Calendar of Events
Facilities
Projects
People
For the Lab Dock Sampling Instruments Links
Spatio-Temporal
Statistical Analysis of
Multi-platform
Optical Ocean Observations
A new Summer Course in
coastal Maine
23 June – 25 July 2003
Brief Course
Description:
A new five-week, intensive, cross-disciplinary, graduate-level
course combining the fields of Optical Oceanography and Spatio–Temporal
Statistics will be taught at the University of Maine’s Darling Marine
Center this summer. This course is
offered under the sponsorship of NSF’s new initiative, Collaborations in Mathematical Geosciences. Click here
for the full class description.
The major theme of the course is the spatio-temporal analysis and
interpretation of optical ocean data.
These data are collected on a variety of spatial and temporal scales by
a diverse array of sensors on a number of different platforms. The course, funded by the National Science
Foundation, represents a new endeavor to enhance mutual understanding of the
scientific issues, problems, and opportunities in geospatio-temporal
analysis. Students
with quantitative backgrounds and interests in oceanography, ocean optics,
geospatial and temporal statistics, and/or visualization techniques are encouraged
to apply. University of Maine graduate
credits (3 credits) will be given for SMS 598.
Optical
measurements serve as proxies for important biogeochemical variables in the
ocean including marine phytoplankton, dissolved organic materials, and organic
and suspended sediment particles.
Optical sensors include passive radiometers as well as active systems
with internal light sources. The platforms include satellites, aircraft, ships,
stationary moorings, Lagrangian drifters, underwater gliders, and powered
autonomous vehicles. The data collected
are all gappy with respect to space and/or time, and each combination of sensor
and platform covers a very different spatial and temporal regime. While data
analysis and interpretation arising from any single configuration is demanding,
the integration and interpretation of data sets arising from multiple
configurations present major challenges to current analytic methodologies.
The
course provides an opportunity for students to explore new research directions
in spatio-temporal statistical modeling, graphical exploratory data analysis,
space time point and continuous processes, Baysian analysis of spatio–temporal
data, and experimental sampling design for ocean data.
The
course goal is to foster cross-disciplinary learning by graduate students and
faculty in oceanography and geostatistics, and thereby accelerate collaboration
and advances in both disciplines.
Oceanography students who participate in this course will be better
prepared to use powerful statistical tools for extracting maximal information
from integrated ocean observing systems.
Mathematics and statistics students who participate will better
understand some of the fundamental challenges of analyzing geospatially and
temporally distributed data.
During
the five-week course students will be exposed to all steps necessary for
comprehensive understanding of spatio-temporal statistical analysis of
multi-platform ocean optical observations
-- including theory lectures in the class room, data gathering (e.g.,
boat cruises in coastal Maine waters or downloading satellite imagery), and
hands-on data analysis in the computer laboratory. Students will have the opportunity to work on
projects as individuals or in teams.
Students are encouraged to bring their own data sets or other data sets
of interest to them.
Instructors:
Dr. Mary Jane Perry Dr.
Mary-Kate Beard Tisdale
Phytoplankton physiology & gliders Spatial Analysis
Darling Marine
Center & Department
of Spatial Information
School of Marine
Sciences Science and Engineering
University of Maine
perrymj@maine.edu beard@spatial.maine.edu
Dr. Andrew Thomas Dr.
Emmanuel Boss
Ocean remote
sensing Ocean
optics & drifters
School of
Marine Sciences School
of Marine Sciences
University of Maine
thomas@maine.edu emmanuel.boss@maine.edu
Dr. Collin Roesler Dr.
Thomas Windholz
Phytoplankton optics & moorings Spatial analysis
Bigelow Laboratory for Ocean Sciences GIS Training and
Boothbay ME
croesler@bigelow.org windthom@isu.edu
Guest lecturers include:
Professor Gerard Heuvelink, Wageningen University;
Dr. John Welhan, Idaho State University;
Dr. Phaedon Kyriakidis, University of California,
Tentative Course Curriculum:
WEEK I (revised
22 June 2003)
Arrive Sunday,
June 22
Day 1 Monday,
June 23
Kresge Classroom
0830 – Welcome, General Introductions, and Course Goals
0855 – Introduce Darling Marine Center staff
0900 – Welcome by Director, Professor Kevin Eckelbarger
0930 - Life at the Darling Center, Ms. Linda Healy, Events Coordinator
1000 – Break: coffee, tea, and muffins
1030 – Darling Marine Center safety, Mr. Tim Miller, Lab Administrator
1130 – Course Mechanics, Expectations, Grading, and
Team-Building assignment for afternoon
Conference Center
1200 - lunch
Classroom in the
1300 - Team-Building exercise, lead by Dr. Mary Jane Perry
1430 - Introduction to DMC laptops and file structure, Mr. Brandon Sackmann
1500 – University of Maine library resources, Mrs. Katherine Sackmann
1530 - Break
1600 - Continue exploration of file structure, etc. on the laptops (Faculty)
Conference Center
1800 – dinner
Day 2 Tuesday, June 24
Kresge Classroom
0830 Overview of Oceanography - Andy Thomas -
How
does the ocean work? currents; vertical structure; time scales of change: seconds, daily cycles, tides, seasonal
cycles, inter-annual variability, climate change.
1000 Break
1030 Overview of Statistics – Thomas Windholz –
Introduction
to statistics – Measurement theory, data types, random variables,
distributions, densities, parametric and non-parametric statistics.
Conference Center
1200 - lunch
Classroom in the
1330 Statistics Laboratory:
One-dimensional Time Series –Thomas Windholz & Emmanuel Boss
Exploration of time series data set from
BATS - monthly profiles of temperature, chlorophyll concentration, and one
nutrient.
Exercise: explore vertical and seasonal structure of the water column.
Apply statistics concept from the morning – means, medians, and variance.
Use different software for different variable – exposure to software. (Goal – to show students that some software works better for specific analyses or plots.)
Plot
data
Student
summary: summary and discussion, lead by
students
Conference Center
1800 - dinner
Classroom in the
1900: Playing with Light –
Emmanuel Boss
Day 3 Wednesday,
June 25
Kresge Classroom
0830 Introduction to Phytoplankton – Mary Jane Perry -
Phytoplankton as biological and optical
particles; time scales and distributions.
1000 Break
1030 Data Analysis and Assumptions in
Statistics - Thomas Windholz
Assumptions,
stationarity and deviations from stationarity, isotrophy, measures of
dependence, covariance functions, variograms for space, trends in space and
time
Conference Center
1200 - lunch
Classroom in the
1330 Statistics Laboratory:
Variograms,
Covariograms, Trend removal - Thomas Windholz & Andy Thomas
Day 4
Thursday, June 26
Kresge Classroom
0830 Inherent
Optical Properties (IOPs) and their measurement – Emmanuel Boss
1000 Break
1030 Data Analysis - Thomas Winholz -
Regression,
ordinary least squares, generalized least squares, goodness of fit, robustness
Conference Center
1200 – lunch
Classroom in the
Marine Culture Building and short excursion on the Ira C.
1330 Optical Measurements:
How
marine particles and dissolved organics interact with light -
Mary Jane Perry, Collin Roesler, and
Emmanuel Boss
Introductory lecture to ac9 - Collin Roesler
Laboratory: ac9, backscattering and chlorophyll measurements by 4 teams.
Goal: what are the uncertainties at each step?
End up with errors for calculating errors in scattering (b = c - a) and RS reflectance. Use data in Friday’s lab.
Day 5 Friday, June 27
Kresge Classroom
0830 Other Absorbers and Scatters in Marine
Waters – Collin Roesler
Other
particles & dissolved organics; processes regulating their time scales and
distributions
1000 Break
1030 Data Analysis: - Thomas Windholz -
errors,
degrees of freedom, confidence tests, significance, residual analysis,
bootstrap.
Conference Center
1200 – lunch
Classroom in the
1330 Statistical error analysis using data from
optics lab, Day 4
Calculate:
scattering correction for ac9; absorption correct for VSF
calculate
uncertainty and propagation of error.
(different
operators, machines; etc.), including propagation of error –
Calculate
uncertainty and propagation of error.
(different
operators, machines; etc.), including propagation of error I
b
= c- a (do
first, easier)
R
= bb /
(a+bb)
Saturday dinner: Lobster Bake at the Darling Center, 1730
WEEK
II
Day 6 Monday,
June 30
Kresge Classroom
0800 Synthesis and lectures
Student-lead
discussion (team reports) of labs from Thursday/Friday
0900 Break
0915 Radiative Transfer Theory and basis of ocean
color remote sensing
–
Collin Roesler
1030 Break
1045 Data Analysis - Thomas Windholz –
Scale,
resolution, data quality, support, and aggregation.
Conference Center
1200 – lunch
Classroom in the
1330 Satellite Data Laboratory
Satellite
data analysis (technical lecture and
lab) – Andrew Thomas
Satellite
data, pixel issues, and scales.
Raw
data, digital image analysis, visualization.
Day 7 Tuesday,
July 1
Kresge Classroom
0830 Spatial and temporal scales of variability
in the ocean – Andrew Thomas
1000 Break
1030 Data Analysis - Thomas Windholz -
Spatial
interpolation and prediction; kriging; likelihood
Conference Center
1200 – lunch
Classroom in the
1330 Statistics Laboratory: Thomas Windholz –-
Interpolation, regression, cross validation, and 2-D kriging
Subsample
at different scales; detrend images;
how to deal with spatially gappy
data
within one image.
Day 8 Wednesday,
July 2
Kresge Classroom
0830 Optical Measurements - Mary
Jane Perry
sensors,
QC, and proxies (including satellites)
1000 Break
1030 Data Analysis: spatial sampling schemes – Thomas Windholz
sampling
schemes (random, stratified, systematic) and asset allocation
Split class into two groups for July 2
Group A:
1200 Lunch in
1330 Satellite Data Analysis - Andy Thomas and Thomas Windholz
Continue
with satellite data analysis from Tuesday, July 1.
Group B:
1200 Sandwiches on the Ira C.
Field sampling on the Ira
C. for remote sensing and in-water optical measurements – Collin
Roesler, Emmanuel Boss, and Mary Jane Perry
Day 9 – Thursday, July 3
Kresge Classroom
Moorings -
Collin Roseler
0830 Optical measurements from moorings
(Eularian)
1000 Break
1030 Data Analysis - Thomas Windholz
Measurement uncertainties, methods for
handling incomplete data sets
Split class into two groups for July 3
(same groups as July 2)
Group B:
1200 Lunch in
1330 Satellite Data Analysis - Andy Thomas and Thomas Windholz
Continue
with satellite data analysis from Tuesday, July 1.
Group A:
1200 Sandwiches on the Ira C.
Field sampling on the Ira C. for remote sensing and
in-water optical
measurements – Collin Roesler, Emmanuel Boss,
and Mary Jane Perry
Day 10 – Friday, July 4 NO CLASS (3 day-weekend)
WEEK III
Day 11 Monday,
July 7
Kresge Classroom
0830 Guest
Lecture: Janet Campbell, UNH: binning; aggregated data in non-linear models
1000 Break
1030 Data Analysis: Introduction to Multivariate
Analysis – Thomas Windholz
Conference Center
1200 – lunch
Classroom in the
1330 Optical Data Laboratory
Finish
analysis of field data from previous week, in preparation for Tuesday’s lab
(put
data into specified table form, etc.).
Day 12 – Tuesday, July 8
Kresge Classroom
0830 Optical Sampling from drifters (true
Lagrangian) – Emmanuel Boss
1000 Break
1030 Data Analysis: Principal Components Analysis -
Thomas Windholz –
Conference Center
1200 – lunch
Classroom in the
1330 Statistics Laboratory
Multivariate
exploratory methods (1st of 2 sessions) - Thomas Windholz
Day 13 – Wednesday, July 9
Kresge Classroom
0830 Hyperspectral
Aircraft data - Marcos
Montes, Naval Research Laboratory
1000 Break
1030 Data Analysis: discussion of kreiging
for the temporal domain
- Thomas Windholz
Conference Center
1200 – lunch
Classroom in the
1330 Multivariate exploratory methods (2nd of 2
sessions)
satellite
data and high spectral resolution aircraft
Satellite
data from Andy; Emmanuel procuring Hycode aircraft swath
Day 14 – Thursday, July 10
Kresge Classroom
0830 AOP
Models - Collin Roesler
Apparent
Optical Properties (AOP) inversion models (analytic models)
1000 Break
1030 Geo-statistical modeling of spatial and
spatial-temporal data – guest lecturer
John Welhan,
1.
Random
function model
2.
Auto
correlation analysis and modeling
Conference Center
1200 – lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory
Auto
correlation analysis and modeling - using WinGLSIB - John Welhan
Conference Center
1600 Darling Marine Center Lecture Series
Directed
sampling ships, AUVS, and gliders – Mary
Jane Perry
Day 15 Friday, July 11
Kresge Classroom
0830 Geo-statistical modeling of spatial and spatial-temporal
data - John Welhan
Part
3: modeling of variability and the
analysis of uncertainty – Kriging
Part
4: modeling of variability and the
analysis of uncertainty – Simulation
1000 Break
1030 Ocean Observatories Initiative – Larry Clark, National Science Foundation
Conference Center
1200 – lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory Lab - John Welhan
modeling
of variability and the analysis of uncertainty
Saturday, July 12
Swimming party and barbeque at Mary Jane Perry’s
house in Whitefield
WEEK
IV
Day 16 Monday, July 14
Kresge Classroom
0830 Introduction and Data Analysis-
spatial variability – guest lecturer Gerard Heuvelink,
1000 Break
1030 Oceanography
- Collin Roesler or Andy Thomas
Conference Center
1200
– lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory - Dr. Gerard
Heuvelink
Day 17 Tuesday,
July 15
Kresge Classroom
0830 Data Analysis- spatial variability –Dr.
Gerard Heuvelink
1000 Break
1030 Oceanography - - Collin Roesler or Andy Thomas
Conference Center
1200
– lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory - Gerard
Heuvelink
Day 18 Wednesday,
July 16
Kresge Classroom
0830 Data Analysis: Gerard
Heuvelink
1000 Break
1030 Data Analysis: guest
lecturer, Dr. Phaedon Kyriakidis,
UCSB
data
integration issues, change of support data integration issues,
change
of support
Conference Center
1200
– lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory - Gerard
Heuvelink and Phaedon Kyriakidis
Day 19 Thursday,
July 17
Kresge Classroom
0830 Fluorescence quenching and merging of proxies
-
Mary Jane Perry
1000 Break
1030 Sochastic simulation for uncertainty
assessment - Dr. Phaedon Kyriakidis
Conference Center
1200
– lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory - Phaedon Kyriakidis
Day 20 Friday,
July 18
Kresge Classroom
0830 Oceanography - Emmanuel Boss
1000 Break
1030 Geostatistical space-time model - Phaedon
Kyriakidis
Conference Center
1200
– lunch
Classroom in the Marine Culture Building
1330 Statistics Laboratory - Phaedon Kyriakidis
Saturday, July 19
Party (location TBD)
WEEK V
Day 21 Monday, July 21
Kresge Classroom
0900 Visualization
-- Guest lecture: Larry Mayer, UNH
Classroom in the Marine Culture
Building
PM projects
Day 22 Tuesday, July 22
Kresge Classroom
0830 Point
processes – Kate Beard
1000 CLASS ROOM is unavailable, due to Ann Simpson’s video conferenced Master’s defense. You can use the dining hall to work with your laptops off line.
Classroom in the Marine Culture
Building
PM projects
Day 23 Wednesday, July 23
Kresge Classroom
0830 Inversion
models – Collin Roesler
1030 Frontiers
in optical measurements – Emmanuel Boss
Classroom in the Marine Culture
Building
PM projects
Day 24 Thursday, July 24
AM projects
1100 Course evaluation
Conference Center
1600 Darling Marine Center
RiverNet: Distributed sensor nets for environmental monitoring
- Arthur Sanderson, Renesselaer Polytechnic
Institute
1800 Dinner with Dr. Sanderson
Day 25 Friday, July 25
Kresge Classroom
0900 Student
presentations
Clean up labs
Saturday, July 26: Students depart
Class Pictures:
Figure 1 Statistics lab
Figure 2 Lecture
Figure 3 AC9 Lab
Figure 4 Ira C sampling for AC9 lab
Figure 5 Ira C sampling for AC9 lab
Figure 6 Filtering chlorophyll samples
Helpful Links:
* About the course: check http://www.dmc.maine.edu/html/courses.html
or contact Dr. Mary Jane Perry, perrymj@maine.edu, (207) 563-3146 ext 245;
* About the applications:
Ms. Linda Healy, lhealy@maine.edu, FAX: (207) 563-3119;
* About the
* A partial list of Matlab tutorials on the WWW. Some are
older than others. Luckily, most basic operations have not changed between
versions.