Viability of Weather Dissemination via Social Networking Technologies
From Somer Erickson (OU SOM) and Kelvin Droegemeier (OU VPR
Office):
With the increase and expansion of technology those who are responsible
for communicating weather information are facing new challenges.
One aspect of this increase in technology is the rise of online social
networking mechanisms, i.e. Facebook, My Space, etc. Many people
of all ages and across cultural, economic and gender identities are
using these mechanisms. The idea of this project is to conduct a
survey of the public in order to understand the usage of these systems
in terms of who is using them and how in order to study the viability
of using them as a means to communicate weather information. The
student, with guidance from mentors, will be helping to administer
a survey, collect data and analyze the results.
Identifying differences between blizzard producing and non-blizzard
producing storms in the Great Plains
From Dr. Heather Reeves and Dr. Michael Congilio (OU CIMMS):
Winter storms pose a significant risk to commerce and human safety
leading to nearly $42 billion in economic losses and over 7000
deaths per year. The most severe form of winter storm is the blizzard,
which is defined by the NWS as having sustained winds of 35 mph
or more and visibility of 0.25 miles or less. Blizzards in the
Great Plains are caused by deep low- pressure systems such as Alberta
Clippers and Colorado Lows. These systems are large and the threat
area for blizzard conditions may cover multiple states, yet blizzard
conditions are usually only reported in a handful of counties for
a given system. Additionally, of the 40+ low-pressure systems that
move over the Great Plains in a season, only 2-3 may produce blizzard
conditions. Those factors that distinguish blizzard-producing storms
from non- producers are not known. In this study we will consider
select case studies of low-pressure systems that move over the
Great Plains (both blizzard producing and not) to see if any criteria
exist that can be used to determine, in advance, what systems have
the greatest potential for producing blizzards.
This project will involve looking at radar data and using some
plotting programs such as GEMPAK or Grads to look at model reanalysis
data. The student working on this project will need to be comfortable
working with unix.
Date-wise, this project will work well for someone coming
late / staying late.
Using a New Technique to Train and Test a Rotation-Detection
Algorithm
From Kiel Ortega, V. "Lak" Lakshmanan, and Travis Smith (OU
CIMMS):
Current rotation detection algorithms are prone to poor performance
due to noise in a Doppler radial velocity field. A new linear least
squares derivatives (LLSD) technique has been developed which is
less prone to problems from noise and outputs a number of fields
diagnosing the radial wind field, including azimuthal shear. Using
a clustering technique on the azimuthal shear field, it becomes
possible to detect and track areas of rotation. The student will
look at a number of cases, subjectively diagnose rotation within
thunderstorms and then compare their diagnoses to the output from
the clustering and tracking. These analyzes and comparisons will
be used to train a new rotation-detection algorithm which uses
LLSD fields.
The student for this project will learn about severe thunderstorm
analysis, the LLSD techinque and algorithm development and training.
The student should be prepared for some light scripting (either
perl, python, etc.) to parse text files for their analysis.
Evaluating Wind Power Forecasts
From Dr. Richard Carpenter and Mr. Brent Shaw (Weather Decision
Technologies; WDT):
Wind power is a rapidly growing industry. Because of the highly
variable nature of this natural resource, accurate forecasts are
critical. WDT was recently awarded a major grant from the State
of Oklahoma to develop a Wind Power Assessment and Forecast System
(WPAFS). This system will couple a mesoscale numerical weather
prediction (NWP) model, the Weather Research and Forecast (WRF)
model, with a microscale model, the Uncoupled Surface Layer (USL)
model, developed by NanoWeather, Inc.
Required skills: Knowledge
of applied synoptic-scale meteorology and weather forecasting.
Desired skills: Knowledge of mesoscale meteorology and experience
analyzing NWP model output will be helpful. Familiarity with any
of the following computing tools and environments will also be
helpful: Linux/UNIX; programming languages (Fortran, C++); scripting
languages (Perl, Python); data analysis and visualization tools
(NCL, GEMPAK); web page development (HTML, CGI); and databases
(MySQL).
Project timeline:
1: Identify cases of interest. The models
will be run over a variety of regions and synoptic regimes. The
first task is to identify cases in which the winds are driven primarily
by the synoptic vs. mesoscale/convective environments.
2: Analyze forecast results. Forecasts will be analyzed by comparing
them with nearby observations (surface stations and towers), as
well as with gridded analyses. Simple statistical measures will
be applied.
3: Compare forecasting techniques. Compare WRF
model runs at varying resolutions with the WRF-USL coupled solution.
What value does the USL model add? What value does increasing the
WRF resolution add?
Reference: “WDT
Awarded Major Grant for High Performance Wind Forecast System,” Press
Release, WDT, 15 Dec 2008.
Verification of European Storm Forecast Experiment (ESTOFEX) Forecasts
From Dr. Harold Brooks (NOAA NSSL) and Dr. Chuck Doswell (OU
CIMMS):
The ESTOFEX group has produced SPC-like convective outlook forecasts for
Europe since 2006. The project will entail evaluating 2 1/2 years of
severe thunderstorm and lightning forecasts from ESTOFEX using state of
the art forecast verification practices. Although computer programming
skills (e.g., FORTRAN, C, etc.) would be useful, they are not
essential. Familiarity with basic statistics would be helpful.
Impacts of SuperResolution Products in Warning Operations
From Dr. Les Lemon, Cynthia Van Den Broeke, and Clark Payne
(OU CIMMS):
The National Weather Service (NWS) began implementing super-resolution
radar data into warning operations in the Spring/Summer 2008. The
goal of this project is to interview forecasters in the NWS to
determine if super-resolution radar data have impacted warning
operations. In addition to the interviews, the student will also
look at cases with super-resolution data to provide first hand
experience in interpreting the data.
- Survey Details: The student will help
in developing a short survey asking forecasters if super-resolution
data have impacted their warning operations. The student will
also administer the survey and analyze the results. The survey
will be administered in person, where possible, or through conference
calls, email, or possibly Survey Monkey. The in-person surveys
will take place at the WFO in Norman, OK and possibly Tulsa,
OK, Wichita, KS, or Dallas, TX. Optimally, the visits will coincide
with an operational day, or near to one. The rest of the survey
responses will be obtained as forecasters complete the survey
that will be sent, or administered, to all of the WFOs. Additionally,
if possible, examples illustrating the forecasters’ experiences
with super resolution data will be obtained.
- Super-Resolution Data Details: This
portion of the project will involve the student looking at super-resolution
case(s) to gain hands-on knowledge of super-resolution data.
Experience interpreting super resolution data will also give
the student the ability to relate to the forecasters’ experiences
with the data.
Project Timeline:
In the beginning, the student will be heavily reliant
on his or her mentors to develop a solid foundation in super-resolution
radar theory and interpretation, and to develop the series of questions
for the survey/interview. As the project progresses, though, the
student should start developing his or her own questions about super-resolution
data and how it might affect warning operations. We hope this is
accomplished through examining data themselves and finding examples
forecasters cite in the interviews, possibly even locating original
examples. In the end, the student will compile the results of the
interview into a report and provide examples, where possible. We
also hope the student will develop some follow-up questions for further
research on super-resolution data in warning operations.
Background Coming In:
Prefer student have strong interest in radar
or have taken radar classes
Background That Will Be Gained:
The student will
use, through the course of the project, Warning Decision Training
Branch (WDTB) lessons, Journal/Conference papers, and interviews.
The student will thereby:
- Develop an understanding of
general radar theory and limitations
- Develop an understanding
of super-resolution radar processing and products
- Develop an
understanding of the use of radar data, and more specifically,
super-resolution data, in warning operations.
Diagnosing Tornado False Alarms
From Dr. Jerry Brotzge (OU CAPS) and Somer Erickson (OU SOM):
Each year across the country over 1,400 tornadoes are confirmed
with about 75% of them warned. However, over 2,700 false alarms
are issued each year, nearly two tornado false alarms for every
confirmed tornado warning. The danger of too many false alarms
is the loss of public credibility – the “crying wolf” syndrome.
So, why so many false alarms, and what can be done to lower this
number?
A data set of over 13,000 tornado false alarms will be
evaluated to determine the cause(s) behind these “failures”.
A subset of 300 events has been identified when no tornadoes touched
down within the Weather Forecast Office County Warning Area, and
these will be compared against a similar number of events with
confirmed touchdowns. This research will involve exploring large
data sets, synoptic analysis, and weather radar climatology. This
work will include some programming in FORTRAN (or C), and some
prior programming experience may be helpful.
This project is informative to CASA,
among others.
Role of Prior Experience on Practices and Policies
From Leads: Gina Eosco (OU SOM+Comm) and Somer Erickson (OU
SOM+Comm). Team members: Kim Klockow (OU SOM), Rachel Butterworth
(OU SOM+Comm), and Harold Brooks (NOAA NSSL):
This project will include two case studies focusing on the role
of prior experience with tornadoes. We will look at what role prior
experience plays in influencing current day tornado practices and
policies for public safety officials including emergency managers,
police chiefs, fire chiefs, etc. The goal is to conduct indepth
interviews with these public safety officials in two locations,
Woodward and Moore, Oklahoma. These locations have been chosen
because of their tornado experiences, but also because of their
strong memory of their respective events. The study will focus
on topics such as siren policies, the utility of weather information,
and the potential use of probabilistic tornado information. With
the help of mentors, the student will conduct, transcribe and code
the interviews. The student will gain practical experience with
social science methods, as well as learn how weather impacts society.
This project is informative to CASA,
among others.
Date-wise, this project will work well for someone coming
late / staying late.
Mitigation of Property Damage from Hail
From Dr. Jerry Brotzge (CAPS), Dr. Dan Sutter (UT Pan American),
and Somer Erickson (OU SOM):
Each year, hail is responsible for approximately $1 billion in
damage to property nationwide. However, consumers spend thousands
of dollars extra each year purchasing hail resistant shingles,
farmers spend hundreds of millions for hail insurance to protect
their crops, and several states routinely pay for cloud seeding
to minimize damage from hail. The real cost savings from these
measures is largely unknown.
This study will analyze the costs
and benefits of hail damage prevention measures by addressing the
question: Where and when is spending money on hail mitigation/insurance
worth the money?
This study will involve several key phases:
- Analyze hail threat across the United States (frequency, size).
- Determine mitigation costs to consumers - damage to roofs,
cars - and farmers – insurance, cloud seeding
- Quantify
those areas/times when mitigation benefits exceed hail damage
costs
This project will involve collection of data from various
state and federal agencies, manipulation of large data sets,
and some statistical analysis. Results from this project will
aid in determining future sensor deployment and mitigation efforts.
This project is informative to CASA,
among others.
Two projects where a particular student has been identified:
Related to Parallel
Algorithms
From Dr. David Stensrud (NOAA NSSL), Dr. Anil Pereira (SWOSU),
and Dr. Warren Moseley (SWOSU):
From the director: This student has been identified (Goree),
as the skills needed are particular to a computer science student
and the project arose from a partnership with SWOSU.
The project will be about parallel algorithms to do mesoscale
modeling. In particular the parallelization of an algorithm to
convert satellite-derived normalized difference vegetation index
(NDVI) from 1 km resolution to vegetation fraction and averaged
to a given model resolution (typically from 2-20 km). Dr. Pereira
specialty is High Performance Computing, Grid Computing and Parallel
Algorithms. Henry Neeman has agreed to help with the Supercomputing
interface if we need it. Our initial approach will be to take the
current algorithm and its associated documentation and study it.
After a brief study we will create a timed related dependency graph
to determine if the current design can be used without changes
to the algorithm. My initial feeling is that this will not create
two many dependencies. The less dependencies the better. After
we create the dependency graph we will work to restructure the
algorithm to eliminate program dependencies. Once we have done
that then will commence coding and testing for the rest of the
system. This type of study is not new but rarely done and the conversion
to dependency graph is and unusual approach to parallelization
but from discussions from people at Nasa Ames, Ut Supercomputing
Center and Henry it seems to have a lot of interest.
Topic TBD
From Tom Connor, Ned Snell, and Mark Leidner (all of AER, Inc.):
From the director: We have one participant who really
wants this project (Viel). Anyone objecting should voice their
concerns right away! Otherwise, we'll consider this project as
being determined.
We want to do a thorough statistical analysis of the daily mean
surface
temperature for various locations in the United States. The analysis
will include:
* Uncovering the trend at a given location
* Employing Fourier Analysis to uncover the seasonally dependent
mean and the seasonally dependent variance
* Remove mean and variance from the time series in order to analyze
the residuals
We will want to find the auto-regressive lag in the residuals
and also
look for the possibility of seasonality in the lag coefficients
as well.
Armed with these statistics, we want to develop a model that will
simulate a possible temperature time series. We would require that
the
model incorporate the seasonality in the variance and capture the
auto-regressive nature of the time series for a particular location.
Once an adequate model is discovered (could be Auto Regressive
Conditional Heteroskedasticity (ARCH) model, for example) simulate
the
coming year's temperature time series many times in order to build
up a
distribution of possible outcomes.
Last Updated: April 24, 2009 |
