The Deep Propagating Gravity Wave Experiment (DEEPWAVE) was designed to quantify gravity wave (GW) dynamics and effects from orographic and other sources to regions of dissipation at high altitudes. The core DEEPWAVE field phase took place from May through July 2014 using a comprehensive suite of airborne and ground-based instruments providing measurements from Earth’s surface to ∼100 km. Austral winter was chosen to observe deep GW propagation to high altitudes. DEEPWAVE was based on South Island, New Zealand, to provide access to the New Zealand and Tasmanian “hotspots” of GW activity and additional GW sources over the Southern Ocean and Tasman Sea. To observe GWs up to ∼100 km, DEEPWAVE utilized three new instruments built specifically for the National Science Foundation (NSF)/National Center for Atmospheric Research (NCAR) Gulfstream V (GV): a Rayleigh lidar, a sodium resonance lidar, and an advanced mesosphere temperature mapper. These measurements were supplemented by in situ probes, dropsondes, and a microwave temperature profiler on the GV and by in situ probes and a Doppler lidar aboard the German DLR Falcon. Extensive ground-based instrumentation and radiosondes were deployed on South Island, Tasmania, and Southern Ocean islands. Deep orographic GWs were a primary target but multiple flights also observed deep GWs arising from deep convection, jet streams, and frontal systems. Highlights include the following: 1) strong orographic GW forcing accompanying strong cross-mountain flows, 2) strong high-altitude responses even when orographic forcing was weak, 3) large-scale GWs at high altitudes arising from jet stream sources, and 4) significant flight-level energy fluxes and often very large momentum fluxes at high altitudes.
The DEEPWAVE experiment employed extensive airborne and ground-based measurements to provide new insights into gravity wave dynamics.
The Deep Propagating Gravity Wave Experiment (DEEPWAVE) was the first comprehensive measurement program devoted to quantifying the evolution of gravity waves (GWs) arising from sources at lower altitudes as they propagate, interact with mean and other wave motions, and ultimately dissipate from Earth’s surface into the mesosphere and lower thermosphere (MLT). Research goals motivating the DEEPWAVE measurement program are summarized in Table 1. To achieve our research goals, DEEPWAVE needed to sample regions having large horizontal extents because of large horizontal GW propagation distances for some GW sources. DEEPWAVE accomplished this goal through airborne and ground-based (GB) measurements that together provided sensitivity to multiple GW sources and their propagation to, and effects at, higher altitudes. DEEPWAVE was performed over and around the GW “hotspot” region of New Zealand (Fig.1, top) during austral winter, when strong vortex edge westerlies provide a stable environment for deep GW propagation into the MLT.
DEEPWAVE airborne measurements employed two research aircraft during a core 6-week airborne field program based at Christchurch, New Zealand, from 6 June to 21 July 2014. The National Science Foundation (NSF)/National Center for Atmospheric Research (NCAR) Gulfstream V (GV) provided in situ, dropsonde, and microwave temperature profiler (MTP) measurements extending from Earth’s surface to ∼20 km throughout the core field program (see Table 2). The GV also carried three new instruments designed specifically to address DEEPWAVE science goals: 1) a Rayleigh lidar measuring densities and temperatures from ∼20 to 60 km, 2) a sodium resonance lidar measuring sodium densities and temperatures from ∼75 to 100 km, and 3) an advanced mesosphere temperature mapper (AMTM) measuring temperatures in a horizontal plane at ∼87 km with a field of view (FOV) of ∼120 km along track and 80 km cross track. AMTM measurements were augmented by two side-viewing infrared (IR) airglow “wing” cameras also viewing an ∼87-km altitude that extended the cross-track FOV to ∼900 km. A second aircraft, the DLR Falcon, participated in DEEPWAVE during the last half of the GV measurement interval. It hosted in situ dynamics and chemistry measurements and a downward-viewing aerosol Doppler lidar measuring line-of-sight winds below the Falcon, where aerosol backscatter was sufficient (see Table 2).
Ground-based DEEPWAVE measurements were likewise extensive (see Table 2). Radiosondes were launched at multiple sites, with those at three sites [two on the South Island (SI) western coast and one in the lee of the Southern Alps] providing frequent soundings during intensive observing periods (IOPs), and others launched from Tasmania and Macquarie Island coordinated with research flights (RFs) to support GW and predictability objectives in those regions. A 449-MHz wind profiler (WP) on the South Island western coast measured three-component winds continuously from ∼0.5 to ∼3–6 km. Additional instruments in the lee of the Southern Alps included 1) a ground-based AMTM measuring the horizontal temperature structure at ∼87 km, 2) a Rayleigh lidar measuring temperatures from ∼22 to 85 km, 3) two all-sky airglow imagers (ASIs) measuring airglow brightness at several altitudes from ∼87 to 96 km, and 4) a Fabry–Perot interferometer (FPI) measuring winds and temperatures centered near ∼87 and 96 km. For reference, the various airglow layers observed by the AMTMs, the ASIs, and the FPI all have full-width half maxima (FWHM) of ∼7–10 km and may vary in altitude by several kilometers about their nominal altitudes. A second Rayleigh lidar and a meteor radar measuring winds from ∼80 to 100 km were deployed at Kingston, Tasmania. Ground-based instrument sites are shown in Fig. 1 (bottom). Figure 2 shows the extent of all DEEPWAVE measurements in altitude and latitude.
DEEPWAVE began with a test flight-planning exercise from 1 to 10 August 2013 to gain experience with forecasting and flight planning and to assess the reliability of such forecasts in preparation for the real field program. This effort, which is summarized and archived online (see appendix A), was judged to be quite successful and led to confidence in the utility of a suite of forecasts and ancillary satellite products in guiding DEEPWAVE IOPs and flight plans.
The DEEPWAVE field program was supported by an extensive operations center at Christchurch International Airport that coordinated all logistical and measurement activities (see appendix B). Forecasting and flight planning was supported by a suite of global, mesoscale, and regional models that proved to be highly valuable and often quite accurate on shorter time scales for final flight planning (see Table 3). These models are now being applied in concert with DEEPWAVE data analysis efforts to answer the science questions posed in Table 1. To aid DEEPWAVE research, a comprehensive DEEPWAVE data archive and management plan has been developed (see appendix A).
GWs, or buoyancy waves, for which the restoring force is due to negatively (positively) buoyant air for upward (downward) displacements, play major roles in atmospheric dynamics, spanning a wide range of spatial and temporal scales. Vertical and horizontal wavelengths, λz and λh, respectively, for vertically propagating GWs are dictated by their sources and propagation conditions and range from meters to hundreds and thousands of kilometers, respectively, with typical scales increasing by ∼10 times or more from the troposphere to the MLT. Intrinsic frequencies (i.e., with respect to the local flow) vary from the inertial frequency to the buoyancy frequency. GWs at lower frequencies dominate the energy spectra, but higher-frequency GWs have larger vertical group velocities and contribute disproportionately to vertical transports of energy and momentum. As a result, smaller-scale GWs (λh ∼ 10–200 km) have larger impacts on atmospheric circulation, weather, and climate, but their effects are much more challenging to quantify. GW influences typically increase with altitude because decreasing density implies increasing GW amplitudes and effects. Large GW amplitudes drive nonlinear (NL) wave–wave and wave–mean flow interactions, instabilities, turbulence, and energy and momentum deposition that result in a strong evolution of the GW spectrum with altitude. These complex dynamics, and their significant dependence on GW sources and the environments through which they propagate, pose major challenges for their parameterizations in global weather and climate models.
Scientific interests and societal needs have motivated many previous studies of GWs from the stable boundary layer and troposphere, through the stratosphere and mesosphere, and into the thermosphere. Among the more important of these are the following:
GWs pose hazards to people and property; examples include sometimes severe downslope winds and severe turbulence at airline flight altitudes;
GWs exhibit a wide range of dynamics and effects that play important roles in atmospheric weather and climate from the surface into the MLT, but many of these are poorly understood at present;
GW motions are incompletely resolved both in global satellite observations and in global numerical weather prediction (NWP) and climate models, and so their effects in large-scale weather and climate models must be parameterized; and
Inadequate understanding and characterization of GW dynamics and effects have resulted in parameterizations of their effects in NWP and climate models that are acknowledged to have major deficiencies.
The importance of GWs in multiple atmospheric processes has led to thousands of papers dealing with diverse GW topics including 1) sources; 2) propagation and refraction in variable environments; 3) linear and nonlinear behavior; 4) wave–wave and wave–mean flow interactions; 5) instabilities and turbulence due to large GW amplitudes and superpositions; 6) energy, momentum, and tracer transports; 7) parameterizations of GW effects in large-scale (LS) models; and 8) GW influences on other processes such as convection, cloud microphysics, chemical reactions, and plasma dynamics and instabilities in the ionosphere. Many other papers have addressed important GW roles in oceans, lakes, other planetary atmospheres, and stellar interiors.
The scope of GW dynamics and roles is reflected in the many seminal papers, reviews, and books describing these various processes. Examples of those addressing atmospheric GW topics of most relevance to DEEPWAVE science include the following:
GW linear dynamics, propagation, conservation properties, and fluxes (Hines 1960; Eliassen and Palm 1961; Bretherton 1969a,b; Booker and Bretherton 1967; Gossard and Hooke 1975; Smith 1980; Nappo 2013);
Below we provide an overview of previous research on atmospheric GWs, focusing on airborne measurement programs, but also noting contributions by other ground-based, in situ, and satellite measurements (a number of which were employed during DEEPWAVE). Numerous modeling studies have likewise addressed GW sources, propagation, linear and nonlinear dynamics, and their various effects. However, we will restrict our overview to those efforts performed specifically for comparisons with observational data or which offer a global perspective on resolved GW sources, propagation, and effects.
The earliest studies of mountain waves (MWs) in the 1930s employed balloons and gliders to sample MW flows in North Africa and Europe (e.g., Queney 1936a,b; Küttner 1938, 1939; Manley 1945). These observations provided key insights into the structure of MWs and lee waves and, together with the Sierra Wave Project (see below), motivated initial theoretical advances (e.g., Queney 1947; Scorer 1949; Long 1953, 1955; see also Grubišić and Lewis 2004). Other observations of plasma motions in the ionosphere (now called traveling ionospheric disturbances) motivated the seminal paper by Hines (1960) that provided the theoretical framework for GW propagation throughout the atmosphere. Brief overviews of subsequent GW research using ground-based, in situ, and satellite measurements, accompanying more recent airborne programs, and employing mesoscale and global modeling are provided below.
Ground-based, in situ, and satellite measurements.
Ground-based and in situ measurement capabilities have improved dramatically since the earliest MW studies. Radiosondes have provided evidence of GW sources, scales, amplitudes, intrinsic properties, and fluxes from the surface into the middle stratosphere for many years (e.g., Tsuda et al. 1994; Allen and Vincent 1995; Sato and Dunkerton 1997; Sato and Yoshiki 2008; Geller et al. 2013). Stratospheric superpressure balloon measurements have likewise defined GW intrinsic properties and momentum fluxes (MFs) in the lower stratosphere and, in particular, their intermittency and potential for infrequent, but very strong, GW events to contribute a large fraction of the total momentum flux (e.g., Hertzog et al. 2008; Plougonven et al. 2008, 2013). Rocketborne falling spheres and newer ionization gauges, lidars, and other probes have measured winds, temperatures, and turbulence from ∼30 to 100 km and enabled studies of energy dissipation rates due to GW breaking, MW filtering during a stratospheric warming, and anomalous MLT mean structure accompanying strong planetary waves (PWs) in the Southern Hemisphere and other dynamics (e.g., Rapp et al. 2004; Wang et al. 2006; Goldberg et al. 2006).
Multiple types of radars have quantified GW amplitudes, scales, spectral character, momentum fluxes, and evidence of various interaction and instability processes from the troposphere to the MLT for about five decades (e.g., Gossard et al. 1970; Atlas et al. 1970; Woodman and Guillen 1974; Sato and Woodman 1982; Vincent and Reid 1983; Balsley and Garello 1985; Fritts and Rastogi 1985; Fritts and Vincent 1987; Smith et al. 1987; Tsuda et al. 1989, 1990; Sato 1994; Thomas et al. 1999; Pavelin et al. 2001; Luce et al. 2008). Rayleigh and resonance lidars have likewise contributed to the definition of GW properties via measurements of temperatures, winds, and/or metallic species densities from very low altitudes to ∼100 km or above (e.g., Chanin and Hauchecorne 1981; Gardner and Voelz 1987; She et al. 1991; Whiteway and Carswell 1994; Williams et al. 2006; Duck et al. 2001; Alexander et al. 2011; Lu et al. 2015). Other optical instruments, especially ASIs and the newer AMTMs, provide valuable information on GW horizontal wavelengths, orientations, phase speeds, and amplitudes, sometimes at multiple altitudes, that contribute greatly to quantification of GW character, propagation, and potential for instability and mean flow interactions (e.g., Gavrilov and Shved 1982; Taylor et al. 1995; Taylor and Hapgood 1988; Hecht et al. 1997, 2001; Walterscheid et al. 1999; Nakamura et al. 2003; Smith et al. 2009; Pautet et al. 2014; Hecht et al. 2014; Fritts et al. 2014).
Multi-instrument measurement programs performed at facilities having extensive ground-based instrument capabilities, such as those that often accompany large radar and/or rocket facilities, have made especially valuable contributions to GW studies. This is because no single instrument can define all of the atmospheric properties and spatial and temporal variability needed to fully quantify the local GW field. Examples of these facilities include the Arctic Lidar Observatory for Middle Atmosphere Research in Norway (69.3°N); the Poker Flat Research Range in Alaska (65.1°N); the Bear Lake Observatory in Utah (42°N); the middle- and upper-atmosphere (MU) radar in Japan (34.9°N); the National Atmospheric Research Laboratory in India (13.5°N); the Equatorial Atmosphere Radar (EAR) in Indonesia (0°); the Jicamarca Radio Observatory in Peru (12°S); the Andes Lidar Observatory in Chile (30.2°S); Buckland Park in Australia (35°S); the Davis (Australia) and Syowa (Japan) Antarctic stations (68.6° and 69°S, respectively); and additional facilities having valuable correlative instrument capabilities in Antarctica, Argentina, Australia, Brazil, Canada, China, France, Germany, India, Puerto Rico, Sweden, and elsewhere.
Measurements of radiances and inferred temperatures by multiple satellite instruments employing limb, sublimb, and nadir viewing have been used to estimate GW temperature variances and momentum fluxes from the lower stratosphere into the MLT for many years. These have provided enticing insights into GWs arising from various sources. In many cases, however, satellite measurements exhibit strong observational constraints because of line-of-sight averaging or weighting-function depths comparable to, or greater than, the smaller, but important, GW scales. Such measurements nevertheless reveal the larger-scale responses to multiple sources, define the global hotspots of GW activity and their seasonal variations, and on occasion capture very strong GW responses under ideal viewing conditions (e.g., Dewan et al. 1998; Eckermann and Preusse 1999; Ern et al. 2004; Eckermann et al. 2007; Wu and Eckermann 2008; Alexander et al. 2009, 2010; Eckermann and Wu 2012; Geller et al. 2013; Hendricks et al. 2014). Figure 3 shows the measurement capabilities of various satellite viewing geometries compared to DEEPWAVE and the GW wavelengths expected to account for the major GW momentum fluxes. Nadir measurements [e.g., Atmospheric Infrared Sounder (AIRS)] extend to relatively small GW λh, but these often fail to capture the smaller-λh GW responses inferred to contribute the largest local momentum fluxes (Fritts et al. 2002, 2014; Hertzog et al. 2012). Nadir measurements also often fail to capture larger-λh GWs when the GW λz is comparable to or smaller than the depth of the weighting function (e.g., Eckermann et al. 2009; Gong et al. 2012).
More recent airborne measurement programs.
The next significant airborne measurement program following those in the 1930s was the Sierra Wave Project. This project employed two gliders in 1951/52 and two gliders and two powered aircraft in 1955, together with radiosondes and ground measurements, and yielded a significantly improved understanding of MW structure and related theoretical advances (Grubišić and Lewis 2004). Subsequent MW and lee-wave studies over the Rockies in the 1960s and 1970s used improved instrumentation aboard various aircraft to sample the MW, lee wave, and turbulence environments accompanying MW breaking. These provided more complete descriptions of the flow structures and evolutions and motivated initial modeling of these events (e.g., Kuettner and Lilly 1968; Lilly and Kennedy 1973; Brinkmann 1974; Clark and Peltier 1977; Lilly 1978; Klemp and Lilly 1978; Lilly et al. 1982).
More recent MW airborne studies benefitted from further expanded measurement capabilities, including dropsondes, GPS positioning, and/or the MTP, ground-based instruments, and associated modeling, for example, over the Welsh mountains (Whiteway et al. 2003); over the Alps during the Alpine Experiment (ALPEX), Pyrénées Experiment (PYREX), and Mesoscale Alpine Programme (MAP) (e.g., Bougeault et al. 1990, 2001; Smith et al. 2002; Doyle and Smith 2003; Doyle and Jiang 2006); over the Sierra Nevada during the Terrain-Induced Rotor Experiment (T-REX) (e.g., Grubišić et al. 2008; Smith et al. 2008; Doyle et al. 2011); and elsewhere (e.g., Brown 1983; Whiteway et al. 2003; Doyle et al. 2005). Additional airborne studies explored the influences of MWs on the formation of polar stratospheric clouds at Arctic and Antarctic latitudes (e.g., Carslaw et al. 1998; Eckermann et al. 2006b).
Other airborne programs targeted more general GW responses. The Global Atmospheric Sampling Program (GASP) employed commercial aircraft for global in situ measurements that enabled comparisons of GW responses to various sources (e.g., Nastrom and Fritts 1992; Fritts and Nastrom 1992). The Airborne Lidar and Observations of Hawaiian Airglow 1990 (ALOHA-90) and the Airborne Lidar and Observations of Hawaiian Airglow/Arctic Noctilucent Cloud Campaign 1993 (ALOHA/ANLC-93) measurement programs employed a lidar and ASI to sample GWs extending from the stratosphere into the MLT (Hostetler et al. 1991; Hostetler and Gardner 1994; Swenson et al. 1995). Several airborne measurements also provided evidence of GWs generated by deep convection and their momentum fluxes at flight altitudes (e.g., Kuettner et al. 1987; Pfister et al. 1993; Alexander and Pfister 1995).
Mesoscale and global modeling of GWs.
Modeling capabilities for mesoscale and global GW studies have improved dramatically in recent years because of ever-increasing computational resources. As a result, various models have been employed in support of GW measurement programs and to identify GW sources and key dynamics spanning larger spatial scales. Mesoscale models have aided the interpretation of MAP, T-REX, and other airborne MW programs, been employed for intermodel comparisons for several events (e.g., Dörnbrack et al. 2001; Smith et al. 2002; Doyle and Smith 2003; Doyle and Jiang 2006; Doyle et al. 2000, 2011), and assessed the dynamical responses and resolution dependence of airflow over small islands (e.g., Vosper 2015). Global forecast and research models (REs) now achieve spatial resolutions of ∼25 km or better that enable direct modeling, rather than parameterization, of GWs extending to horizontal scales as small as ∼100 km. As examples, Yamashita et al. (2010) showed that the European Centre for Medium-Range Weather Forecasts (ECMWF) T799 model described GWs having λx > 100 km that agreed reasonably with a much higher-resolution Weather Research and Forecasting (WRF) Model simulation and AIRS observations of GWs due to a typhoon, exhibited similar GW variance distributions as the Microwave Limb Sounder (MLS), but underestimated GW amplitudes by ∼2 times compared to Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) measurements. Shutts and Vosper (2011) employed the Met Office (UKMO) and ECMWF global models and a very-high-resolution (4 km) unified model to examine the MW energy and momentum fluxes over the southern Andes. They found a peak in the fluxes at λx ∼ 400 km, with approximately half the fluxes at λx < 200 km. Sato et al. (2012) used a high-resolution middle-atmosphere general circulation model (GCM) to explore the stratospheric dynamics of MWs having λx ∼ 200 km and larger arising from the southern Andes. They found that the MWs refract strongly into the polar vortex because of horizontal wind shears and yield downward-propagating responses below ∼40 km because of nonlinear dynamics at higher altitudes. Similar improvements in characterization of MW and more general GW influences at higher spatial resolution were also found to occur in the Community Atmosphere Model, version 4.0 (CAM4), and the Whole Atmosphere Community Climate Model (WACCM) (Bacmeister et al. 2014; Liu et al. 2014).
FIELD PROGRAM AND EPO OVERVIEW.
The DEEPWAVE field program was complex and was made possible by the participation of a large number of individuals from the NSF principal investigator (PI) teams, NCAR, Naval Research Laboratory (NRL), DLR, National Institute of Water and Atmosphere Research (NIWA), Australian Antarctic Division (AAD), and other colleagues and students in New Zealand, Australia, and Austria. Altogether, over 100 people contributed to various aspects of the program. DEEPWAVE participants and their roles are listed in appendix D. The various tasks included aircraft logistics, operations and maintenance, ground-based instrument installations, weather forecasting and updates, flight planning and debriefs, personnel scheduling, education and public outreach (EPO) activities, and local outreach. Most activities were performed during daytime, but, because of the extensive use of the new GV lidars and imagers, most research flights and all ground-based optical measurements were performed at night. The major components of the program are discussed further below.
Weather forecasting, briefings, and updates.
Daily weather forecasting began each morning, with efforts coordinated by a lead forecaster and contributed to by a team including scientists, students, and NIWA staff using local weather observations and forecasts and mesoscale and global forecast models (FCs; see Table 3). The forecast models often proved to be quite accurate on short time scales and hence very valuable for these purposes. The focus was on events having GW responses expected to penetrate into the stratosphere and MLT and weather impacting GV operations. Weather briefings occurred each day at 1300 local time (LT) [0100 univeral time (UT)] and typically reviewed the weather for that day (if there was a research flight scheduled) and 1–3 days out for flight planning purposes. On days having research flights scheduled, an additional weather update was also provided ∼2 h before flight departure.
Flight planning typically involved submission of flight proposals by individuals or teams designed to address specific DEEPWAVE science questions. Occasionally, flight plans looked farther ahead and anticipated a combination of flights, for example, predictability and verification or successive sampling of a multiday event. Often, alternative flight proposals were merged to optimize the expected results and/or address common measurement goals. A subcommittee of scientists that changed weekly determined the final flight plan in the event of competing proposals. The selected flight plan was then sent to the Earth Observing Laboratory (EOL) team for review and feedback.
Research flights and large-scale context.
All research flights [RFs and Falcon research flights (FFs)] for the GV and the Falcon were part of an IOP ranging from 1 to 4 days to facilitate coordination with ground-based measurements. GV flight durations ranged up to ∼9 h and flight distances ranged up to ∼8,000 km. Falcon flights had maximum durations and lengths of ∼3.5 h and ∼3,000 km, respectively. The large majority of RFs and FFs were performed at high altitudes, ∼12–13.7 km for the GV and ∼10–11 km for the Falcon. For the GV, this was done for fuel efficiency and because the GV lidars were not allowed to operate at lower altitudes. Both aircraft also performed a number of flight segments at lower altitudes to sample interesting events on various occasions. MW flights targeted strong and weak forcing to span a range of responses at higher altitudes. The IOPs, dates, research targets, and flight summaries for all RFs flown during DEEPWAVE are listed in Table 4.
IOPs are shown in the context of the large-scale ECMWF horizontal winds from 0 to 80 km in Fig. 4 (top). The dominant feature is the polar night jet with a maximum wind often exceeding 100 m s−1 at ∼50–60 km that is presumably modulated in strength by PWs on time scales of ∼5–10 days. The poleward jet associated with frontal systems exhibits episodic maxima of ∼30–50 m s−1 at ∼8–12 km on similar time scales. Also seen in the second half of July are two intervals in which the polar night jet decreases, first to ∼60 m s−1 (∼15–20 July) and then to ∼30 m s−1 (beginning ∼29 July). These intervals accompany significant enhancements in the zonal wavenumber 1-PW amplitude that yield both a weak stratospheric warming and westward wind perturbations that account for the weaker ECMWF winds at these times.
As noted above, DEEPWAVE was supported by extensive ground-based measurements. The WP operated continuously from 28 May to 28 July. Radiosondes were launched daily at Hokitika from 24 May to 18 July, at Lauder from 13 June to 1 August, at a higher cadence during IOPs at these sites and at Haast, and at Hobart and Macquarie Island to support flights or predictability objectives in those areas. The AMTM, ASIs, and FPI at Lauder and Mt. John Observatory (MJO) performed routine nighttime observations spanning the DEEPWAVE core measurement interval. The DLR lidar at Lauder operated from 19 June to 6 November. The Kingston lidar operated in coordination with GV flights over Tasmania and the Tasman Sea, and the meteor radar at Kingston operated continuously beginning 10 June. The altitudes sampled by these various instruments are shown with vertical bars in Fig. 2. Additional ground-based IOPs were designated on nights for which interesting responses were observed that correlated with the forecast models and measurements at lower altitudes. These events are listed in Table 5.
Four examples of radiosonde measurements at Hokitika and Lauder relevant to specific cases discussed further below are shown in Fig. 5 (bottom). Shown in Fig. 5 (top) are RF-mean or nightly mean temperatures obtained with the GV airborne and Lauder Rayleigh lidars for each available measurement over South Island. These illustrate some of the diversity of GW propagation environments from the surface to 60 km during the DEEPWAVE program.
DEEPWAVE EPO efforts had two primary objectives: 1) to increase the awareness of students in kindergarten–grade 12 of the field of atmospheric science by exposing them to research methods through engaging presentations and interactions with early-career scientists and 2) to increase public awareness of the DEEPWAVE science objectives and societal benefits on an international level. The program consisted of targeted student enrichment activities including 10 presentations to 565 middle and high school students; Internet-based outreach efforts that included 11 educational web pages with 2,000 views in a 104-day period, 15 Facebook posts, blog posts, and tweets from postdoctoral scholars in the field; a research aircraft public open house with over 250 visitors; media visits resulting in several high-profile pieces broadcast in New Zealand; and various printed information. Additionally, 26 undergraduate and graduate (grad) students from eight organizations and universities were directly involved with DEEPWAVE research and operations, gaining valuable experience in observational fieldwork.
INITIAL MEASUREMENTS AND RESULTS.
Initial DEEPWAVE data analysis efforts are addressing a number of topics and yielding a variety of tantalizing results. Example “first results” that will be discussed briefly below include 1) strong variability of MW energy fluxes among, and within, the various MW flights; 2) evidence of MW breaking at flight altitudes; 3) predictability targeting and influences; 4) MWs arising from weak forcing attaining large amplitudes at higher altitudes; 5) strong three-dimensional (3D) MW responses at high altitudes over Auckland Island; 6) GWs in the stratosphere apparently generated within the jet stream; 7) responses to weak MW forcing over several days that yielded intermittent MW breaking in the MLT; and 8) comparisons of DEEPWAVE measurements with model forecasts and AIRS temperature observations.
MW flight-level responses and predictability.
An initial assessment of MW propagation employing GV flight-level (FL) MW energy flux estimates 〈p'w'〉 (where p' and w' are the in situ GV measurements of pressure and vertical velocity perturbations, and brackets denote horizontal averaging) for each MW RF is shown in Fig. 6 (top). WRF Model estimates of these fluxes at 4, 12, and 30 km for initial conditions specified by the ECMWF Integrated Forecasting System (IFS) model are shown in Fig. 6 (bottom) and were computed following Kruse and Smith (2015). The WRF GW energy flux maxima typically accompany frontal systems that bring strong lower-level flow over South Island. RF energy fluxes are positive (negative) for upward (downward) MW propagation, suggesting strong variability in MW strength and propagation within individual MW events. Modeled energy fluxes suggest variable MW propagation and dissipation at higher altitudes depending on the MW forcing strengths and propagation environments. The numbered circles on the x axis in Fig. 6 (bottom) are the RFs for which computed energy fluxes are shown in the top panel.
One of the strongest MW events during DEEPWAVE occurred during RF12 on 29 June. The GV flew a box pattern with repeated flight segments over Mt. Aspiring and Mt. Cook. Data from segments 14 and 22 along Mt. Aspiring flight-track 2 (MA2; see Fig. 1) are shown in Fig. 7. Most notable are the very different responses separated by only 1.5 km in altitude. At 12.2 km, the along-track wind accelerated to 25 m s−1 and then decelerated to 12 m s−1 over the high terrain. At 13.7 km, the disturbance was stronger and decelerated to ∼0 m s−1, which is expected to accompany wave breaking. The vertical velocity fields (top panel) were also different at the two levels. At 12.2 km, these mostly showed a quasi-periodic train of small-scale (SS) waves downwind of the highest orography. These were likely trapped waves having small energy and momentum fluxes. At 13.7 km, a burst of high-frequency turbulence occurred over the high terrain, likely accompanying wave breaking.
An example of the predictability component of DEEPWAVE is illustrated in Fig. 8 for 13 June 2014 (RF3). The Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS) forecast and adjoint models (Amerault et al. 2008; Doyle et al. 2014) were used to compute the forecast sensitivity to the initial state, and these regions of high sensitivity were targeted for additional dropwindsonde (DWS) observations. As an example, the color shading in the Tasman Sea (Fig. 8a) highlights the upstream regions where the 24-h COAMPS forecast kinetic energy in the lowest 1 km above the surface in the gray box is most sensitive to the initial-state 700-hPa u-wind component. The sensitive regions most strongly influence MW launching and amplitudes over South Island 24 h later. Green dots along the flight track show the DWS deployments for this assessment. The evolved perturbations (24 h) based on the sensitivity scaled to a maximum of 1 m s−1 at the initial time (Fig. 8b) exhibit a maximum over South Island with growth of ∼10 times for the u-wind component perturbations in this case. The GV flight the following day on 14 June served as the verification flight to assess the degree to which the targeted DWSs improve the prediction of MWs over South Island.
MW responses in the stratosphere and MLT accompanying weak surface forcing.
A major surprise during the DEEPWAVE field program was the observation of large-amplitude, breaking MWs in the MLT on a night that the flight planning team had elected not to fly a MW mission because of the forecast of weak MW forcing conditions. This quickly sensitized the team to conditions for which weak surface forcing can nevertheless lead to large MW amplitudes at high altitudes potentially because of the largely linear MW propagation and an absence of instabilities and breaking in the stratosphere, in contrast to strong forcing events (e.g., Fig. 7).
One example of these MW dynamics was observed during RF22 (13 July), a case having weak cross-mountain flow and MW forcing but favorable vertical propagation conditions with strong eastward winds through the stratosphere and above. A subset of observations from the GV lidars and the AMTM and wing cameras is shown in Figs. 9 and 10. Figures 9c and 9d show two successive cross sections along Mt. Cook flight-track 1 (MC1; see Fig. 1) of stratospheric temperatures from 20 to 60 km and sodium density perturbations obtained with the GV lidars. Figures 9a and 9b show corresponding cross sections of sodium mixing ratios for the same two cross sections. Rayleigh lidar temperatures are shown together with perturbation temperature contours from the ECMWF IFS that contributed significantly to DEEPWAVE flight planning and were interpolated to the GV location in space and time for this comparison. Note, in particular, the very close agreement of the MW scales and phase structures between the GV lidar data and a composite of IFS analyses and 1-h predictions, including the MW growth with altitude and the changing MW vertical wavelength λz accompanying the stronger winds extending to ∼60 km and above. The major differences are that the IFS results underpredict (by ∼2–5 times) the large-scale MW amplitudes, and they appear not to capture some of the smaller-scale MWs contributing to the lidar temperature perturbations above ∼40 km.
At higher altitudes, sodium mixing ratios measured by the GV sodium lidar reveal very large vertical displacements because of the smaller-scale MWs and other GWs. Peak-to-peak displacements as large as ∼3–8 km imply these smaller-scale GWs have T' ∼ 5–20 K or more and very large momentum fluxes. Rough estimates based on the observed GW scales and amplitudes measured on RF22 are ∼100–500 m2 s−2 or larger, which are ∼1–2 decades larger than the expected mean values at these altitudes (e.g., Fritts and Alexander 2003; Fritts et al. 2014).
An example of a combined GV AMTM and wing camera cross-mountain image of OH airglow brightness is shown in Fig. 10a for the vertical cross section shown in Fig. 9d. This reveals the same λh ∼ 200–300-km MW seen by the Rayleigh lidar and multiple additional MWs and other GWs at smaller horizontal wavelengths, λh ∼ 30–80 km, at ∼87 km. Additional horizontal cross sections of the IFS horizontal divergence at 2 hPa (∼43 km) at 0900 UT (Fig. 10b), and AIRS brightness temperature (radiance) perturbations at 2 hPa (Figs. 10c and 10d) suggest that the GV imagers observed the upward extension of the larger- and smaller-scale MW field seen by the GV lidars. The IFS vertical and horizontal cross sections in Figs. 9 and 10 captured both the vertical and horizontal structures of the large-scale MW and the associated trailing waves (TWs) for this event quite well.
Jet stream GW responses.
Jet streams also represented a significant source of larger-scale GWs predicted by the NWP models during DEEPWAVE. Thus, several flights over the Southern Ocean (SO) specifically targeted these GWs. An example of one cross section through an apparent jet-generated GW, and its prediction by the IFS model, is shown in Fig. 11. As seen in the MW observations on RF22 (Fig. 9), Rayleigh lidar temperature measurements again reveal surprising agreement in the GW spatial structures and refraction with altitude with the changing environment. But again, GW amplitudes tended to be underestimated by the model fields interpolated to the GV locations and measurement times by up to ∼2 times or more. While our initial comparisons employed only the IFS model, we note that other global and regional models supporting DEEPWAVE achieved similar successes in characterizing GW responses to the various sources for which the GW spatial scales were well resolved. These comparisons will be highlighted in future papers.
MW responses over small islands.
Given the potentially strong MW responses at higher altitudes to flow over small SO island orography (e.g., Alexander and Grimsdell 2013), several DEEPWAVE flights overflew SO islands when deep MW forcing was expected. An example of these measurements over and in the lee of Auckland Island by the GV imagers on RF23 with strong surface flow from the northwest is shown in Fig. 12a. This image reveals ship wave temperature structure at ∼87 km having a dominant λh ∼ 40 km and evidence of a stronger trailing wave response to the north, likely resulting from filtering by the intervening winds. The GV AMTM also revealed a peak amplitude of T' ∼ 20 K or larger immediately in the lee of Auckland Island. A MW response computed with the NRL Fourier–Ray (FR) linear model (Eckermann et al. 2006a) using upstream forcing profiles from NWP models and GV DWSs for this day captures some key features of the observed MLT MW field (wavelength and approximate amplitude) in Fig. 12b. Three GV passes over Auckland Island ∼3–4 h later revealed breaking and instabilities that destroyed the MW field at ∼87 km. As for RF22 (Fig. 9), the large amplitude and small λh of this response also imply a very large, but spatially localized, MW momentum flux.
MW breaking observed on 21 June.
Finally, we illustrate ground-based MW observations that alerted the team to the importance of weak forcing events at high altitudes. This event occurred near the end of an interval of sustained weak MW forcing first observed on RF7 on 19 June to the southeast of South Island (e.g., AIRS images show continuous large-scale MW and trailing wave responses in the middle stratosphere throughout this interval). Three images of OH (∼87 km) temperatures obtained with the AMTM at Lauder at 30-min intervals are shown in Figs. 13a–c. These reveal relatively stationary MWs exhibiting λh ∼ 10–70 km, phases oriented largely north (N)–south (S), and maximum T' > 20 K. The images also exhibit pronounced “sawtooth” patterns in the temperature fields seen as gradual decreases in temperature from warm to cold followed by sudden transitions back to warm in progressing from east to west that are indicative of GW nonlinearity, including steepening, overturning, and breaking. The Lauder AMTM images cover only a portion of the larger-scale MW response also seen simultaneously by the Lauder ASI (Fig. 13d) and by AIRS ∼2 h later (Fig. 13e), both of which indicate that these MWs extend well upstream and downstream of the orographic source. They also appear for only ∼1 h on this day, suggesting that filtering by variable winds at these or lower altitudes must modulate these MLT responses, given that the AIRS responses are essentially continuous throughout ∼4 days. As noted for the MWs seen on RF22 and RF23 discussed above, these strong breaking MWs over Lauder must likewise have very large momentum fluxes extending in this case over a large area.
The DEEPWAVE field program was successfully executed because of the major efforts by many people and organizations (see appendix D) and an unprecedented and comprehensive suite of airborne and ground-based instrumentation (see Figs. 1 and 2; Table 2). DEEPWAVE was also the first research program to systematically measure GW dynamics arising from various sources in the troposphere and stratosphere to altitudes of dissipation extending up to ∼100 km. DEEPWAVE measured GWs generated by orography, jet streams, frontal systems, deep convection, and secondary generation processes and spanned a range of forcing, propagation, and dissipation conditions. The various DEEPWAVE measurements led to the initial identification of a large number of anticipated research targets (see Tables 4 and 5) and also yielded a number of surprises. These include the following:
highly variable MW energy fluxes at flight altitudes for weak and strong forcing;
the interruption of vertical MW propagation and resulting absence or strong attenuation of MWs at higher altitudes in cases of strong forcing and breaking in the stratosphere;
the detection of secondary GW generation in regions of strong MW breaking;
the potential for MWs due to weak forcing to penetrate to very high altitudes and achieve very large amplitudes and momentum fluxes;
the penetration of MWs having very small horizontal wavelengths of λh ∼ 10–30-km to ∼80–100-km altitudes under weak forcing conditions;
the generation of ship wave patterns due to small islands at small scales and large amplitudes in the MLT;
the ubiquitous presence of larger-scale GWs from nonorographic sources in the stratosphere and mesosphere;
strong and coherent responses to orography and other GW sources at larger scales that were often remarkably consistent with the predictions of mesoscale and global models employed in DEEPWAVE forecasting and analysis efforts; and
regions of initial condition sensitivity diagnosed from adjoint models were nearly always in areas of very active weather including jet streaks, fronts, and convection that played a prominent role in GW launching the following day.
Initial conclusions from our DEEPWAVE measurements include confirmation of 1) the important roles of multiple sources of larger-scale large-amplitude GWs (λh ∼ 200–300 km or larger) that readily penetrate to higher altitudes; 2) the frequent refraction of larger-scale GWs into the polar vortex, including large-scale trailing MWs; 3) the importance of environmental wind and temperature fields in defining their evolving characteristics and the altitudes to which they penetrate; and 4) links between GW sources and characteristics at higher altitudes. Initial DEEPWAVE observations and analyses also suggest that smaller-scale GWs 1) arise preferentially from orography, deep convection, and secondary GW generation in the stratosphere; 2) readily penetrate into the stratosphere and mesosphere under suitable propagation conditions; 3) are less likely to exhibit strong refraction into the polar vortex; 4) often attain very large amplitudes at higher altitudes; and 5) typically dominate the total momentum fluxes in these regions.
DEEPWAVE measurements also have implications for modeling of GWs arising from various sources. The high-resolution mesoscale and global models that supported DEEPWAVE appear to capture important aspects of MW generation and propagation when the MW scales are well resolved. The global models also perform well in defining the character of GW responses to various sources for larger-scale GWs. Compared to FL and lidar stratospheric measurements, however, these models typically underestimated the measured GW amplitudes in the stratosphere and above.
Specific questions suggested by initial DEEPWAVE observations and modeling that further studies will attempt to resolve include the following:
How do environmental conditions modulate the deep propagation of GWs from various sources?
What roles do nonlinear dynamics and instabilities play in interrupting GW penetration to higher altitudes?
Which GW sources and spatial scales contribute most to total momentum fluxes as a function of altitude, and can these be quantified by current models and satellite measurements?
Which GW sources and spatial scales account for the largest latitudinal transport of momentum?
What dynamics account for the spatial and temporal intermittency of energy and momentum fluxes at different altitudes?
What are the dynamics and consequences of multiscale GW superpositions throughout the lower and middle atmosphere?
Our DEEPWAVE research team is actively pursuing multiple research topics and we anticipate that a number of results will be available to the community in the near future.
The DEEPWAVE program was made possible through financial and/or in-kind support from many U.S. and international organizations, including NSF, NCAR/EOL, NRL, DLR, NIWA, AAD, the New Zealand MET Service, PAE Ltd., ECMWF, and NOAA–NCEP. Many individuals also contributed greatly in various ways (see appendix D). Financial support for U.S. participants was provided by NSF under Grants AGS-1261619 and AGS-1338646 (GATS); AGS-1338655 (Yale University); AGS-1061892 and AGS-1338666 (Utah State University); AGS-1338309 and AGS-1343097 (Boston University); and AGS-1338557 (CPI). The NRL investigators (Doyle, Eckermann, Reinecke, and Reynolds) are supported by the Chief of Naval Research through the NRL Base Program (PE 0601153N). We are indebted to Kerry Chuck and Phil Ambler for their very able assistance with airport logistics in Christchurch and to John Robinson and his very helpful NIWA staff at Lauder.
DATA MANAGEMENT, FIELD CATALOG, AND ACCESS.
Development and maintenance of a comprehensive data archive is a critical step in meeting the scientific objectives of DEEPWAVE. The goal is to make the dataset and documentation available to the scientific community as soon as possible following the DEEPWAVE field program via a permanent DEEPWAVE web page. This web page is available online (at www.eol.ucar.edu/field_projects/deepwave). The web page includes information on operations, logistics, facilities, instrumentation, mailing lists, meetings and presentations, education and outreach, and data management throughout the DEEPWAVE program.
EOL will maintain a DEEPWAVE data management portal that provides a long-term archive and access to DEEPWAVE datasets for the DEEPWAVE PIs and the scientific community (http://data.eol.ucar.edu/master_list/?project=DEEPWAVE), including the main archive at EOL and DEEPWAVE archives at other organizations. EOL will also ensure that “orphan” datasets (i.e., smaller regional and local networks) will remain available through the EOL DEEPWAVE archive. DEEPWAVE data will be available to the scientific community through a number of designated DEEPWAVE Data Archive Centers (DDACs), coordinated by NCAR/EOL and the main archive website noted above.
General users will have free and open access to all DEEPWAVE data, subject to procedures at the various DDACs and the terms of the DEEPWAVE data policy. Key elements of this policy include the following: 1) timely submission of preliminary and final data to an archive; 2) exclusive access to the DEEPWAVE datasets by DEEPWAVE science team members from 29 January 2015 to 29 January 2016; 3) full public data access on 1 February 2016; 4) prompt notification of data providers and offers of coauthorship or attribution by data users; and 5) proper dataset citation using digital object identifiers (DOIs) and acknowledgment of DEEPWAVE data including the project name, data providers, and funding agencies.
An online DEEPWAVE field catalog (http://catalog.eol.ucar.edu/deepwave) was hosted by EOL during the DEEPWAVE field program to support mission planning, product displays, documentation of activities, and “browse” tools for use in postfield analyses. The DEEPWAVE field catalog can access and replay flight missions and supports real-time mission coordinator and geographical information system (GIS) catalog maps display tools. The 2013 DEEPWAVE flight planning exercise is documented online ("http://catalog.eol.ucar.edu/deepwave_2013"). An example of the field catalog maps display is shown for reference in Fig. A1.
DEEPWAVE FIELD OPERATIONS.
Operational support for the DEEPWAVE field program included several major components. The DEEPWAVE Operations Center and aircraft support were located at the U.S. Antarctic Program (USAP) Christchurch International Airport (CHC). Major logistical support was provided by PAE Ltd., the local New Zealand contractor funded by NSF. The project occupied two buildings and adjacent ramp space and served as the focus for aircraft support, forecasting and in-field science analyses, logistics, and communications. Broadband Internet access facilitated communications with remote participants in New Zealand and elsewhere.
The major deployments of ground-based instruments and aircraft for DEEPWAVE occurred over the period from late May to early August 2014, though several instruments or capabilities remained up to several months longer at Lauder. More information on these efforts and related activities can be found online (www.eol.ucar.edu/field_projects/deepwave).
The science leadership, operations coordinators, and facility project managers were key components of the DEEPWAVE in-field management team. DEEPWAVE had a daily planning meeting (DPM) 7 days a week to discuss relevant operations issues, resources and status, science objective status, current weather and outlook, and PI science mission proposals. An interesting aspect of DEEPWAVE was that all GV flights but one were conducted at night to allow the new GV optical instruments to perform optimally. The DPM was convened at 0100 UTC (1300 LT) 7 days a week to allow participation by as many groups as possible across 10 time zones. ReadyTalk web conferencing linked participants with full audio and video capabilities. The DPMs led to the definitions of the various IOPs and RF and GB measurement scheduling.
Real-time support for the project including tracking of, and interactions with, the GV utilizing the DEEPWAVE field catalog and the EOL/Research Aviation Facility (RAF) Aeros and catalog maps tools for displaying real-time aircraft position, flight-level data displays, satellite and model data overlays, dropsonde launches and plots, and lidar and AMTM data sharing.
A unique aspect of DEEPWAVE was the ability to make real-time dropsonde deployment decisions at specific points over New Zealand and widely over the Southern Ocean. These data were relayed via satellite to the ground for quality control and processing by EOL-trained student participants before forwarding to the Global Telecommunications System for assimilation into global weather center model forecasts.
DEEPWAVE PARTICIPANTS AND ROLES.
Table C1 shows the DEEPWAVE participants and their roles during the experiment.
This article is included in the Multi-Scale Dynamics of Gravity Waves (MS-GWaves) Special Collection.