Infrasonics

I got interested in acoustics whilst trying to find a cause of weather. I recently came across another magnificent US effort to aid mankind.

When you come across this sort of thing you wonder why they have a republican party.
And voted for a monkey.
Twice. …

Inframatics Newsletter June 2004

Storms

Storms are frequently recorded by infrasound
arrays. An example provided here occurred on
August 28, 2003. Infrasound waves due to the
storm were recorded at IS31. In Figure 19 the nine
traces corresponding to 8 micropressure variations
(BDF channels) and to the wind speed (LWS
channel), are shown.
The correlation between the wind speed and
the micropressure variations is very clear.
The coherence of the waveforms recorded
between 03:30 and 04:00UT is clearly visible
when stacking the signals (Figure 20) and their
energy content, mainly concentrated within 1 Hz,
is visible in the spectrograms of Figure 21.

Figure 19. The infrasound raw data recorded at IS31
on 2003/08/28: the correlation between micropressure
and wind data is clearly visible. Amplitudes are
expressed in Pascals (micropressure data) and in m/s
(wind speed).

Figure 20. Stacking of four of the eight waveforms
recorded at IS31: on the x-axis time is expressed in
seconds; on the y-axis amplitudes are expressed in
Pascals.

Figure 21. Four of the eight spectrograms associated
to the records at IS31 on 2003/08/28. The energy
associated to the four arrivals is concentrated below
1 Hz. On the x-axis time is expressed in seconds, on
the y-axis the frequency is expressed in Hz. The color
bars are based on the computation of the power
spectral densities (expressed in Pa2/Hz).

*******

Geomagnetic Storms
In several studies ([7], [8]) it has been shown
that the auroras borealis and australis generated
by geomagnetic storms, produce infrasound
waves. Here a quite unusual example is provided.
The infrasound signals have been recorded at
about 42.1°S at IS05, Hobart, Tasmania,
(Australia), on 21 November 2003, when a large

geomagnatic storm occurred. The records of the
7-element array, obtained by filtering data
between 0.08 and 0.1Hz are shown in Figure 22.
The signature of the infrasound waves generated
by auroras and the correlation between the
waveforms are clearly visible.
Several other examples of aurora occurring
at mid-latitudes have been observed during this
period.

Figure 22. The infrasound data recorded at IS05 on
2003/11/21: data are filtered between 0.08 and 0.1Hz.
The correlation between micropressure and wind data
is clearly visible. The amplitudes are expressed in
Pascals.

*******

Inframatics Newsletter March 2004

Atmospheric Specifications for Infrasound Calculations
Douglas P. Drob
The primary research goal of inframatics.org is to coordinate worldwide research on
understanding how to use low-frequency sound waves to characterize distant natural and
man-made atmospheric events. There are a number of different problems being investigated.
These include;
• Quantification of the impact of the assumptions and levels of approximations in
propagation codes and atmospheric specifications.
• Development of remote sensing capabilities for significant natural geophysical
phenomena of the solid earth (e.g. earthquakes, volcanoes, and land slides); ocean
processes (e.g. microbaroms, littoral surf interactions); and atmospheric processes
(e.g. stratospheric dynamics, upper atmospheric tides, and lower thermospheric
heating).
• Operational infrasound source location including; event detection and location,
interactive event screening and location, and ground-truth event location and analysis.
The solution to any one of these problems is
not mutually independent. Whatever the specifics
may be, infrasound propagation calculations
require specifications of the atmospheric state
variables – wind, temperature, density, and
pressure from the ground to 170 km. These fields
are needed on a case-by-case basis for the analysis
of historical events, in real-time for operational
processing, and over extended periods of time for
climatological investigations. Ideally these
specifications should be global in nature and have
as much spatial and temporal resolution as
possible. Furthermore, it is important that the
software tools to exploit these data for infrasound
propagation calculations should be simple,
reliable, easy to use, and independent of the type
of atmospheric specification selected. Recent and
continuing improvement in the atmospheric
science community’s ability to specify and
provide detailed atmospheric specifications is
helping to advance infrasonic research. There are
several types of specifications available; a) raw
atmospheric observations coincident with
infrasonic events of interest, b) climatologies or
empirical models, c) global and regional
operational Numerical Weather Prediction (NWP)
analyses, and d) hybrid models.
The purpose of this article is to increase
awareness in the infrasound community about the
sources, availability, utilization, and basic aspects
of the various types of specifications available.
This article will provide a perspective from the
point of view of an atmospheric scientist on the
relative strengths and weaknesses of these
different specifications for infrasound propagation
calculations. The ultimate goal is to reach a point
where improvements in atmospheric
specifications no longer improve the performance
of infrasound monitoring systems. This point has
not yet been reached. To help define this issue
more clearly the uncertainty and errors of any
inferred infrasound source characteristics are the
combination of errors and uncertainties in; 1) the
measured signals from each array element, 2)
array waveform processing and detection
algorithms, 3) the propagation models relating the
observables back to the source, and 4) the
specifications of the background atmosphere.
Items three and four are tied together by the
fundamental physics of infrasound propagation.

Direct Measurements
Direct measurements in the vicinity of the
source or receiver within a few minutes or hours
of an event provided a way to specify the
infrasound signal propagation conditions.
Typically these can be radiosonde (0 – 35 km)
and/or rocketsonde (35 – 75 km) wind and
temperature profiles. Radiosonde measurements
are being collected on a daily basis around the
world by various weather services. Other direct
measurements such as satellite temperature and
wind profiles are also available but difficult to
obtain, understand, and utilize in infrasound
propagation calculations. These observations then
need to be combined with climatologies and/or
weather analysis products. Unfortunately, direct
measurements typically include localize wave
structures that may or may not be part of the larger
synoptic scale structures (200 to 1000 km); i.e.
observable at both the source and receiver.
Furthermore, direct measurements may include
unchecked calibration biases and measurements
errors. This makes using raw radiosonde and other
single point measurements rather dubious for
some infrasound propagation calculations. In
most cases, direct measurements are only useful
for short to medium range propagation
calculations where more than one profile is
available and it is safe to assume range
independence. Direct in-situ surface
measurements, however, remain the best available
source of wind information used for noise
characterization and reduction experiments.
Climatologies
A decent climatology can capture the main
aspects of the general circulation and temperature
structure of the atmosphere determining the
behavior of infrasound propagation.
Climatologies are convenient in that they
circumvent the need to deal with direct
measurements or gigabytes of operational
numerical weather prediction data. As such, they
are also a useful tool for developing and testing
propagation codes. They also provide a convenient
resource for calculating static travel time tables.
Unfortunately the represented geophysical
variations in climatologies can often be
overshadowed by naturally occurring stochastic
variations and lead to erroneous interpretations
of infrasound observations.
The most popular climatologies in today’s
infrasound community are the Mass Spectrometer
and Incoherent Radar Model (MSIS-90,
NRLMSISE-00) [1, 2] for temperatures, densities,
and pressures, and the Horizontal Wind Model
(HWM-93) [3] (henceforth the HWM/MSIS
empirical models). These two models were
originally developed at the NASA Goddard Space
Flight center but are now the primary
responsibility of the Naval Research Laboratory,
Upper Atmospheric Modeling Section.
Constructed from a 40-year historical database of
upper atmospheric research measurements, the
HWM/MSIS models provide a good way to obtain
estimates of winds, temperatures, pressures, and
major species concentrations in the mesosphere
and lower thermosphere (55 to 150 km) for
infrasound propagation calculations. Referenced
over 750 times in the scientific literature, the
empirical models are widely used by the
atmospheric science research community.
There are several reasons why the HWM/
MSIS models are convenient for infrasound
propagation calculations. As FORTRAN
subroutines with embedded empirical coefficients,
there is no data management requirements
associated with the models. Furthermore only
minimal programming experience is required to
independently run or integrate these two models
directly into existing infrasound propagation
codes. With interpolation in space and time
intrinsically provided, the specifications are
continuous functions over the full range of
latitudes, longitudes, altitudes, seasons, local
times, solar flux, and geomagnetic conditions.
There are however, disadvantages to using the
HWM/MSIS models for infrasound propagation
calculations. Other than the most cyclical or
repeatable patterns, these models make no attempt
to resolve of the atmosphere’s random synoptic
scale weather patterns and waves. The coverage
of the available historical observations is often
sparse over sections of the models temporal and
spatial domain; therefore they only represent the
grossest salient dynamical features of the general
circulation of the lower, middle, and upper
atmosphere. Though the output precision is
infinite, the effective spatial resolution of the
modeled atmospheric variations is approximately
10° x 10° degrees horizontally and about 1-2 km
vertically. Compared to other options the
spatiotemporal resolution of these models is
limited.
To understand what this implies for infrasound
propagation calculations the statistical
performance of the HWM wind model was
evaluated [4]. Biases in the magnitude and
position of the stratospheric wind jets as compared
to 15-day zonal averages of stratospheric
Numerical Weather Prediction (NWP) analysis is
on the order of 25 m/s and can persist for several
months, particularly in the southern hemisphere.
The calculated geophysical variances within the
15-day zonal averages of the NWP data are on
the order of 20 to 30 m/s. Within a given daily
zonal average, the variances due to the synoptic
scale weather patterns are an additional 20 to 30
m/s. As a result, the HWM wind climatology can
underestimate the magnitude and direction of the
stratospheric wind jets by as much as 50 m/s over
large spatial regions for extended periods of time.
These errors are large enough to result in
erroneous predictions of stratospheric ducting and
can lead to significant errors in calculated travel
times and azimuth deviations. The situation for
the accurate representation of tropospheric ducting
phenomena is even worse [5].
Additional biases of 20 to 60 m/s and RMSE
(root mean square error) variances of 40 to 60 m/
s in the 80 to 120 km region of the HWM model
also exist. Other published scientific evaluations
of the MSIS model indicate the existence of
occasional biases and unresolved variances in the
80 to 120 km region. Fortunately, however,
atmospheric temperature profiles have
significantly less climatological variance as
compared to the wind vectors. The unrepresented
variability in both models has negative
implications for the calculation of the propagation
characteristics of thermospheric arrivals. Work
by the infrasound research community to
understand and access the impact of these
problems is proceeding on a number of fronts.
As we shall see, the observed variability and
climatological model biases of the lower
atmosphere (0-55 km) can be eliminated by
combining the HWM/MSIS models with
Numerical Weather Prediction (NWP) data. In
order to correct the current biases and improve
RMSE errors in the lower thermospheric portion
of the HWM/MSIS models the data used to
characterize the biases will need to be assimilated
into the models. Work on this will continue at
NRL over the next few years. Related research
by the atmospheric science community is also
continuing to develop and apply techniques to
routinely measure and model the 60 m/s RMS
variances, or synoptic meteorology, of the 80 to
120 km region.
In summary, climatologies are based on
historical data and by their nature are subject to
biases and systematic observational errors. They
will always depart from reality for specific events
or conditions. The extent to which this occurs is
latitudinally and seasonly dependent. Even
though the HWM/MSIS models provides a decent
frame of reference and are convenient to use, they
are less than ideal for performing accurate source
location calculations at certain locations and times
of the year.

Global Scale Numerical Weather Prediction Analyses

Lower atmospheric Numerical Weather
Prediction (NWP) is an important national and
international activity. A number of organizations
and agencies such as the World Metrological
Organization (WMO), (US) National
Oceanographic and Atmospheric Administration
(NOAA), and various Defense Departments build
and maintain networks of ground-based weather
stations and meteorology satellites. Operational
measurements are continuously assimilated into
complex numerical models using the combination
of rigorous statistics and geophysical fluid
dynamics. There are several well known
operational atmospheric modeling centers; the
European Center for Medium Range Weather
Forecasting (ECMWF) [6], NOAA National
Centers for Environmental Prediction (NCEP) [7],
and the US Navy’s Fleet Numerical Meteorology
and Oceanographic Center (FNMOC) [8]. These
centers produce both weather forecasts and
observational summaries called analyses. The
spatial resolution of these global analysis products
and forecasts is typically 1o x 1o or better (about
4500 km2 at mid-latitudes). To achieve kilometer
scale resolutions nested mesoscale models like the
Navy’s Coupled Ocean/Atmosphere Mesoscale
Prediction System (COAMPSa) are available [9].
These mesoscale systems include highly detailed
oceanic, topographical, and nonhydrostatic
effects.
Unfortunately, none of the operational centers
regularly measure, specify, and produce forecasts
much beyond the upper stratosphere (50 to 55
km), with several only providing output for
general consumption below 35 km. This is in part
due to the fact that civil and military demand for
specifications above 35 km is minimal, but also
because it is more difficult and costly to make
measurements above this altitude. Presently
monitoring of stratospheric temperatures and
inferred winds fields are obtained by only one or
two space-based operational sensors. Operational
NWP centers and climate researchers have,
however, identified the importance of extending
the upper boundary of the data assimilation and
prediction models, if only for the reason that it
has been shown to improve their forecast skill in
the lower atmosphere. As a result, the upper
boundaries of forecast and data assimilation
systems are gradually being extended into the
mesosphere and beyond. Consequently, reliable
operational observations are needed in these
regions to drive and constrain the models. Thus
in the near term, current climatologies are the
infrasound community’s only readily available
source of environmental information above 55
km.
Eventually new scientific instruments to
measure and infer the temperatures and wind
fields from the stratosphere on upwards will
become a reality. For example, a new temperature
sounder was launched last year onboard the DMSP
F-16 satellite. Once the instrument is calibrated
and validated it promises to provide improved
upper stratospheric temperatures for use by
operational numerical weather prediction centers.
The development of space-based instruments
capable of making direct wind measurements
throughout the middle and upper atmosphere are
also being researched and tested, though it may
be some time before dedicated operational sensors
are flown. Revolutionary prototypes have already
provided vast amounts of high quality, quasi-
global wind measurements that will be used to
improve the HWM wind climatology.
Despite some limitations, there are a number
of significant advantages to utilizing operational
NWP specifications in infrasound calculations.
Several NWP databases developed for climate and
atmospheric research exist and are publicly
accessible [11]. In the 0 to 35 km region they
extend back to 1960, and in the 0 to 55 km region
they extend back to the early 1990s. These
specifications are based on observations such as
NOAA/DMSP temperature sounders, GOES
radiances and cloud drift measurements, weather
radars, and world wide radiosonde observations.
Compared to the direct use of single observations
for infrasound propagation calculations there are
established infrastructures to validate,
appropriately filter, and assimilate the host of
measurements into reliable global and regional
atmospheric specifications. The infrasound
research community can accept these NWP
analyses as data, instead of as theoretical models,
as does much of the atmospheric science research
community.
As with anything, there are also challenges
associated with using the NWP data in infrasound
calculations. As already mentioned, the majority
of the operational measurements and
specifications are focused on the 0 to 35 km region
of the atmosphere. Above this altitude the
specifications are typically not included as part
of the standard NWP center operational products
and are based on a limited number of
observational sensors. Without additional
information, NWP analyses are only useful for
modeling infrasound propagation that is limited
to tropospheric and stratospheric ducting. Other
technical hurdles include the fact that these
specifications are usually provided in pressure or
sigma coordinates and may contain a large number
of extraneous fields. Additionally these analyses
are only available at discrete time intervals (e.g.
6- to 12-hours) and are specified on discrete
Cartesian grid points that are over dense and
problematic near the poles. The conversion from
pressure to altitude coordinate is also nontrivial,
especially when considering orography.
Sophisticated interpolation schemes are required
for profile extraction and range dependent
propagation modeling. At 1° x 1° resolution
typical NWP analysis file sizes can exceed 20
megabyte every six hours and can be written in
any number of quasi-standardized data formats
and naming conventions. Compared to the solid
earth and hydroacoustic models the data
management and storage are significant.

Hybrid or Composite Specifications
By intelligently combining the NWP
specifications with the HWM/MSIS empirical
models it is possible to side-step many of the
disadvantages of the different data types while
maintain some of the advantages of each. The
heart of this problem lies in merging the various
pieces of information into a single coherent
specification which can be readily applied to an
arbitrary time, location, and altitude. The first
approaches to constructing composite models for
infrasound calculations was to combine the data
and climatologies in physical space with some sort
of single or multi-dimensional cubic splinning.
For example, a number of researchers have
appended radiosonde and NWP specifications
with HWM/MSIS to investigate aspects of
infrasound propagation [11, 12]. Whenever the
NWP specifications depart dramatically from
climatological specifications at the information
interface, however, it is easy to introduce spurious
artifacts when combining the two data sets.
To avoid this problem, and solve a few others,
it is possible to fuse the NWP analysis,
climatologies, and any available direct
observations together in a vertical pressure
coordinate system in the spectral domain via
vector spherical harmonic transforms. A unique
atmospheric specification system was developed
at the Naval Research Laboratory (NRL) to do
just this. The NRL-Ground to Space semi-
empirical spectral model (NRL-G2S) combines
all available atmospheric data into a highly
resolved, self-consistent, global and/or regional
specification, ranging from 0 to 170 km. This
system provides a very efficient way to synthesize,
store, transmit, and reconstruct large global
volumes of environmental information for
infrasound event analysis. It eliminates the
difficulty or need to work with NWP data in its
various esoteric formats and resolutions. It also
solves the pole problem and significantly
compresses the data. All the information needed
to perform range dependent infrasound
propagation calculations anywhere in the world
can be provided in a single G2S coefficient file
corresponding to a specific time interval. It is
further possible to account for time dependence
by the interpolation of several temporally adjacent
spectral coefficients sets. A mathematical
discussion outlining the NRL-G2S system can be
found in [5].
The operational prototype of the G2S system
running at NRL system obtains and archives the
operational numerical weather analyses from the
NOAA, FNMOC, and other sources and then
fuses them together with the HWM/MSIS
empirical models to produce coefficient sets at 6-
hour intervals with a time delay of a few hours.
Using several historical NWP databases, G2S
specifications are also produced on a case-by-case
basis for infrasound ground-truth event analysis.
The G2S specification are typically provided to
clients in two triangularly truncated spectral

resolutions, T-72 (~2.5°) for near-real-time and
long-term database distribution, and T-121
(~1.25°) for event driven investigations. A suite
of platform independent client software tools for
writing MATLAB and FORTRAN software
applications to construct environmental profiles
anywhere on the globe from the G2S coefficient
sets are available. The MATLAB client software
is simple enough to use at the command prompt,
yet powerful enough to integrate directly into
existing applications. Through the Tau-9 [13] and
InfraMap Toolkit [14] the NRL-G2S atmospheric
specifications are slowly becoming available to
the infrasound community. A climatological
database of G2S coefficients for 2003 to current
now exists and will continue to expand. This
database will soon become available to the
infrasound research community.
For these investigations, it is important to
remember that during certain seasons and
locations the instantaneous atmospheric
conditions in the lower atmosphere may not
deviate much from the climatological average.
Because the G2S model and the HWM/MSIS
models are essentially the same above 55 km there
will effectively be no difference between the
models and any resulting calculations. On the
other hand, at other times and locations conditions
can be quite different from the monthly average
climatology (e.g. winter mid- and high-latitudes).
Furthermore, when investigating this issue it is
very important to consider the extent to which
information is lost by any assumptions and
approximations in the propagation models.
Currently there is hesitation to abandon and/
or adapt existing seismic source location
infrastructures to address inherent atmospheric
time dependence. This is because of; 1) the
moderate increase in computational and data
storage capacity required, 2) the lack of easily
accessible codes to efficiently for incorporate this
information, 3) limited resources and manpower
to modify existing codes to exploit this
information, 4) skepticism that this information
will improve the existing monitoring systems.
Unless the number of available IMS and
research arrays is dramatically increased, it is
unlikely that it is possible to engineer around the
problem of having to know the details of the
atmosphere. A continuing challenge to the
infrasound community is therefore to improve our
ability to perform propagation and source location
calculations utilizing the wealth of atmospheric
information that is available.
Given the advanced state of today atmospheric
specifications and available computational
resources, there is no reason why source location
travel time tables and other information needed
for operational event detection and location can
not be calculated in near-real-time. A number of
possibilities from hourly, daily, to weekly
environmental and travel time table updates are
available. If implemented correctly this would
almost certainly reduce false associations caused
by the presence of the highly variable and sporadic
Conclusions/Recommendations
Insight into the relationship between
infrasonic observables, propagation models,
atmospheric specifications, and source
characteristics has been gained by studying
ground-truth infrasound events in combination
with different atmospheric specifications,
propagation models, and assumptions. Compared
to the travel time errors achievable by the seismic
monitoring community and the number of ground
truth events and detectors at their disposal, the
infrasound community must use every resource
available. Ground-truth studies and model
validation efforts will continue to be helped along
by the efficient integration of near-real-time
environmental updates, propagation model
improvements, and the availability of large
environmental databases. The issue of whether
detailed atmospheric specifications can improve
infrasound source location calculations is still an
open question, as a number of independent
theoretical and observational investigations into
this matter have produced conflicting result,
though most favor the need for better atmospheric
specifications [5, 15, 16, 17].

Conclusions/Recommendations
Insight into the relationship between
infrasonic observables, propagation models,
atmospheric specifications, and source
characteristics has been gained by studying
ground-truth infrasound events in combination
with different atmospheric specifications,
propagation models, and assumptions. Compared
to the travel time errors achievable by the seismic
monitoring community and the number of ground
truth events and detectors at their disposal, the
infrasound community must use every resource
available. Ground-truth studies and model
validation efforts will continue to be helped along
by the efficient integration of near-real-time
environmental updates, propagation model
improvements, and the availability of large
environmental databases. The issue of whether
detailed atmospheric specifications can improve
infrasound source location calculations is still an
open question, as a number of independent
theoretical and observational investigations into
this matter have produced conflicting result,
though most favor the need for better atmospheric
specifications [5, 15, 16, 17].

For these investigations, it is important to
remember that during certain seasons and
locations the instantaneous atmospheric
conditions in the lower atmosphere may not
deviate much from the climatological average.
Because the G2S model and the HWM/MSIS
models are essentially the same above 55 km there
will effectively be no difference between the
models and any resulting calculations. On the
other hand, at other times and locations conditions
can be quite different from the monthly average
climatology (e.g. winter mid- and high-latitudes).
Furthermore, when investigating this issue it is
very important to consider the extent to which
information is lost by any assumptions and
approximations in the propagation models.
Currently there is hesitation to abandon and/
or adapt existing seismic source location
infrastructures to address inherent atmospheric
time dependence. This is because of; 1) the
moderate increase in computational and data
storage capacity required, 2) the lack of easily
accessible codes to efficiently for incorporate this
information, 3) limited resources and manpower
to modify existing codes to exploit this
information, 4) skepticism that this information
will improve the existing monitoring systems.
Unless the number of available IMS and
research arrays is dramatically increased, it is
unlikely that it is possible to engineer around the
problem of having to know the details of the
atmosphere. A continuing challenge to the
infrasound community is therefore to improve our
ability to perform propagation and source location
calculations utilizing the wealth of atmospheric
information that is available.
Given the advanced state of today atmospheric
specifications and available computational
resources, there is no reason why source location
travel time tables and other information needed
for operational event detection and location can
not be calculated in near-real-time. A number of
possibilities from hourly, daily, to weekly
environmental and travel time table updates are
available. If implemented correctly this would
almost certainly reduce false associations caused
by the presence of the highly variable and sporadic

tropospheric and stratospheric ducts which have
propagation characteristics very different from the
climatogically predicated conditions. Beyond
improvements to automated infrasound detection
and source location, human interactive event
analysis and screening should be performed with
high resolution atmospheric specifications, and
eventually exploit resources like the COAMPS
mesoscale analysis system in conjunction with
hybrid models. The quality of the meteorological
information from NWP centers, and available
thought the near-real-time G2S specification and
other resources, are more than adequate for both
these purposes.
Independent of what a particular infrasound
application calls for, future updates of the HWM/
MSIS empirical models in the mesosphere and
lower thermosphere will also be very beneficial
to the infrasound community. Finally, though not
discussed in this article, continuing investigations
into the mathematics of geophysical inverse
problems with infrasound, both for atmospheric
remote sensing and source location, provides a
very powerful way to understand and quantify the
relationship between infrasonic observables and
the atmospheric specification problem. With all
of these advances, unknowns, and new techniques
it is truly an exciting time to be involved in
infrasonic research.

Advertisements

3 thoughts on “Infrasonics

  1. So..The site allows massive blocks of text so long as it is not copy edited?Interesting.The pdf readers that come with Linux distro will readily copy to text and to graphics -unless copy-protected.Most online journals disseminating new scientific thought copy protect. It's an hang-over from the dark ages when only those who could both afford an education and had an ability with Latin, could get one.I like to put stuff from pdfs onto a text editor or word processor as I wish to add commentary or copy salient extracts if I think they will go somewhere.I haven't had time to do much with the above just yet as there is just too much for me to read through.I've got to go through all their letters, pick out the plums, plumb the thinking, look for pit-falls and pratt-falls, edit what notes I took and analyse what I presume to be an aspect that underlines my thinking/ideas.Then I have to sleep on it. And await inspiration.Then re-do it all or throw it away.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s