ANDROSCOGGIN COUNTY, ME LAND COVER CHANGE ANALYSIS: A SUCCESSFUL COLLABORATION

1
ANDROSCOGGIN COUNTY, ME LAND COVER CHANGE
ANALYSIS A SUCCESSFUL COLLABORATION

• Russell G. Congalton, Jennifer L. Bourgeault,
  Mimi L. Becker
• GLOBE Land Cover/Biology Team
• 215 James Hall, 56 College Rd
• University of New Hampshire
• Durham, NH, USA 03824

2
PARTNERS

• GLOBE Land Cover Investigation Team at the
  University of New Hampshire
• Auburn Land Lab
• Over 500 teachers and students from Auburn
  schools over a period of three years

3
MOTIVATION

• The Land Cover/Biology Team had a great resource
  in Patty Gaudreau. She was a dedicated middle
  school GLOBE teacher who helped start the Auburn
  Land Lab where environmental learning took place
  for all the Auburn schools.
• Patty believed that all teachers should be GLOBE
  trained and GLOBE was a great vehicle for
  learning.
• All 7th graders at the Auburn Middle School were
  receiving laptops and teachers were encouraged to
  use them.

4
MOTIVATION

• The land cover and land use in Lewiston-Auburn
  had changed drastically in recent years -
  students had noticed.
• Students wanted to connect the existing
  recreational trails together throughout Auburn
  but needed more information about the land cover
  in order to do this.
• All Auburn teachers had to use activities and
  lessons that directly met the State of Maine
  Learning Results.
• The community government had active and
  interested GIS experts.

5
INTRODUCTION

• Study combined student-collected data with
  researcher-collected data in a land cover
  classification and change detection analysis of
  Androscoggin County, Maine.
• Teachers, assisted by the Land Cover Team,
  trained their students, developed a lesson plan
  that met their Learning Results objectives and
  used the computers all 7th grade students had
  been given to cover the Land Cover Protocols.

6
RESEARCH AREA

• The study site Androscoggin County, Maine.
• The county contains Maines second and fourth
  largest communities Lewiston (population 35,690
  in 2000), and Auburn (population 23,203 in 2000).

7
Study Site Maine with Androscoggin County
highlighted on the left. Androscoggin County with
Townships labeled on the right.
8
THE DATA

• GLOBE students collected land cover data from 200
  usable sites using the Land Cover/Biology
  Protocols.
• land cover class
• latitude, longitude, and elevation from a GPS
• photos for quality control
• Student data supplemented data collected by the
  UNH GLOBE Land Cover Team.
• Land cover types that were rare in the study area
  and not randomly sampled frequently were
  supplemented using image and photo interpretation
  and GIS data layers.
• Wetlands (MUC 6) and Barren lands (MUC 5).
• The majority of the student-collected data were
  used in the accuracy assessment of the 2000 land
  cover map.

9
Distribution and amount of data collected by
GLOBE students and the UNH Land Cover Team.
This figure reveals the significant data
collection effort completed by the GLOBE
students. These data are critical to this
project.
10
THE DATA

• The Landsat images
• 16 September, 1993 (Landsat 5 TM)
• 28 August, 2000 (Landsat 7 ETM)
• Why appropriate
• proximity of collection month and day
• study area was cloud free in both
• reduced the chances of detecting false changes
  due to differences in vegetation phenology caused
  by the images being from different seasons of the
  year.

11
Example of the Landsat TM imagery for
Androscoggin County, Maine.
12
QUALITY CONTROL

• All student-collected data were subjected to
  initial quality control measures.
• Examined by MUC class
• Is it possible within the study site? (e.g. there
  is no tundra in Androscoggin County).
• GPS coordinates recorded correctly
• Is it in decimal degrees?
• Are there four decimal places? (e.g. 44.1212 was
  acceptable where 44.12 was not)
• The points were mapped on digital ortho quads
  (DOQs). Points outside the areas where the
  student teams visited were removed.
• If the MUC class did not match photo
  interpretation they were not used.

13
PARTNERSHIP

• In Year 2, the Land Cover Team provided the
  student teams with route maps of the area each
  group would cover during data collection.
• Goal Spread collection of data points over a
  large area
• Student teams were instructed to delineate a
  polygon on the map for each LCSS visited.
• Maps were used as quality control for the
  latitude and longitude.
• error in the unit or
• with the recording and calculation of the
  latitude and longitude by the students

14
Sample Map given to the student teams collecting
land cover data during the spring of 2003.
15
PARTNERSHIP

• Student teams collected land cover data near or
  within the 15km x 15 km area depicted on their
  Landsat imagery centered on Auburn Middle School
  (AMS)
• covered approximately 18 of the study area
• Additional land cover data were collected by the
  UNH Land Cover Team.
• using the exact same protocols as the students
• outside of the area covered by the student teams
• increase the sampling area to encompass the
  entire county

16
Study area for students participating in the
GLOBE Program at Auburn Middle School, compared
to entire study site.
17
MUC-A-THONS

• 500 students and teachers were trained in Land
  Cover Protocols and collected data.
• UNH Land Cover Team provided
• initial training and on-going support
• assistance for the first and second MUC-A-THONs
• Subsequent MUC-A-THONs conducted by teachers with
  support from the Auburn Land Lab.

18
TRAINING FOR A MUC-A-THON

• Classroom Training
• First year MUC only
• Second year lesson plan for a Land Cover Unit
  was developed, many of the field training
  stations were covered before the actual field
  training day
• Field Training (just before the MUC-A-THON)
• Student training took place at LC stations
• Data Sheet and practice data entry
• Compass and photos
• Canopy and ground cover, estimating tree height
  (5 meters)
• GPS and pacing
• Chaperones were trained separately

19
FIRST YEAR

• 250 students, teachers, community and parent
  volunteers (7 buses with 5-8 teams per bus)
• Ticks, mud, laughter, and lots of land cover
  sites by the end of the day!
• Each bus route covered a set of sampling sites
• A bus dropped off anywhere from 5-8 teams of
  students along a designated route and picked them
  up a couple hours later.
• Each team consisted of one high school student
  (sophomore), 2 middle schoolers, one elementary
  student and one chaperone.
• The students were in charge since they were
  trained in the protocols - they had to take the
  lead.

20
SECOND AND THIRD YEAR

• Limited to 7th graders
• Focus changed from a multi-level learning
  experience to meeting the Maine State Learning
  Results Requirements and using the laptops
• Planned as a unit rather than a field exercise

21
RESULTS

• Land cover maps were generated from Landsat TM
  imagery for both the 1993 and 2000 images.
• LC changes occurring between these two dates were
  identified
• Accuracies of the land cover maps were in the mid
  60 with the highest accuracy of 68 for the 2000
  land cover map.

22
Land cover maps generated from Landsat TM imagery
for 1993 and 2000.
23
RESULTS

• Key Data Quality Issues student data were uneven
  for positional accuracy.
• Cause Failure to record GPS data to 4 decimals
  (highest level of precision) resulted in rounding
  off of the coordinates with direct loss of
  positional accuracy.
• Mostly corrected by using the route maps where
  students had sketched the polygons of their
  sites.
• Critical to Accuracy all future GPS locations be
  collected with the highest level of precision
  recorded on the GPS device. (DO NOT round off
  the coordinates as this results in a significant
  loss in positional accuracy of these key
  monitoring locations.)

24
OUTCOMES

• Detailed land cover/vegetation change detection
  analysis from Landsat Thematic Mapper (TM)
  imagery acquired in 1993 and 2000.
• The students will use these results to aid in
  planning a recreation trail within the county.

25
ACKNOWLEDGEMENTS

• Funding for this research was provided by the
  National Science Foundation (NSF) through the
  GLOBE Program under grant GEO9801750 and is
  gratefully acknowledged.
• Thanks to David West, Larry Ryan, and Amanda
  Scally for their time and energy put into the
  GLOBE Program, training the teachers and
  students, and facilitating the MUC-A-THONs.
• Special Thanks to Patricia Gaudreau at the
  Auburn Land Lab. Without her dedication to the
  GLOBE Program and her students interest in the
  trail system, this research collaboration would
  not have taken place.

26
7th Grade GLOBE Land Cover Unit

• Steve Robertson, Nicole Melcher, Leslie Magalis,
  Cameron Parker, Patty Gaudreau, Jennifer
  Bourgeault (UNH Land Cover Team)

27
PRE-MUC AND SUGGESTIONS

• Encourage math teachers to incorporate
• calculation of percentages
• basic trig for Clinometers
• metric system
• Encourage social studies teachers to incorporate
  
• latitude/longitude lessons
• GPS coordinate systems
• Letters to teachers and chaperones. Teacher
  letter had class schedule and explanation of
  curriculum.
• If students have questions about any of the GLOBE
  process, use the Ask the Scientist on the GLOBE
  website.
• Land Lab/UNH staff will come to team meetings
  prior to MultiSpec lesson to discuss upcoming
  curriculum and program (roles, overview of
  lessons, set-up)
• Add Investigation Process to Standards assessed
  with this program.

28
WHAT IS MUC? LESSON

• Before first lesson GLOBE Website scavenger hunt
  and quiz.
• MUC Guide Lesson
• discuss MUC codes
• global land cover
• vocabulary
• Taught by science teachers
• Assessment
• file of 20 images
• Use pictures from file and each student will MUC
  them individually.
• Before next lesson
• Show Remote Sensing Video from GLOBE. Video
  quiz assessment.
• Have students install MultiSpec folder onto
  laptops.

29
MULTISPEC LESSON

• 80-90 minute lesson in classrooms
• Taught by UNH/Land Lab Staff
• Start with hard copies of change images. Use to
  ask succession and land cover change questions
• Finding the familiar (orienting themselves).
• Is our school still surrounded by the same land
  cover?
• Find an area, compare and contrast this between
  images.
• What causes difference, how do we find out if it
  is true (introduce scientific method)
• Differences between true and false color images.

30
MULTISPEC LESSON

• Walk through MultiSpec.
• Basic skills.
• Create cluster maps using MultiSpec.
• Save clustered maps in ArcView format.
• Next year, with a little advanced work, open
  these maps in ArcVoyager, drop the MUC-A-THON
  points right on top and look at accuracy of the
  clustered maps.
• Assessment
• Take Pretest on succession. (groups of 2-3)
• Keep them to hand back after Post-test.
• Before next lesson Do MultiSpec tutorial with
  students.

31
PROTOCOL LESSON

• Taught by Land Lab Staff _at_ Land Lab
• Three rotating sessions
• GPS/Pacing
• Compass/Photo
• Forest/Woodland protocols
• Look at real data sheets and GPS forms.
• Assessment MUC Exam

32
MUC PRACTICE LESSON

• Taught by science teachers
• Use site behind AMS to discuss 90x90 sites and
  data entry form.
• Walk through MUC guide.
• Reiterate homogeneous sites and pacing.
• Assessment Have students MUC a site near their
  homes.
• Before next lesson
• Double check chaperones, cell phones, digital
  cameras.
• Students should know their roles and team
  assignments before day of MUC-A-THON.
• Reserve time for next days data entry lesson.

33
MUC-A-THON AND DATA ENTRY

• Taught by chaperones and teachers
• Download pictures onto laptops after MUC-A-THON
• send to GLOBE on Data Entry day
• Data entry the next day
• students will learn how to do this together after
  MUC-A-THON
• lesson and website
• Students with photos meet with computer
  technology person to forward them to GLOBE.
• Assessment Reflection
• Before next Lesson Prepare map with MUC points.

34
MAPPING/COMPARISON LESSON

• Taught by science teachers
• Bring up cluster maps and discuss where they
  MUC-ed.
• Cluster maps were labeled using their own terms
  (forest, water, field, urban, etc.).
• Change common names to MUC classes so they can
  use the MUC-A-THON data for accuracy assessment.
  
• Have a map in ArcInfo to look at and find MUC
  sites.
• AA Matrix. The students had no problem with the
  accuracy assessment matrix.
• Compare the classes they gave to their cluster
  maps and correct them with ArcInfo information.
  
• Assessment Post-test Succession assessment.
• Hand back pre-test to see if MUC-ing helped
  answer their question.