European Tropical Forest Research Network

Experiences in the elaboration, implementation and follow-up of forest management plans in natural tropical forests using computers, computer software and other technological packages: a Case Study on Borneo.

By Froylan Castaneda & Christel Palmberg-Lerche

This is a summary of a Paper based on the work of Michael Kleine, with contributions from Evelyn Jugi, Yosep Ruslim, Robert C. Ong, Albert Radin, Bernd Hahn-Schilling and Alexander Hinrichs. August 2001. Forest Management Working Papers, Working Paper 2, Forest Resources Development Service, Forest Resources Division. FAO, Rome (unpublished).

This case study summarises the experiences in the use of computers, computer software and other technological packages in the planning, implementation and monitoring of forest management in natural tropical forests. The methods and procedures presented here have been tested and are currently being used in three projects in the Indonesian and Malaysian territories on the Island of Borneo, Asia. The projects located in East Kalimantan, Sabah and Sarawak are collaborative efforts between the local forest authorities, the private sector and the Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ), the German Agency for Technical Co-operation. They aim to develop management systems suitable for the sustainable utilisation of tropical forests within the local socio-economic context.

The report is organised according to the basic management concept for sustainable tropical forest management as pursued in all three projects.

In the first section the reader is introduced to the forests and forestry sectors of East Kalimantan, Sabah and Sarawak. Brief descriptions are provided on the ecological and socio-economic environment, the current condition of the forest resources and the institutional framework under which forest management takes place. This section also outlines the basic management concept pursued in all three projects. This approach is characterised by distinct levels and phases of management ranging from planning, implementation and monitoring to control. In order to better understand the function and use of the various computer-supported methods and procedures of forest management, they have been organised according to this structured management concept.

The second section presents three key components of strategic planning in forest management, i.e. forest zoning or land use planning, forest inventory and yield regulation. The various aspects of land use planning are processed using a comprehensive process of data collection, compilation with computer software and spatial representation with the help of GIS technologies. The inventory data processing and yield regulation procedure make use of specially designed program packages. Their applications are explained in detail.

Operational planning at the compartment level is the component within the forest management system dealt with in the third section. A computer-assisted method to design improved topographic and tree location maps is presented. These are important for reduced-impact harvesting operations. In addition, this section also contains a description of a silvicultural decision-support system that is based on silvicultural diagnostic survey. This survey applies aerial photographs to pre-stratify the forest area followed by a more detailed field survey.

In the fourth section two monitoring and control systems that are fundamental to sustainable forest management are described. Both systems are complementary and build on each other, in terms of data procurement and computer-supported processing. The so-called compartment register provides the most detailed information on forest conditions, input and output at the lowest level of forest management. A comprehensive cost and resource accounting system uses the databases generated at the compartment level. Through a structured process the information is compiled and aggregated allowing an overall financial analysis and profit calculation for the entire FMU.

For further information please contact:
Froylan Castaneda, Forestry Officer (Tropical Forest Management) or Christel Palmberg-Lerche, Chief
Forest Resources Development Service, Forest Resources Division, Forestry Department FAO
Viale delle Terme di Caracalla, I-00100 Rome (Italy)
Christel.palmberglerche@fao.org

Emerging Technologies for Forest Resources Appraisals and Land Use Planning

By K. D. Singh

Introduction
This paper presents a short account of recent technological developments such as GPS (geo-positioning systems), GIS (geographic information system) and Modelling, which could be used with advantage in national (forest as well as agroforest) inventories. These technologies, in no way, obviate the importance of good field work or good interpretation of images. Their contribution consists of reducing the cost and/or time of inventories and enhancing the analytical capacity, undoubtedly a contribution of great importance.

Geo-positioning systems
The GPS is a satellite-based navigation system, which can be used for determining the coordinates (x-y-z) of any observation point. It is accurate within a few metres and enables one to map points, lines and areas quickly and easily. The accuracy depends on the equipment, field conditions (in particular, canopy cover), as well as the procedure used. Combined with survey maps such systems could replace the traditional method of locating field plots with maps, chain and compass. An added benefit is that operations such as clearing paths and bypassing obstacles, are no longer required. In the past these operations significantly added to the cost of forest inventory. Through GPS, the plot location becomes near-foolproof, fast and in-expensive.

The trend is toward cheaper, more flexible GPS. This generally means cellular phone-size equipment. Hand-held models allow one to collect and maintain spatial data for GIS use which is an ideal solution for mapping and managing spatial data in many applications, including natural resource mapping and environmental studies. One may also record customized attributes of information. These field data can then be incorporated seamlessly into GIS.

Geographic information system
The term GIS is currently used in two different senses. In the narrow sense, it refers to computer hardware and software for handling geographic information. In the broader sense, it includes all GIS equipment and operations, including the hard and software, input data, editing and storing, processing of data and reporting (see figure 1). The value of GIS to deal with complex problems such as land use and environmental planning, resources management, integrated area development, etc., has been known for a long time. However, recent advances in computer technology such as data base management systems (DBMS), computer- assisted drafting and mapping (CAD/CAM) and modelling techniques have made GIS a very powerful and at the same time affordable tool even at the individual level. Using a relatively inexpensive PC, it is now possible to input, integrate and store information in the form of data as well as maps; analyse and manipulate data and combine these with complex modelling algorithms and display results in the form of computer graphs, maps or tables. Together with the word processing capability of computers, the final report can be produced directly.

Experience indicates that input data for GIS has to be very well prepared if cost and time associated with a GIS project is to be minimized. In fact, to get the best out of a GIS, a total planning is required, starting with survey and analysis of user's output requirements, followed by input design, reliable and relevant data collection and modelling.

Figure 1

Modelling
Models are an abstraction of reality in a mathematical statistical language with a view to study the interaction among various components or dynamic behaviour of the entity or the process of change under alternative assumptions. Models need not include all associated observations or measurements, but a subset of them which are relevant to the problem. A good model should, in fact, exclude unrelated details and focus on fundamental aspects of the reality under study.

Because models are used in such a wide variety of contexts, it is difficult to define even broad types of usage without ambiguity. One major division is between descriptive and normative (or predictive models). The former is concerned with some description of the real world, such as a scale model, a map, a series of equations, and some other analogue; the latter, an ideal type, is what might be expected to occur in the real world under certain conditions. Models have also been classified as static (steady state) or dynamic according to their changing nature; deterministic and stochastic according to degree of probability associated with their prediction; or holistic and reductionist according to the level of detail included in the model.

In complex modelling work, a start is made with establishment and testing of descriptive models to determine causal relationships between various variables. Once causal relationships are known, it is possible to develop a hypothesis, theory, or model, which can predict outcomes given the same or similar circumstances in another area. Well-established relationships can make for very powerful normative models essential for pro-active land management.

An example of modelling: formulation of alternative wood-energy policies
This section presents a model with the acronym APM (Area Production Model), which was originally developed by Nilsson and the present author for a World Bank investment study in India back in the seventies. Since then it has undergone a long period of improvement (see Nilsson 2000). The model simulates possible developments of land use and primary area production. It can be used in connection with production and consumption studies at the level of civil district, commune or a development block to demonstrate impact of alternative developments in the future land use. If possible, it should be run based on the experience of the current and past land use.

The use of the model will be illustrated here to analyse the impact of alternative policy decisions on fuelwood production and consumption in a district of India viz. Adilabad in Andhra Pradesh. Four scenarios are presented using the available forest resources and socio-economic data. Effort was made to make the model as realistic as possible. The alternative scenarios and underlying assumptions are given in Table 1 and results in Figure 2 in a graphic form.

Table 1: Assumptions made in the model regarding future state of the district

Figure 2

The following conclusions could be drawn from the model study:

• even in a district like Adilabad, with a forest cover of 40% today, a long term program is essential to balance the supply with the demand. In the best case scenario, it will take about 25 years to balance supply and demand.
• Both increasing production and controlling consumption are necessary. This suggests the need for an integrated development planning.
• The program for forest plantations must be high yielding to achieve self-reliance in the shortest possible time.
• The establishment and development of plantations must be secured with people participation with control of grazing and illicit removals.
• Adequate finance on a continuing and long-term basis must be secured. This will call for an effective, creditworthy and efficient management environment.

References
Nilsson Nils-Erik 2000, The Area Production Model: Background and Design, (Extracted from a manuscript under preparation)
Singh K. D. 2001, Guidelines on National Inventory of Village Forests, CIFOR, Bogor, Indonesia

For further information Please contact:
K. D. Singh, Adjunct Professor
University of Massachusetts at Boston, USA
. Email: karan.singh@umb.edu

Application of remote sensing and Geographic Information Systems for the monitoring and protection of a Special Protected Area in Greece.

By Ioannis Meliadis, Alexandros Tsiontsis

During an E.E. project in Greece for Special Protected Areas, the techniques of remote sensing and Geographical Information Systems (G.I.S.) have been used for monitoring and protection purposes. The primary objective for the Special Protected Area (SPA) was to reduce pressure on natural habitats while supporting sustainable development. The success of such activities depends to a large extent upon knowledge of both the ecological and socioeconomic environments.

The main objectives were to:

• Develop image processing techniques to identify land cover changes using multi-temporal satellite imagery
• Develop a G.I.S. data bank which will be the baseline for the monitoring project
• Incorporate remote sensing and G.I.S. data for management and protection of the study area
• Identify the environmental parameters that are critical for the area.

The study area was the Mountain Antichasia - Meteora found in the central part of Greece. The size is almost 826.127 ha and it belongs to the Mediterranean zone. The SPA includes a mosaic of different vegetation types of a Sub-Mediterranean character. Most of the area is covered by Quercetalia pubescentis and only a small part of the higher altitudes is covered by vegetation of the beech zone. The area has a poor economic development; the main sources of livelihoods are agriculture, animal husbandry and forestry. This zone has been characterised as an Important Bird Area (IBA) according to the EC Instruction 79/409/EC and it is considered very important for the coherence of the Special Protection Network.

For the study area the following data sources were used:

• Two LANDSAT TM images dated 28/7/1989 and 1999
• Aerial photographs at a scale of1:20.000
• Management plan maps at a scale of 1:20.000
• Topographic map at a scale of1:5.000
• Land Resource Map at a scale of 1:50.000.

For the analysis two multi-temporal images were used. The methodology adapted for this study involved the classification of the LANDSAT data for both dates into land cover classes. A supervised classification using all available bands was performed on a portion of a 30-m resolution TM image. With the most recent TM data available, a map of land cover classes was produced. This map contains: Forests, Partially Forested Areas, Agricultural Lands, Rangelands, Shrublands and Urban Areas. These six classes are sufficient to accurately capture biomass of the area and estimate the habitats for the wild fauna.

In order to obtain land use change data, the digital image of 1989 was used. The photo interpretation of the aerial photos of 1987 was used for the delineation of the land cover classes which were used as reference in the classification of the '89 LANDSAT image. The same digital processing procedure was used for the older image. The change detection analysis was conducted by subtracting the two classified images and developing a cross – tabulation operation.

Comparison between the images indicated that the areas occupied by forests, partially forested land and shrublands decreased during the ten year period, while the remaining classes showed an increase. A new map was produced, indicating the areas where land cover had changed.

The new map was overlaid with G.I.S. layers. Showing the contour lines, point elevations, road and river networks, habitat, land use, soil, geology, land ownership, watersheds, game refuges, culture, and protected areas. New information was derived by the processing of the above layers, such as the aspects, slopes and the Digital Elevation Model.

The data bank of the G.I.S. was used for spatial research on the parameters that most influence the area. For example, land ownership layers and access were determined to evaluate the necessity of conducting wild fauna surveys on non-public lands which would require access permits and authorization from individual land owners. Buffer zones around the protected areas served as restriction zones for any annoyance of the wild fauna. The "habitat polygons" were also used as an input for the imagery-based analysis of potential habitat. Individual positions of birds and expert delineation of historic habitat polygons served as the primary sampling unit for the purposes of habitat prediction. In this analysis, the temporal dynamics of wild fauna areas were evaluated as a function of the temporal change in vegetation spectral signatures obtained from satellite imagery. Changes in vegetation communities are related to features of the landscape such as vegetation phenology and condition.

The main conclusions can be summarised as follows:

• Effective Monitoring requires techniques that have an acceptable cost and provide information at the required level of accuracy.
• Digital remote sensing data are very flexible and show potential for monitoring programs.
• The combined use of multi-date satellite imagery and the digital ancillary data sets of G.I.S. has proved a useful technique for producing the baseline for monitoring projects.
• Multiple G.I.S. layers can easily be used for a more detailed spatial analysis of the ecosystems and isolation of environmental parameters that influence the monitoring project.

Proper use and monitoring of our land and environmental resources for quality of life and sustainable growth requires that timely, accurate data on land cover and land use be available continually. The complex questions being addressed internationally require that researchers take advantage of new technologies including remote sensing, Geographic Information Systems (G.I.S.) and Environmental Information Management Systems that may lead to simulation models for land management decision-making processes.

Dr. Ioannis Meliadis
NAGREF – Forest Research Institute, Lab of Remote Sensing & G.I.S.
Vassilika, Thessaloniki, 570 06 Greece
Meliadis@fri.gr

Dr. Alexandros Tsiontsis
NAGREF – Forest Research Institute, Lab of Forest Soils
Vassilika, Thessaloniki, 570 06 Greece
tsiontsi@fri.gr

Weather Independent Tropical Forest Monitoring in Indonesia

By Vincent Schut and Arjen Vrielink

This article demonstrates some of the current possibilities of Remote Sensing (RS) techniques for weather independent monitoring of tropical forests on different scales, both temporal and spatial. Traditional forest monitoring and inventory techniques in the field suffer from three major problems: relatively small areas; no replicability / timeliness; relatively high costs. In other words: it is really expensive to have a fieldwork campaign for a large area every two months. Research has shown that satellite and airborne RS techniques can overcome these problems. However, one major constraint of optical techniques is that their observations are hindered by cloud cover and haze and are limited to observations by daylight.

Contrary to optical systems, radar is characterised by the possibility of unhindered observation during cloud cover, smoke (forest fires) or during the night. Therefore radar is the ideal instrument for forest monitoring, especially in the humid tropics.

An important parameter of radar systems is the frequency band. The most common bands are: C-band (5.3 GHz), L-band (1.2 GHz) and P-band (440 MHz). C-band has a relatively high frequency for earth observation and is characterised by low penetration in vegetation layers. At lower frequencies the penetration is much stronger and consequently also characteristics of the terrain underneath closed vegetation can be made visible, eg (forest-) inundation, logs or biomass.

There are many kinds of radar systems. By addition of extra antennas modern radar systems can do a lot more than just imaging the earth. The so called interferometric radar can make three dimensional observations. Polarimetric radar uses different polarisations through which it can observe additional structural characteristics of the terrain and vegetation.

SarVision is currently implementing several monitoring systems in Indonesia. In this article, we focus is on two of these systems. First a nationwide optical monitoring system is discussed: SPOT vegetation mapping. Second a regional radar monitoring system is discussed: monitoring the Mawas Reserve.

SarVision has developed a new system for large scale weather independent forest monitoring. Using SPOT Vegetation data, this system gives a fast overview of the remaining forest resources of a large area, in a resolution of 1 x 1 km. An algorithm has been developed to aggregate the raw data into a cloudless, contrast-enhanced and classifiable product. The resulting forest and forest change maps can be used for ecological monitoring, (forestry) law enforcement, forest fire damage investigation, etc. They are ideal for a multisensor approach, pointing towards deforestation hotspots that can then be investigated in more detail using spaceborne or airborne radar, or other (future) sensors.

The system is applicable worldwide and is perfect for extensive and long term forest monitoring. Due to the processing algorithm cloud and haze influences are effectively eliminated. Patches of recent deforestation can easily be found and then investigated by other sensors, e.g. MODIS (250 m. resolution), spaceborne radar or even airborne radar for a very high resolution investigation. Our straightforward and automated processing algorithms and the guarantee of future availability of SPOT Vegetation data ensure a previously unknown stability and accuracy in large scale forest monitoring.

Large forest areas in Indonesia are under severe threat because of illegal logging, poaching and increased susceptibility to fire due to recent forest degradation. One of the areas of major interest is a 273,000 ha area in Central Kalimantan, east of Palangkaraya: the Mawas reserve. This area still comprises substantial areas of pristine lowland forest and the largest remaining orang-utan population (3,000 individuals). Under co-ordination of the government, local authorities and the Balikpapan Orang-utan Survival Foundation (BOS) a three-component program is executed. These components are: (1) fast detection of illegal logging, encroachment or any other threatening conditions by remote sensing satellites, (2) a flying team to collect evidence and prosecute offenders, and (3) moving local population to buffer zones and providing them income and training.

SarVision is setting up a monitoring system based on spaceborne radar observation to deal with the first of the above mentioned components. Of course, because of cloud cover, radar remote sensing plays a major role. The objective is to set up an operational service to be able to inform local teams as fast as possible. It is planned to do this at least every two months, with a delay of less than one month.

Currently 25 m resolution ERS-2 (European Radar Satellite, C-band) data are used, along with historical data (mainly ERS-1/2 and JERS-1 (Japanese Earth Resources Satellite, L-band)). In the future the use of ENVISAT-ASAR (recently launched) and ALOS-PALSAR (L-band system) is foreseen. L-band seems superior to C-band for forest monitoring. Visibility of drainage and hydrology in these swamp forests are much better than in C-band (or in optical sensors such as Landsat TM). It is therefore very likely that L-band will contribute to refined ecological forest type mapping of this area. Within two years the service should be completely operational. This monitoring concept can then also be applied to other similar (forest) areas in the world.

Until now large scale weather independent RS monitoring is primarily discussed in relation to research projects. However, the examples above show that SarVision is already operating RS monitoring systems in Indonesia. There are Plans to extend the current monitoring areas to more forest areas in Indonesia and also to other countries with tropical forests. SarVision is currently working hard to develop more advanced monitoring techniques. The aim is to combine observations from different sensors in one monitoring system. Also high resolution (3D) airborne radar mapping systems are being developed for detailed monitoring such as individual tree mapping.

For further information please contact:
Vincent Schut, email: schut@gissrv.iend.wau.nl
Arjen Vrielink, SarVision BV
Oudlaan 37, 6708 RC Wageningen The Netherlands
Tel: +31 317 421526, Fax: +31 317 428880
Email: info@sarvision.com