ESSE21 logo
LLU-ESSE21 LULC Module banner
Geobrain project logo

LAB GUIDE

QUICK LINKS

Instructions for Learners and Teachers Where to Get the Data Review of LULC (Land Use/Land Cover) Classification Systems How to do a Supervised and Unsupervised Classification Doing Change Detection Criteria for a Good Map

Instructions for Learners and Teachers

Overview Instructions:

The lab exercise included below (PDF format) is meant to supplement regular remote sensing teaching materials from textbooks, other lab exercises, e.g. from the Remote Sensing Core Curriculum or learning modules from ERDAS Imagine, IDRISI Selva, QGIS or similar image processing software. It is NOT meant to be a step-by-step instructional guide. We utilize datasets from theHonduras--Case Study No. 1 and its Virtual Tour--for illustrative purposes. And we link to other resources online from FAOs GLCN-Land Cover Topic Centre. See also:

Sally J. Westmoreland, University of Redlands

No. 2 > April 4, 2006 - UNAH-OACS Tim Foresman, University of Maryland and ISRSE

 

Other Supplementary Resources:

NOAA - Coastal Services Center:

Conversion Tools, Utilities, other Key Resource Aides:

Other Classification Exercises:

Back to TOP

Where to get the Data and
Tools and Utilities for Conversion...

A) Using provided datasets (.img - ERDAS format)

B) Using provided LANDSAT and ASTER images

C) Download your own data (from free web sources)

D) Conversion tools and utilities and other resources

 

A. If you use included datasets (download from lists below):

Note: These images were prepared as subsets of larger LANDSAT images of the northern coastal region of Honduras centered on the area between La Ceiba Honduras and Tela where Cuero y Salado Wildlife Refuge is located and they are already formated for use in ERDAS Imagine, IDRISI Selva, QGIS or similar image processing software.

2003 DATA FILES - ERDAS COMPATIBLE

2000_03_29_sub.img 10-Apr-2006 12:37 12M

2000_03_29_sub.rrd 10-Apr-2006 12:37 1.2M

2001_03_16_sub.img 10-Apr-2006 12:38 12M

2001_03_16_sub.rrd 10-Apr-2006 12:38 1.2M

2003_03_06_sub.img 10-Apr-2006 12:38 12M

2003_03_06_sub.rrd 10-Apr-2006 12:38 1.2M

 

 

 

 

 

 

 

 

 

2001 DATA FILES - ERDAS COMPATIBLE

ihs_fcc_0316_01.img 14-Apr-2006 16:41 211M

ihs_fcc_0316_01.rrd 14-Apr-2006 16:42 18M

subset_6b_031601.img 14-Apr-2006 16:42 107M

subset_6b_031601.rrd 14-Apr-2006 16:42 9.8M

subset_pan_03601.img 14-Apr-2006 16:43 70M

subset_pan_031601.rrd 14-Apr-2006 16:43 6.1M

 

 

 

 

 

 

 

 

Back to TOP

B. If you use select LANDSAT and ASTER images provided:

 

LANDSAT/MSS & ASTER DATA FILES

North Coast

ASTER Data:

a) Montagua Valley (Guatemala/Honduras border = AST_L1B_00302212004162932_03122004095231_VNIR.jpg

b) Omoa-Puerto Cortez coastline = AST_L1B_00302122004163537_03032004105610_VNIR.jpg

c) Trujillo region = AST_L1B_003_03122002162352_03292002082147_VNIR.jpg

d) Tela Bay region - coastal zone to Cuero y Salado Reserve (west of La Ceiba) = AST_L1B_003_03062003162917_03242003154406_VNIR.jpg

e) Tela Bay - Interior, e.g. Texiguat Preserve and upper San Juan River basin = AST_L1B_003_11082000164217_07192001022937_VNIR.jpg

f) La Ceiba / Utila region - Coastal zone from Cuero y Salado Reserve to about Sambo Creek (east of La Ceiba) = AST_L1B_003_09242001163447_10112001094803_VNIR.jpg

LANDSAT/MSS Data:

g) Trujillo region and Roatan/Guanaja / Cayos Cochinos (visible-color image) 1985 = p017r49_5t850201nn1.jpg

h) Trujillo region and Roatan/Guanaja / Cayos Cochinos (False-Color Infrared image) 1973 = p018r49_1m1973121901.jpg

i) Omoa Bay to La Ceiba & Utila/West End/Roatan (Visible color) 1987 = p018r49_5t870318nn1.jpg

j) Omoa Bay to La Ceiba & Utila/West End/Roatan (Visible color) 2000 = p018r049_7t20000329_z16nn10.jpg

k) Omoa Bay to La Ceiba & Utila/West End/Roatan (Visible color) 2001 = p018r049_7t20010316_z16nn10.jpg

l) Omoa Bay to La Ceiba & Utila/West End/Roatan (Visible color) 1979 = p019r49_3m1979020401.jpg

 

Source: JPL - ASTERweb - TerraLook Collections - - PAA_Collections_Preliminary - Categories

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LANDSATMSS & ASTER DATA

Pacific Coast

ASTER Data:

a) ASTER - Golfo de Fonseca (Cholute) region (visible-color) 2003 =AST_L1B_00310162003162927_05262004164023_VNIR.jpg

LANDSAT/MSS DATA:

b) Golfo de Fonseca (Choluteca) region - to Lake Nicaragua (False-color infrared) 1979 = p018r51_2m1976012301.jpg

c) Golfo de Fonseca (Choluteca) region - to El Salvador (visible-color) 1990 = p018r51_5t900206nn1.jpg

d) Golfo de Fonseca (Choluteca) region - to El Salvador (visible-color) 2002 = p018r051_7t20020506_z16nn10.jpg

e) Golfo de Fonseca (Choluteca) region - to El Salvador (False-color infrared) 1979 = p019r51_3m1979020401.jpg

Source: JPL - ASTERweb - TerraLook Collections - - PAA_Collections_Preliminary - Categories

 

 

 

 

 

 

 

 

 

 

 

 

 

Back to TOP

C. If you download your own data:

GET LANDSAT DATA from GLCF (Global Land Cover Facility)

LANDSAT 7 data for Northern Honduras (Path 018 Row 049)

ETM+
WRS-2, Path 018, Row 049
2000-03-29
EarthSat
Ortho, Geocover
Guatemala, Honduras
Online: 042-430

Scenes: 042-431 Acq. date = 2001-03-16

 

 

 

 

Online Imagery and Data Sources HERE:

Back to TOP

 

D) Conversion Tools, Utilities, other Key Software Resource Aides:

Review LULC (Land Use/Land Cover) Classification Systems - FAO, USGS, IGBP

Digital image data are frequently the basis to derive land use/land cover information over large areas. Classification of image data is the process where individual pixels that represent the radiance detected at the sensor are assigned to thematic classes. As a result the image is transformed from continuous values, generally measured as digital numbers (DN) or brightness values (BV) to discrete values that represent the classes of interest. Traditionally the algorithms employed for this process differentiate and assign pixels based on the values recorded at that pixel for each of the wavelength regions (bands) in which the sensor records data.

A universal land use/land cover classification system does not exist. Instead, a number have been developed to reflect the needs of different user(s). Typically, the systems are hierarchically arranged with the ability to consolidate lower level classes into the next highest level and with a consistent detail for all classes at a given level in the hierarchy. In selecting a classification system for use with remotely sensed data, the classes must have a surface expression in the electromagnetic spectrum. For example, a crop such as pineapples reflects electromagnetic radiation, but an automatic teller machine (ATM) on the side of a building cannot easily be detected, especially from most down-looking sensors. In addition, the resolution characteristics of the imagery selected must be compatible with the classification system. That is, the imagery must have the spatial detail, spectral discrimination and sensitivity, and temporal characteristics required for the classes of interest.

For purposes of applying the system for image classification, we must differentiate the information classes represented by the land use/land cover classification system from the spectral classes that we can obtain from the imagery. Often an information class will have a range of spectral responses that represent the inherent variability within a class that is intended to capture like activities. This may be due to composition of covers that are necessary to express the class, e.g., residential class would include materials for roads, lawns/gardens, rooftops and other building materials. Or, multiple land covers may individually satisfy the criteria for a given land use class, e.g., crops of barley, corn, lettuce, sugar beets, etc are all in the class “field crop”, but would have different spectral responses. The diurnal and seasonal aspect also contributes to spectral variability due to variation in planting dates, vegetation phenology, and illumination. Thus multiple spectral classes, called signatures or training sites, that capture the variability are required to represent a single information class.

Description of Study Area

In this module, we work through techniques to classify a portion of the North Coast of Honduras using Enhanced Thematic Mapper imagery for March 2003. The classification scheme is one used by Forestry Department in Honduras and is based on the FAO Land Cover classification system. For purposes of this exercise, we are working with 30m spatial resolution data and that is consistent with Level II of the system and in some cases Level III. The image processing routines described are based on ERDAS Imagine software. Below are other resources from a workshop held in Honduras (April 2006) that will give you further introduction to LULC.

Select Powerpoints from LULC Workshop in Honduras:

No. 1 > April 3, 2006 - UNAH-OACS Tim Foresman, University of Maryland and ISRSE

Evening April 3 - to OACS Students Sally J. Westmoreland, University of Redlands

Sally J. Westmoreland, University of Redlands

Other resources Online:

Study and evaluate other Global, US, Canadian, European landuse/land cover classification systems and compare their advantages and disadvantages over the FAO/GLCN system. See for example:

Back to TOP

How to do a Supervised and/or Unsupervised Classification

See the PDF lab module attached HERE

Back to TOP

Doing Change Detection

 

The goal of doing LULC classification is primarily to do change detection--that is compare land cover classes at different periods of time and then project/describe the extent of change in use OR cover for purposes of monitoring trends in human management of the landscape or to assess "natural" or "human-induced" changes.

There are many examples of tutorials and other resources that describe change detection on the web. Some of the best focused on coastal zone change are from NOAA - Coastal Services Center--see below:

Essentially the procedure is to:

  1. create a supervised classification of images from two different time periods,

  1. overlay the two classified images, and then

  2. highlight the changed pixels between time 1 > 2,

  3. then, through careful ground-truthing and image interpretation, assess what are the "driving forces"--causes--of the changes observed.

This is the hardest part and requires some local knowledge of the environment and knowledge domain so one can sort out spurious (pixel artifacts) from actual/real change on the ground. See for example the following images of changes between 2001-2003 on the North Coast of Honduras (these are from the data/images used in the LULC TUTORIAL see below:

Back to TOP

Criteria for a Good Map

Here is a basic Map/Graphic checklist to help you assess the final output of your lab work which in most cases will be a map or series of maps. You may also be asked to produce a portfolio or poster with additional explanatory or content information--see criteria/rubric for policy-briefs or papers...


Checklist for Evaluation of Maps and Graphics

          Key Map Elements included: Title,
          Scale (graphic and Ratio), Legend,
          Compass Rose or North Arrow).....................
          1     2     3     4     5

          Uses appropriate projections and datums.......1     2     3     4     5

          Grammar, spelling, fonts, lettering
          (appropriate for users and audience).............1     2     3     4     5

          Visual appeal (color/line quality)....................1     2     3     4     5

          Accuracy/types of spatial data and
          analytic methods fully documented.................1     2     3     4     5

          Conciseness and focus (not too busy).............1     2     3     4     5

          Relevance to user needs and audience............1     2     3     4     5

          Persuasiveness of map--to the point...............1     2     3     4     5

          Timeliness of content and data........................1     2     3     4     5

          Quality and diversity of sources cited
          (all sources cited appropriately)......................1     2     3     4     5


                         TOTAL POINTS POSSIBLE   =     50

          Points Earned X 2   =    ___ (Total out of 100 points)

LLU-ESSE21 LULC Module logo

Back to TOP