:CMOPS>2.0.CO;2.
Browning, K. A., 1990: Organization of clouds and precipitation in extratropical cyclones. Extratropical Cyclones: The Erik Palmén Memorial Volume, C. W. Newton and E. O. Holopainen, Eds., Amer. Meteor. Soc., 129–153.
Browning, K. A., M. E. Hardman, T. W. Harrold, and C. W. Pardoe, 1973: Structure of rainbands within a mid-latitude depression. Quart. J. Roy. Meteor. Soc., 99, 215–231, doi:10.1002/qj.49709942002.
Carlson, T. N., 1980: Airflow through midlatitude cyclones and the comma cloud pattern. Mon. Wea. Rev., 108, 1498–1509, doi:10.1175/1520-0493(1980)108<1498:ATMCAT>2.0.CO;2.
Catto, J. L., 2016: Extratropical cyclone classification and its use in climate studies. Rev. Geophys., 54, 486–520, doi:10.1002/2016RG000519.
Catto, J. L., C. Jakob, G. Berry, and N. Nicholls, 2012: Relating global precipitation to atmospheric fronts. Geophys. Res. Lett., 39, L10805, doi:10.1029/2012GL051736.
Crespo, J. A., and D. J. Posselt, 2016: A-Train-based case study of stratiform–convective transition within a warm conveyor belt. Mon. Wea. Rev., 144, 2069–2084, doi:10.1175/MWR-D-15-0435.1.
Eckhardt, S., A. Stohl, H. Wernli, P. James, C. Forster, and N. Spichtinger, 2004: A 15-year climatology of warm conveyor belts. J. Climate, 17, 218–237, doi:10.1175/1520-0442(2004)017<0218:AYCOWC>2.0.CO;2.
Graf, M. A., H. Wernli, and M. Sprenger, 2017: Objective classification of extratropical cyclogenesis. Quart. J. Roy. Meteor. Soc., 143, 1047–1061, doi:10.1002/qj.2989.
Harrold, T. W., 1973: Mechanisms influencing the distribution of precipitation within baroclinic disturbances. Quart. J. Roy. Meteor. Soc., 99, 232–251, doi:10.1002/qj.49709942003.
Hawcroft, M., L. Shaffrey, K. Hodges, and H. Dacre, 2012: How much Northern Hemisphere precipitation is associated with extratropical cyclones? Geophys. Res. Lett., 39, L24809, doi:10.1029/2012GL053866.
Heideman, K. F., and J. M. Fritsch, 1988: Forcing mechanisms and other characteristics of significant summertime precipitation. Wea. Forecasting, 3, 115–130, doi:10.1175/1520-0434(1988)003<0115:FMAOCO>2.0.CO;2.
Hewson, T. D., 1998: Objective fronts. Meteor. Appl., 5, 37–65, doi:10.1017/S1350482798000553.
Hoskins, B., and K. Hodges, 2002: New perspectives on the Northern Hemisphere winter storm tracks. J. Atmos. Sci., 59, 1041–1061, doi:10.1175/1520-0469(2002)059<1041:NPOTNH>2.0.CO;2.
Hoskins, B., and K. Hodges, 2005: A new perspective on Southern Hemisphere storm tracks. J. Climate, 18, 4108–4129, doi:10.1175/JCLI3570.1.
Madonna, E., H. Wernli, H. Joos, and O. Martius, 2014: Warm conveyor belts in the ERA-Interim dataset (1979–2010). Part I: Climatology and potential vorticity evolution. J. Climate, 27, 3–26, doi:10.1175/JCLI-D-12-00720.1.
Molod, A., L. Takacs, M. Suarez, and J. Bacmeister, 2015: Development of the GEOS-5 atmospheric general circulation model: Evolution from MERRA to MERRA2. Geosci. Model Dev., 8, 1339–1356, doi:10.5194/gmd-8-1339-2015.
Naud, C. M., and B. H. Kahn, 2015: Thermodynamic phase and ice cloud properties in Northern Hemisphere winter extratropical cyclones observed by Aqua AIRS. J. Appl. Meteor. Climatol., 54, 2283–2303, doi:10.1175/JAMC-D-15-0045.1.
Naud, C. M., A. D. Del Genio, M. Bauer, and W. Kovari, 2010: Cloud vertical distribution across warm and cold fronts in CloudSat–CALIPSO data and a general circulation model. J. Climate, 23, 3397–3415, doi:10.1175/2010JCLI3282.1.
Naud, C. M., D. J. Posselt, and S. C. van den Heever, 2012: Observational analysis of cloud and precipitation in midlatitude cyclones: Northern versus Southern Hemisphere warm fronts. J. Climate, 25, 5135–5151, doi:10.1175/JCLI-D-11-00569.1.
Naud, C. M., D. J. Posselt, and S. C. van den Heever, 2015: A CloudSat–CALIPSO view of cloud and precipitation properties across cold fronts over the global oceans. J. Climate, 28, 6743–6762, doi:10.1175/JCLI-D-15-0052.1.
Naud, C. M., J. F. Booth, and A. D. Del Genio, 2016: The relationship between boundary layer stability and cloud cover in the post-cold-frontal region. J. Climate, 29, 8129–8149, doi:10.1175/JCLI-D-15-0700.1.
Neiman, P. J., and M. A. Shapiro, 1993: The life cycle of an extratropical marine cyclone. Part I: Frontal-cyclone evolution and thermodynamic air–sea interaction. Mon. Wea. Rev., 121, 2153–2176, doi:10.1175/1520-0493(1993)121<2153:TLCOAE>2.0.CO;2.
Pfahl, S., E. Madonna, M. Boettcher, H. Joos, and H. Wernli, 2014: Warm conveyor belts in the ERA-Interim dataset (1979–2010). Part II: Moisture origin and relevance for precipitation. J. Climate, 27, 27–40, doi:10.1175/JCLI-D-13-00223.1.
Posselt, D. J., A. R. Jongeward, C.-Y. Hsu, and G. L. Potter, 2012: Object-based evaluation of MERRA-simulated cloud physical properties and radiative fluxes during the 1998 El Niño–La Niña transition. J. Climate, 25, 7313–7327, doi:10.1175/JCLI-D-11-00724.1.
Rienecker, M. M., and Coauthors, 2011: MERRA: NASA’s Modern Era Retrospective Analysis for Research and Applications. J. Climate, 24, 3624–3648, doi:10.1175/JCLI-D-11-00015.1.
Ruf, C., A. Lyons, M. Unwin, J. Dickinson, R. Rose, D. Rose, and M. Vincent, 2013: CYGNSS: Enabling the future of hurricane prediction. IEEE Geosci. Remote Sens. Mag., 1, 52–67, doi:10.1109/MGRS.2013.2260911.
Ruf, C., and Coauthors, 2016: New ocean winds satellite mission to probe hurricanes and tropical convection. Bull. Amer. Meteor. Soc., 97, 385–395, doi:10.1175/BAMS-D-14-00218.1.
Schemm, S., I. Rudeva, and I. Simmonds, 2015: Extratropical fronts in the lower troposphere—Global perspectives obtained from two automated methods. Quart. J. Roy. Meteor. Soc., 141, 1686–1698, doi:10.1002/qj.2471.
Shapiro, M. A., and D. Keyser, 1990: Fronts, jet streams, and the tropopause. Extratropical Cyclones: The Erik Palmén Memorial Volume, C. W. Newton and E. O. Holopainen, Eds., Amer. Meteor. Soc., 167–191.
Simmonds, I., K. Keay, and J. A. T. Bye, 2012: Identification and climatology of Southern Hemisphere mobile fronts in a modern reanalysis. J. Climate, 25, 1945–1962, doi:10.1175/JCLI-D-11-00100.1.
Sinclair, V. A., 2013: A 6-yr climatology of fronts affecting Helsinki, Finland, and their boundary layer structure. J. Appl. Meteor. Climatol., 52, 2106–2124, doi:10.1175/JAMC-D-12-0318.1.
Small, R. J., and Coauthors, 2008: Air–sea interaction over ocean fronts and eddies. Dyn. Atmos. Oceans, 45, 274–319, doi:10.1016/j.dynatmoce.2008.01.001.
Wernli, H., 1997: A Lagrangian-based analysis of extratropical cyclones. II: A detailed case-study. Quart. J. Roy. Meteor. Soc., 123, 1677–1706, doi:10.1002/qj.49712354211.
Wernli, H., and H. C. Davies, 1997: A Lagrangian-based analysis of extratropical cyclones. I: The method and some applications. Quart. J. Roy. Meteor. Soc., 123, 467–489, doi:10.1002/qj.49712353811.