Meyer, H.; Thies, B. & Bendix, J. (2011): <b>Land use/land cover map of South Ecuador 1987.</b> <i>Available online (http://www.tropicalmountainforest.org/data_pre.do?citid=1038) from DFG-FOR816dw.</i>
email:
bendix <at> staff.uni-marburg.de
Faculty of Geography
Deutschhausstraße 10
Philipps University of Marburg
Laboratory for Climatology and Remote Sensing
35032 Marburg
Germany
Abstract:
The map shows land use/land cover (LUC) of the Loja-Zamora region in 1987. It consists of 6 LUC classes: Forest, Subparamo, Paramo, Pasture, Urban and Water. The classification is mainly based on a Landsat TM image from 1987/03/26. However, since it features high cloudiness, a composite of 8 further images around the year 1987 was used for filling the gaps.
Additional Infos:
The map is available in UTM 17S (WGS 84) projection. <br/>
The basis for the classification was the Landsat Scene LT50100631987085XXX02 (accessed in New Earth explorer (http://edcsns17.cr.usgs.gov/NewEarthExplorer/) from 1987/03/26. The following TM scenes were used for filling the clouded areas of 1987: <br/>
-1985/03/20 (LT50100631985079AAA04)<br/>
-1986/12/04 (LT50100631986338XXX03)<br/>
-1986/11/02 (LT50100631986306XXX09)<br/>
-1989/12/28 (LT50100631989362XXX02)<br/>
-1990/03/02 (LT50100631990061XXX02)<br/>
-1998/07/14 (LT50100631998195XXX01)<br/>
-1998/08/15 (LT50100631998227XXX01)<br/>
-1998/12/21 (LT50100631998355XXX01)<br/>
<br/>
The procedure steps 1-5 were accomplished for each of the 9 Landsat images.
Keywords:
| land use | Loja | land cover | landsat | remote sensing |
Associated entities to this dataset:
--------
1
. spatial raster entity --------
Raster Name:
classification1987
Tech. Details ...
Physical Structure Description:
Object Name:
classification1987.rst
Size (byte):
5814555
External Format:
Idrisi Ratser Format
raw image file
Object Name:
classification1987.rdc
Size (byte):
693
External Format:
Idrisi Ratser Format
description file
Info about Georeference:
Number of Bands:
1
Rows:
2649
Columns:
2195
Verticals:
0
Spatial Reference System:
WGS_1984_UTM_Zone_17S
Cell geometry:
pixel
Cell size X:
30.0
Cell size Y:
30.0
Raster Origin:
upperLeft
Horizontal Accuracy:
not assessed
Vertical Accuracy:
not assessed
Corner Point:
xCoordinate:
691590.0
yCoordinate:
9519830.0
pointInPixel:
lowerLeft
corner:
lowerLeft
xCoordinate:
757440.0
yCoordinate:
9599300.0
pointInPixel:
upperRight
corner:
upperRight
Attribute(s) ...
Attribute Name -- Attribute Label
Unit
Attribute definition
Land_use
--
land use
no unit
Land use
Method
(Sampling/processing description of the values of
this attribute by one ore more steps.)
--- Step
1
---
Title
:
Cloud mask development
Description
:
Training sites of clouded and unclouded areas were selected. A maximum likelihood classifier was then used for cloud classification. Since cloud edges and especially cloud shadows could not be identified in this way, they were digitized manually.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
--- Step
2
---
Title
:
Atmospheric Correction (Cos(t) model)
Description
:
The Atmospheric correction was accomplished using the Cos(t) model (Chavez 1996) which is implemented in IDRISI (Eastman 2006). The information used for the correction could be found in the metadata of the Landsat images.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
Citation:
Eastman, J
.
(2006):
IDRISI Andes guide to GIS and image processing
. Available online
(http://tropicalmountainforest.org/publications.do?citid=1037) from DFG-FOR816dw.
Citation:
Chavez, P
.
(1996):
Image-based atmospheric corrections – Revisited and improved
. Available online
(http://tropicalmountainforest.org/publications.do?citid=1036) from DFG-FOR816dw.
--- Step
3
---
Title
:
Topographic correction
Description
:
The topographic correction was done by a statistical- empirical approach described in Eastman (2006). For this purpose, a SRTM 90m digital elevation model and metadata from the Landsat image were used.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
Citation:
Eastman, J
.
(2006):
IDRISI Andes guide to GIS and image processing
. Available online
(http://tropicalmountainforest.org/publications.do?citid=1037) from DFG-FOR816dw.
--- Step
4
---
Title
:
Removal of stripes
Description
:
Since the first 2 channels of some images were affected by along-scan horizontal banding, a principal component analysis (PCA) was accomplished. The first 5 of 6 principal components were back calculated into the first two channels by inverse PCA.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
--- Step
5
---
Title
:
Adaptation of reflectances
Description
:
A linear regression with the Landsat 1987 scene was done with each image individually. The coefficients were used for calibrating the image using the ?calibrate? module in IDRISI. This was important since there was a bias in the reflectances of the different images.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
--- Step
6
---
Title
:
Image composite creation
Description
:
The 1987 image was the basis for the image composite. The cloud mask of the 1987 image which was created in step 1 was stepwise replaced by cloudfree pixels of the other images. The order of the images which were replacing the clouded pixels was:
1986/11/02; 1986/11/02; 1985/03/20; 1989/12/28; 1990/03/02; 1998/07/14; 1998/08/15; 1998/12/21.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
--- Step
7
---
Title
:
Image Splitting
Description
:
Since they were some image-dependent differences of reflectances in the composite, it was splitted into 12 parts for better control in the classification procedure.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
--- Step
8
---
Title
:
Choice of training sites
Description
:
Training sites of LUC were chosen for each of the 12 parts of the composite by visual interpretation and comparison to the 2001 classification (Göttlicher 2009).
Göttlicher D. (2009): Landcover classification Ecuador (10 classes). Available online (http://www.tropicalmountainforest.org
/data_pre.do?citid=704) from DFG-FOR816dw.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
--- Step
9
---
Title
:
Image classification
Description
:
Spectral signatures for the LUC classes Forest, Paramo, Pasture, Urban and Water were developed (Eastman 2006). A maxlike classifier was then used for supervised image classification (Eastman 2006). In order to flatten the results, a 3x3 mode filter was applied to the image. Finally the splitted image parts were merged to one image.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
Citation:
Eastman, J
.
(2006):
IDRISI Andes guide to GIS and image processing
. Available online
(http://tropicalmountainforest.org/publications.do?citid=1037) from DFG-FOR816dw.
--- Step
10
---
Title
:
Distinguishing forest from subparamo
Description
:
The spectral differences in the composite deriving from the different input images did not allow to distinguish between different forest types. Thus, Subparamo was distinguished from other forest types using information from the 2001 classification (Meyer et al. 2011) Under the assumption that Subparamo does not change into other forest types, forest in the 1987 classification was reclassified as Subparamo if it was Subparamo in 2001.
Meyer H. , Thies B. , and Bendix J. (2011): Land use/land cover map of South Ecuador 2001 . Available online (http://www.tropicalmountainforest.org/data_pre.do?citid=1039) from DFG-FOR816dw.
Instrumentation:
Instrument:
Description:
|
Instrument: Idrisi
|
Vendor: Clark Labs
|
email:
bendix <at> staff.uni-marburg.de
Faculty of Geography
Deutschhausstraße 10
Philipps University of Marburg
Laboratory for Climatology and Remote Sensing
35032 Marburg
Germany
