Canopy EOF analysis
Edward Patton, Roger Shaw (University of California, Davis) and John Finnigan (CSIRO, Australia) are using empirical orthogonal function (EOF) analysis to extract the “characteristic eddy” from large-eddy simulations of turbulent flow above and within a plant canopy. The analysis reveals the three-dimensional structure of organized canopy motions with only limited a priori assumptions The eddy structure that emerges from the analysis appears much like the double roller vortices characteristic of plane mixing layers (see Figure 38). Future work will use the EOF analysis to develop a low-dimensional model of canopy exchange processes for use in assimilating measurements and parameterizations of land-surface exchange.
Canopy turbulence under stable stratification
Using high-resolution large-eddy simulations, E. Patton and Peter Sullivan are investigating the interplay between stable stratification and canopy turbulence. Compared to simulations with no vegetation, a stably stratified canopy reduces the height of the nocturnal jet and the pressure drag alters the momentum balance resulting in lower turbulence levels and a reduced PBL depth (see Figure 39). Future research will investigate turbulent flow over canopy covered complex terrain. This work is ultimately aimed at improved upscaling of carbon dioxide measurements and parameterizations of canopy exchange in larger-scale models.
Energy transfer under nocturnal stable conditions
Jielun Sun, in collaboration with Sean Burns at NCAR, Richard Coulter (Argonne National Lab), and Robert Banta (ETL/NOAA), continued analyzing the data collected from the Cooperative Atmosphere-Surface Exchange Study-1999 (CASES-99). She focused on the statistical characteristics of various atmospheric disturbances – density currents, solitary waves, and internal gravity waves. She found that these disturbances were more likely to occur with northerly winds of magnitude less than 4-6 m/s. A temperature drop associated with the density current is more obvious at upper levels due to the temperature increase with height at night. High frequency internal gravity waves are frequently observed at higher levels as their counterparts near the surface are easily wiped out due to strong-shear generated turbulence mixing. Under weak wind conditions, density currents are observed about 50% of the time in the early evening and are most often associated with wind perturbations from the northeast. This implies that drainage flows are a possible cause of the observed density currents. In addition, nocturnal temperature oscillations tend to be associated with pressure oscillations that have long periods. Further information can be found at http://www.mmm.ucar.edu/science/abl/cases/cases.html. Future work will compare tower and remote sensing data to identify influences of ground drainage flows and upper low-level jets on intermittent turbulence. Next, she will examine the statistical impact of intermittent turbulence on the nocturnal boundary layer structure. This work will lead to further understanding and parameterization of the nocturnal boundary layer.
CO2 transport over complex terrain
Jielun Sun, along with Russ Monson, Susan Buhr (both University of Colorado), Dave Schimel, Britton Stephens, Steve Oncley, Tony Delany (ATD), Dennis Shoji Ojima (Colorado State University), Leonel Sternberg (University of Miami), and Dean Anderson (USGS), collaborated on an NSF sponsored program to study carbon sequestration in complex terrain. She participated in the aircraft field campaign (ACME, Airborne Carbon in the Mountains Experiment) in May and July 2004 and a ground field campaign (CME-04, Carbon in the Mountains Experiment) that has been underway since May 2004. Also, she continued analyzing data collected at the Niwot-Ridge AmeriFlux site. She found that the spatial distribution of CO2 is sensitive to major steep slopes, small gullies embedded in these steep slopes, and that nocturnal drainage flows associated with topography plays an important role in nighttime CO2 transport. In the early morning, a sudden decrease of CO2 storage in the canopy layer is observed and results from upslope flows and upward turbulent transport. Preliminary analysis of the aircraft data also confirm an accumulation of CO2 at low levels even on large scales. These results provide critical information on the many on-going long term observational programs, which currently ignore horizontal transport in the global CO2 budget. Additional information is provided in http://www.mmm.ucar.edu/science/abl/cme/cme.html. In the future, she will continue to work closely with colleagues funded by the NSF project and from the NCAR Biogeoscience initiative on analyzing data collected during CME-04 and ACME. Meanwhile, she will work on a new research plan to further investigate CO2 transport over complex terrain on regional scales.
Carbon in the Mountains Experiment: Tower Work
S. Burns in collaboration with Russ Monson (University of Colorado) collected and analyzed meteorological, carbon dioxide, and flux data from the University of Colorado Ameriflux tower. Ameriflux is a network of towers located across North and South America that is attempting to better quantify carbon budgets on continental scales. The Colorado Ameriflux Tower is also part of CME (Carbon in the Mountains Experiment) that took place from May 2004 to July 2004. The tower will continue to collect data throughout FY05. The following animation (Figure 40 animation) depicts CO2 evolution at the CME site.
Click image for a larger version.
S. Burns in collaboration with Tony Delany (ATD), Mark Williams (INSTAAR), and Russ Monson (University of Colorado) examined carbon dioxide diffusion rates within the snow pack of a sub-alpine forest. Transport of CO2 through a snow pack determines the rate at which biologically-generated CO2 in the soil is released to the atmosphere and can significantly impact global CO2 budgets. The rate of CO2 transfer in snow depends on several parameters including: wind speed, snow density, and ice layers. Using "HYDRA" (a multi-inlet CO2 measuring system developed by Tony Delany at ATD) a preliminary study was carried out in 2004 with a more extensive study planned for winter 2004-2005. An example of CO2 transport is shown in Figure 41. Here CO2 is injected into a snow pack and the resulting CO2 concentration is monitored at different inlet locations within the snow pack.
Click image for a larger version.
Margaret LeMone, Diane Strassberg (MMM student assistant), Joe Alfieri (RAP student assistant), and Robert Grossman (CoRA) compared IHOP data to CASES-97 data to illustrate the important role that vegetation (in this case winter wheat and grass) plays in the distribution of sensible (SH) and latent heat (LE) near the surface. Panel a) of Figure 65 shows the distribution of winter wheat (green areas on lower right) along the flight track (dots) located southeast of Wichita, Kansas. This flight track is the same one used in CASES-97. The remaining frames show the impact of the surface cover on radiometric surface temperature and sensible heat flux. Note that the IHOP radiometric surface temperature Ts in panel b) of Figure 42 is relatively warm over the winter wheat area along the western portion of the flight tack, with a secondary maximum corresponding to a second wheat area around -98.7o longitude. For CASES-97, there are Ts minima at the same two locations. The changes correspond to the winter-wheat life cycle: the wheat is green and rapidly growing in late April to early May, matures by late May, and is harvested by the middle of July. Note that the weaker horizontal variation on 17 June is related to recent rainfall (15 June); the other IHOP days had more than five days since the last heavy rain. For CASES-97, there was rain on 7-8 May and there was no rain for over a week before 29 April. The warm surfaces correspond to higher sensible flux. For IHOP, the two warm winter wheat areas have higher sensible heat flux (Figure 59). Similarly, the cool growing winter wheat areas have correspondingly low sensible heat flux. The pattern is less strong on 10 May because of more recent rains and less contrast between the grass and winter wheat (panel d) of Figure 65. LeMone is also collaborating with Songlak Kang, Ken Davis, and Kenneth Craig of Pennsylvania State University on explaining horizontal variability in surface fluxes along the "Western Track," which runs north-south across the eastern Oklahoma Panhandle.
Click image for larger version.
Click image for larger version.
Figure 43. Fair-weather noontime data from IHOP_2002 flux stations 1-3, differences between refractivity estimated from 2-m measurements, and average refractivity estimated from standard flux-profile relationships, as a function of (top) surface latent heat flux and (bottom) relative humidity.
One of the objectives of IHOP_2002 was to test Frederic Fabry's radar-based method to obtain near-surface refractivity (and hence, information about the near-surface mixing ratio and temperature fields). Using software developed for the GLOBE program to calculate surface-layer profiles, M. LeMone showed that changes in temperature and mixing ratio with height accounted for at least part of the systematic humidity-related change in bias between refractivity determined from radar and refractivity calculated from measurements taken at 2 m found by Tammy Weckwerth and Crystalyne Pettet of ATD, Frederic Fabry and ShinJu Park of McGill, and James Wilson of RAP. M. LeMone assumed that radar estimates of refractivity, based on propagation time from the radar to ground targets, corresponded to averages in the lowest 10-20 m of the atmosphere. Flux-profile relationships predict that larger latent-heat fluxes are associated with steeper falloff of specific humidity with height. This leads to a "low" bias for the radar with high latent-heat fluxes. Because high relative humidity tends to be associated with high latent heat flux for the data used, the low bias is also seen at high relative humidities (see Figures 43a and 43b).
Teddy Holt (Naval Research Laboratory), Dev Niyogi (North Carolina State University), Fei Chen (RAP), and Kevin Manning and M. LeMone from MMM compared the sensitivity of precipitating convection modeled in the Navy COAMPSTM model to soil-moisture initiation and LSM representation of evapotranspiration. The differences in simulated fluxes, boundary layer height, and dryline behavior were significant, resulting in significant differences in initiation, horizontal distribution, and propagation of convection, even in a strongly-forced situation. Comparison of modeled and surface near-surface fluxes and means was not straightforward, since small horizontal displacement of modeled features resulted in large differences at a point.
Jielun Sun, in collaboration with Larry Mahrt and Dean Vickers (both
Oregon State University), Tom Jackson (NASA), Ian MacPherson (National
Research Council, Canada), Paul Houser (NASA Goddard Space Flight Center),
and Eleanor Burke (University of Arizona), investigated the response
of atmospheric moisture flux to temporal and spatial variations in soil
moisture and vegetation variability. She reviewed common modeling approaches
and investigated alternate formula for computing evapotranspiration taking
into account soil moisture and surface type. The aircraft data collected
during the Southern Great Plains (SGP) over various surface types are
used to examine a simple formula for evapotranspiration. For further
Click image for a larger version.
Robert Sharman (RAP) and P. Smolarkiewicz conducted numerical studies of urban boundary layer flows and contaminant dispersion in the vicinity of the Pentagon using EULAG (Figure 44). The objective is to quantify airflow characteristics under various meteorological conditions, with emphasis on predicting the concentration levels of hazardous contaminants as they disperse and collect in recesses. A series of simulations for both high and low Reynolds number flows is needed to complement and compare to field observations and wind tunnel measurements. So far, they have focused on high Reynolds number flows for various choices of physical and numerical parameters, in order to assess the optimal model configuration. Among others, the simulations using standard terrain-following coordinates were compared with an immersed-boundary approach. Contrary to experience with other simulation models, EULAG represents steep urban structures as terrain-following orography and produces results qualitatively similar to the immersed-boundary approach. Generalization of this approach should lead to better representations of internal boundaries in atmospheric and oceanic numerical simulation models.
Peter Sullivan, John Wyngaard (Pennsylvania State University), James
Edson (Woods Hole Oceanographic Institute) along with NCAR investigators
in MMM and ATD planned, implemented, and carried out the Ocean Horizontal
Array Turbulence Study (OHATS) utilizing the air-sea interaction tower
(ASIT) operated by Woods Hole Oceanographic Institute. The ASIT is located
approximately 3 kilometers off the coast of Martha's Vineyard in 15 meter
deep water with a southwesterly exposure open to the Atlantic ocean.
The experimental design and layout of the sensors in OHATS targets the
measurement of turbulence variables in the marine surface layer along
with simultaneous wave information. Turbulence (three velocity components
and potential temperature) was measured using 18 sonic anemometers and
wave information (height and propagation direction) was collected using
3 laser altimeters. This data will be used to construct subfilter scale
motions that are modeled in LES codes and will potentially establish
correlations between marine surface layer turbulence and wave states.
The duration of the field campaign was approximately two months starting
1 August 2004 and ending 1 October 2004 with continuous data collection.
The observations span a regime of light to modest strength winds, unstable
to very stable stratification, and a variety of wave states. A preliminary
sweep through the dataset has identified more than 280 hours of data
that will be interrogated in future analysis. Further information, including
a gallery of photographs and preliminary data, can be found at http://www.atd.ucar.edu/rtf/projects/OHATS04 and
Click image for a larger version.
P. Sullivan, James McWilliams (UCLA), and Ken Melville (Scripps Institute of Oceanography) are developing a turbulence resolving LES model for wind-wave equilibrium ocean mixed layers that contains surface wave effects; both wave breaking and vortex forces generated by Stokes drift are included in the LES. During the past year, we incorporated the wave breaking model described by Sullivan et al. (JFM, 2004) into a working LES code that also contains Stokes drift. With the assumption of wind-wave equilibrium, we are able to use empirically derived formulas from field observations to specify the total momentum and energy supplied by the atmosphere to the ocean, the shape of the surface wave spectrum, and the characteristics of the wave breaking field. Essentially, the primary input to the LES is then the reference wind speed at a height of 10 meters above the water. LES results illustrate close connections between surface wave conditions and vertical mixing in the OBL. We find that the mean currents, turbulence variances (and TKE), scalar and momentum fluxes, and entrainment at the thermocline all exhibit considerable sensitivity to the surface wave field. One of the most striking results is that the greatest mixing arises when wave breaking and Langmuir circulations (resulting from Stokes drift) act together. Flow visualization and animations indicate that excess horizontal momentum generated by a breaking wave at the surface can be trapped in a lateral convergence zone by Langmuir circulations and subsequently deflected downward into the mixed layer as a vigorous downwelling jet. This sequence of events is depicted in the (x-y) snapshots of Figure 45. We are able to trace the downwelling jets in panels c) and d) to surface sites where wave breaking and Langmuir circulations are both active at earlier times. These downwelling jets are not found in LES with Stokes drift or wave breaking acting alone. They found that the vertical velocity skewness is a strong indicator of the presence of these coherent structures.
P. Sullivan and Stephen Belcher, short term visitor from U. Reading (UK), began a numerical investigation to explore the formation and dynamics of critical layers induced by moving wavy surfaces. The objective is to interrogate the Miles critical layer mechanism for wave growth in the presence of a wave group. A series of canonical DNS problems were designed to focus on the vortex dynamics induced by a wave group near a boundary. They find that a flat surface with a fixed sinusoidal varying vertical velocity as a lower boundary condition can interact strongly with the mean profile and thereby induce strong vortical motion. By tracking the solution in time they are able to examine how the critical layer induces a downward flux of vertical momentum which can then lead to substantial wave growth. An animation of this flow is provided in Figure 46 animation.
Click image for a larger animation.
Jielun Sun, in collaboration with S. Burns, Douglas Vandemark (NASA
Goddard Space Flight Center), and Mark Donelan (University of Miami),
explored the possibility of retrieving two-dimensional wave spectra using
three laser altimeters on board the LongEZ aircraft. They found that
a wavelet analysis method is able to retrieve wave propagation direction
and wavenumber using the datasets collected from both the Shoaling Waves
Experiment (SHOWEX) and Coupled Boundary-Layers/Air-Sea Transfer (CBLAST-low)
experiments. She processed all the directional wave spectra from the
LongEZ aircraft collected during the CBLAST-Low pilot experiment in 2001.
She also worked with James Edson (Woods Hole Institute of Oceanography),
and S. Burns and P. Sullivan, Dean Vickers and Larry Mahrt (Oregon State
University), and Djamal Khelif (University of California, Irvine), on
analyzing the CBLAST tower data and Pelican aircraft data. The role of
surface waves on turbulent transport in the marine atmospheric boundary
layer is being examined in detail. Eventually, the improved understanding
of this complex process will lead to improved models of air-sea coupling
and weather forecasts. Web sites http://www.mmm.ucar.edu/science/abl/showex/showex.html
and http://www.mmm.ucar.edu/science/abl/cblast/ provide additional information. In FY05, she will continue to work with colleagues involved in CBLAST-Low analyzing and comparing the Pelican aircraft and tower data to better understand the characteristics of energy transport in wave influenced layers.
Coherent vortical motions just beneath the ocean surface are known to control the transport of soluble gases like CO2 across the air-water interface. However, mechanisms that determine these vortical motions are poorly understood. Possible mechanisms include surface waves, interaction between wave and shear turbulence, and pure shear turbulence. To isolate the turbulence effect from surface waves, Wu-Ting Tsai, Shi-Ming Chen (National Central University, Taiwan) and Chin-Hoh Moeng performed a direct numerical simulation of a stress-driven turbulent layer bounded by a free surface. They observed that even in the absence of waves elongated high-speed streaks and localized low-speed spots are observed on the surface Figure 47. Floating Lagrangian particles aggregate and travel along these streaks similar to those found in the laboratory and Langmuir circulations found in the open ocean. Elongated streamwise vortical motions are also observed just beneath the surface, however, these vortices are shorter in length and less correlated than those arising from Langmuir circulations.
Click image for larger version.
Large amplitude solitary waves affect underwater naval operations through their effects on buoyancy and acoustic transmission. A sudden encounter of a lower density area at a depth of high ambient density can cause an underwater vehicle to ascend suddenly to the surface. Underwater signals are perturbed by solitary waves (near-surface solitary waves have been observed to cause large anomalous acoustical signal loss of about 15 dB). As a collaborative effort, a team of researchers led by Alex Warn-Varnas (NRL) and P. Smolarkiewicz, simulated three-dimensional solitary wave generation and propagation using EULAG coupled to the barotropic NCOM tidal model. Newly developed dynamic grid-deformation technology was used. The scenario examined was the site of the ASIAX experiment, Figure 48a, where the evolution of solitary waves is particularly complex. The tidal motions in the Strait of Luzon, generate internal bores that subsequently disintegrate into solitary wave packets, which propagate into the South China Sea towards the continental shelf where large-amplitude solitary waves of depression, up to 150 m, are observed. Figure 48b, shows simulated isohalines after ~3.3 tidal cycles, using EULAG in an idealized high-resolution setup.
Click image for larger version.
Atmospheric diurnal variability in humidity and winds controls the diurnal variability of fires, including their tendency to flare up in the afternoon and subside in intensity at night. Janice Coen has extended NCAR's coupled atmosphere-wildland fire – environment model (CAWFE) to include another feedback loop between the atmosphere and fire allowing fuel moisture, a critical parameter in the fire environment, to respond to changing atmospheric conditions. She has also extended the model capabilities to represent more complex fuel complexes and fire behaviors by adding new fuel models. This new capability allows fire modeling in chaparral, a dominant vegetation type of southern California. In collaboration with Dar Roberts (Dept. of Geography, Univ. of California Santa Barbara) and Phil Dennison (Dept. of Geography, University of Utah), she simulated the Simi Fire, which ignited near Simi Valley, CA, during the firestorms of October, 2003. In this case, fire spread rapidly under severe Santa Ana winds complicated by small-scale canyon winds through extremely dry fuel and complex topography of the Santa Susana Mountains (Figure 49 animation). Although verification of the simulation continues, this case is being used to test the capabilities of the model to represent extreme rates of fire spread in complex, spatially heterogeneous fuels.
Click image for a larger animation.
With the support of an NSF Information Technology Grant, J. Coen, collaborators at the University of Colorado Denver (Jan Mandel, Leo Franca, Lynn Bennethum, and Craig Johns), the University of Kentucky (Craig Douglas), Rochester Institute of Technology (Tony Vodacek and Bob Kremens), and Texas A&M University (Wei Zhao) are developing a dynamic data-driven wildland fire model. The goal of this work is to develop methods capable of dynamically assimilating out-of-sequence field data such as airborne infrared imagery of fires and autonomous atmospheric environment detectors into running ensembles of atmosphere-fire models and return results to the field. In this first year, the team has developed a stochastic reaction-diffusion partial differential equation to replace the current fire code, and developed a framework for introducing data into ensemble members. Aside from improvements that are expected in simulating fires by guiding them with real data, the techniques for assimilating data have wide application to other problems where the probability distribution of the assimilated field is not Gaussian (for example, the temperature is either that of ignited fuel or UN-ignited, not a centered distribution).
J. Coen has tested CAWFE configured for a faster-than-real-time application to wildfires in Colorado during summer 2004. A two-times-daily 48-hr MM5 forecast for the continental U.S. –done by Jim Bresch (MMM)- was used to initialize the domain and update boundary conditions. The atmosphere-fire model used four nested domains to nest from synoptic scale to 556m in the vicinity of an ignition. Skip Edel (Colorado State Forest Service) provided the Colorado fuel database, and Christine Wiedinmyer (ACD) adapted it for a gridded model. Ignition information and local fuel moistures were estimated from the web, and CAWFE simulated the fire progression throughout that day. These results are still being examined for strengths and weaknesses, but demonstrates a first capability to routinely simulate fire behavior, smoke transport, and the fire-affected environment using modest computational capabilities.
John Daily (University of Colorado Boulder) and Shankar Mahalingam (University of California Riverside) are collaborating with J. Coen to develop techniques to parameterize fine-scale pyrolysis and combustion processes for the spatial and temporal scales of atmospheric models. Daily has developed a detail representation of the complex chemical reactions that occur when wildland fuels are heated and decompose into flammable gases that are burned in wildfires, and has been developing a reduced set of chemical equations that can be used in coupled kinetic-transport models. Mahalingam has:
E. Patton and J. Coen are actively porting the fire component of the
Clark-Hall model into the WRF framework. Implementing the fire code into
WRF will benefit the community by allowing new users to take full advantage
of the many services that come with a community model. In particular,
WRF is fully supported, it uses state of the art technology, it runs
on many computing platforms in both serial and parallel environments,
it is easy to switch and/or add physics or numerics, and comes with pre-built
analysis tools. Other reasons motivating this conversion is the ability
to immediately link with the emissions and chemistry components of WRF,
as well as the readily available real-time data initialization and assimilation.
The fire-component has been separated from the Clark-Hall model and re-written
for implementation into WRF. Implementation of the fire module into WRF
is ongoing. We anticipate a working version to be available in the first
six months of FY05. This is joint work between MMM and the Wildland Fire
Next Topic: Planetary Boundary Layers