Chapter 4: Monitoring Forest Operations using Medium Resolution (30 m) Imagery
4.1 Materials and Methods
4.1.2 Materials
Chapter 4: Monitoring Forest Operations using Medium
included to try and include some of the main site factors, such as geology, soil and climate, in the study sites. This process was undertaken using ArcGIS 8®(ESRI, 2000), and resulted in suitable compartments being identified to form the study area.
A set of 162 compartments in 20 plantations were identified from which ground- truthing sites were selected. For display purposes only 15 of these compartments are shown (see Sample Illustration Site - Figure3.1.1).
4.1.2.2Satellite Imagery Data
The satellite imagery applied in this study was standard Landsat 7 multi-spectral imagery purchased from the Satellite Applications Centre (SAC) of the CSIR. The selection of the Landsat 7 sensor was based primarily on a cost/benefit basis, and was a balance between obtaining imagery with a resolution small enough to detect the critical spectral information required, and being available at a reasonable cost.
Another factor to be considered was the repeat cycle, especially as this study required imagery at a much closer interval than is normally the case for change detection processes. The revisit cycle of sixteen days for Landsat 7 was suitable, especially over periods of high cloud cover, as it allowed a greater chance of obtaining a cloud-free image.
The specific Landsat images used in this study were selected on the basis of the following criteria:
1. Imagery with a maximum amount of cloud-free coverage over the areas of interest.
2. Imagery covering as much of the seasonal variation as practically possible, but with at least a mid-winter and mid summer image, together with an inter-seasonal image. As many of the forestry operations that were to be monitored as part of this study reach a peak in mid to late summer (January to May), this was a particular period of interest.
3. Imagery that covered all the areas of interest within the study site in a single image.
The images were obtained from SAC as geo-referenced to Transverse Mercator, Lo 31°, Clark 1880 spheroid, Cape Datum (Le. equivalent to USGS L1G processing level). They were supplied on compact disc in ERDAS Imagine® .IMG format.
A total of four images were purchased, covering the period from March 2002 to April 2003. Technical specifications of the sensor are provided in Appendix 1, together with a sample of a raw image.
The four images obtained for this study were captured on the following dates:
First image - 30.03.2002 (referred to as the March image in the text). Second image - 18.06.2002 (referred to as the June image in the text).
Third image - 28.01.2003 (referred to as the January image in the text).
Fourth image - 02.04.2003 (referred to as the April image in the text).
Details of the images are summarised in Table 4.1 below.
Table 4.1 Summary of Landsat 7 images used in study
Image Date Time Frame Cloud Sun Sun
Acquired Acquired Details: Cover Azimuth Elevation (GMT) Path/Row
March 30.03.2002 07:39:33 K-J 168/080 0000 52.11° 42.89°
June 18.06.2002 07:39:16 K-J 168/080 0000 36.19° 28.00°
January 28.01.2003 07:39:16 K-J 168/080 0000 81.2]0 54.15°
April 02.04.2003 07:39:29 K-J 168/080 0000 50.94° 42.24°
Source: SAC, 2003.
These images were standard Landsat 7 (ETM+) images, including the seven multi- spectral bands (Le. Bands 1 to 7), with a 30 m spatial resolution, plus a panchromatic band image with a 15 m resolution. The panchromatic images were not utilised in this study. For analysis purpose, all multi-spectral bands (excluding the thermal infrared (TIR) band) were used. However, for display purposes, an RGB composite image was used comprising Bands 4, 3, and 2.
4.1.2.3 Forest Management Attribute Data
The forest management information is recorded on a monthly basis by the field staff and captured into an Informix relational database called the Forest Management System (FMS). All operational information is captured into this database, which is linked to the GIS using key fields. Reports can be drawn from FMS, either using the standard menu system, or using SQL.
Specific information that is required by the GIS on a regular basis is available through the creation of standard "views" or query tables in FMS. These include the compartment register, which lists all the base information for every compartment, and certain planning information such as felling or planting plans.
A record is kept of all operations that have an operation code, for the lifespan of a compartment, including some information if a compartment is merged with another compartment. If the compartment is deleted however, the history is deleted as well.
This historical aspect of the database has a key role to play when it comes to comparing the data derived from the satellite imagery with the database data.
Compartment data, both as base data and historical operational data, were extracted from FMS and used in this study to compare with the data classified by the imagery, in order to test the hypothesis proposed for this study. Specific compartment information extracted from FMS included species, felling dates, planting and coppice (see glossary) dates and growth cycle. Historical information extracted included when burning, planting and other establishment operations had occurred, as well as when weed control operations had been done. This information was queried on a compartment basis.
4.1.2.4 Geographic Information System (GIS) Data
In addition to the forest management information, a complete set of digital spatial data for all forestry landholdings in the study area was maintained through the application of GIS technology. This information included the legal cadastral information, compartment (commercial and non-commercial) information, roads, rivers, contours, buildings, dams and a considerable amount of derived information such as slope classes, climatic data and other such information.
The cadastral information provided a means to enable the image processing functions to be focussed on the specific areas of interest. This reduced the processing time as well as reducing the disk space requirements, although these still proved to be very intensive. The compartment information provided the units of observation on which the image processing and analysis were based. This data also provided the link in terms of integrating the image analysis results with the FMS data and in so doing enabled a comparison reporting process to be devised,