Background: Senegal is a coastal country lying on the westernmost point of Africa. The country has a surface area of approximately 197,000 km2 and is covered by vegetation of variable density increasing from North to South from arid steppe to forest with transitional savannas in between. Every year, much of Senegal is affected by bush fires, which have a considerable impact on the development of the vegetation. The burning season starts in November at the end of the rains and extends until May. The timing of fires coincides with the Harmattan winds, which carry hot, dry continental air from the Sahara desert and which dry out the herbaceous vegetation layer, thereby supplying a ready source of fuel.
The monitoring of fires by remote sensing has been carried out by the Centre de Suivi Ecologique (CSE) since 1990. The principal aim of this activity is to assist the Direction des Eaux, Forêts, Chasse et Conservation (DEFCCS) of the Senegalese government with the identification and management of bush fires.
Data: CSE carries out satellite monitoring of bush fires using NOAA-AVHRR imagery. This is received on average twice daily by the CSE’s receiving station, installed in 1992. Prior to 1992, images were obtained from the AVHRR receiving station at Maspalomas (Canary Islands) with an average delay of two weeks from acquisition to delivery at CSE. This delay made CSE’s job of real time fire monitoring virtually impossible, and thus until recently CSE was unable to alert government authorities concerned with controlling and extinguishing active fires.
The raw images received at CSE are geometrically corrected and mapped to a UTM projection. The correction modules, which apply orbital parameters obtained from the satellite, are integrated in the mapping routines of CHIPS, an image processing software package developed by the University of Copenhagen in collaboration with CSE.
Classification: Classification is undertaken by applying a threshold to channel 3 (thermal infrared) on night images to identify active fires. The threshold capitalizes on the thermal contrast between active fires and unburned or recently burned surfaces. In general, the greatly decreased incidence of solar radiation at night time minimizes the risk of confusing grey body surfaces (that is, surfaces which behave almost like black body emitters) with burned areas having spectral characteristics similar to that of laterites.
Daytime images are then utilised to estimate the burned surface area. These images are converted to false colour composites with channels 1 (visible), 2 (near infrared), and 3 (thermal infrared) assigned to the red, blue and green image plans respectively. These false colour images allow on-screen identification of burned areas, which are then sampled for their radiometric characteristics. The image samples, or training sites, are used in a supervised classification employing the Box class method and the surface area is subsequently calculated by counting pixels classified as burned in the resultant images.
Verification: CSE regularly carries out field verifications of these classifications. This is necessary because of the difficulties inherent in mapping burn scars in areas where tree cover often prohibits direct view of burned surfaces from space. Verification is done with the assistance of Global Positioning System (GPS) receivers, which permit the exact location of burned areas to be identified and related to the satellite images mapped in UTM. In addition aerial statistics are derived after a correction factor of 0.8 has been applied to the original estimates. This correction factor takes account of the coarse resolution (1.1 km at nadir view angles) of the AVHRR sensor and the fact that the AVHRR tends to over estimate (by about 20%) the amount of burned area.
Operational Monitoring: Bush fire classifications are done almost daily during the burning season. This results in the production of daily maps in A4 format, with exact positions of fires given by administrative units (arrondissements), which are then transmitted directly to DEFCCS. In addition, monthly and annual maps are produced that reveal the full extent of burning over these longer time periods. These images are incorporated into Geographic Information System (GIS) software (usually either Idrisi or Arc/INFO) to facilitate further cartographic enhancements and analysis of results.
Summary of Results in 1993/94
The results of the analysis of 1993/94 images reveal that approximately 750,000 ha were burned and 1,677 fires were identified among all the arrondissements of Senegal. Bush fires were most extensive in terms of burned area during the three month period of December-February, with a pronounced peak during February. During the next trimester (March-May) the number and the extent of fires diminished probably due to the low herbaceous cover at the end of the dry season. However, these late dry season statistics may include certain extreme fire events which can result from ignition of highly inflammable herbaceous material.
Fig.1. Wildfire statistics 1993-94 (will be added later)
Fig.2. Burn scar map of Senegal for the 1993-94 fire season derived from NOAA AVHRR imageries.
The annual fire map on the previous page shows the predominance of bush fires in the southern part of the country, where fires were highly dispersed yet prevalent. In this zone, vegetation production, as measured by CSE’s vegetation monitoring unit, reaches an annual rate of approximately 4,000 kg of dry matter per hectare. These areas tend to be dominated by Combretum spp., which are less extensive toward the south-west, where the more humid sub-Guinean climate regime along with well-conserved dense forests probably retard fire development.
The northern portion of the country had fewer instances of bush fires, although the areas burned tended to be more continuous. This region is part of the Sahelo-Sudanian zone, characterised by high production of grasses such as Eragrostis tremulata and Schoenefeldia gracilis during wet years. In these areas, fires spread extensively across the continuous herbaceous layer, where there are generally few fire breaks. In such zones, the surface area of burns sometimes exceeded 50,000 hectares. Moreover, the passage of fire is considered an adverse impact on the so-called “sylvo-pastoral” zones, where grazing is the primary occupation and land use. In particular, fire lessens the pasturage and subsequently leads to longer distance movements of herds and humans in search of forage, a system which is known widely as transhumance.
The cluster of isolated fires at the extreme north of the map corresponds to the burning of sugar cane fields, conducted by the Senegalese Sugar Company to stimulate further production. These are the so-called “feux de la canne”.
Application of the Results
Intervention Strategies: The bush fire monitoring is carried out by CSE as a service to the DEFCCS to help them to better manage bush fires. Products are supplied in accordance with protocols agreed between CSE and DEFCCS, which ensure timely delivery and certain standards of cartographic products and statistical indicators. However the means for dealing with fires on the ground after information has been transmitted by CSE are still somewhat limited. This serious handicap has led in the past few years to a change in strategy concerning the fighting of bush fires, which now emphasizes raising the awareness of local populations. Numerous village fire committees have been created across Senegal and have been given light equipment to limit and stop the spread of bush fires.
In this context, the cartographic information remains extremely important given its role in planning interventions and impact assessment in affected zones. Overall, it serves as a means for taking decisions towards better coordinated actions at national level. Moreover, the use of remote sensing for bush fire detection has reversed traditional monitoring schemes by enabling information to flow from CSE to the field.
Sub-regional Cooperation: The Republic of Guinea contains within its territory the upper basins of the Senegal and Gambia Rivers as well as a portion of the Niger River. To the extent that fire threatens the ecological balance of the basin areas, these basins and the role of fire in them are important in the regional context.
With the support of the Swiss development cooperation agency (Cooperation Suisse), CSE is currently leading a trial programme with the Direction des Forêts et Faune of Guinea aimed at raising awareness of the problems posed by bush fires. Information on fire dynamics gathered by CSE’s AVHRR fire monitoring activity is transmitted by radio in rural areas in order to alert local populations to the urgency of the problem and the need to control it. It is expected that the impact of these radio campaigns in certain test areas will determine whether the approach is feasible elsewhere.
Global Change: The results of the satellite monitoring of bush fires in Senegal and Gambia during the 1991/92 season have served as a base for calculating the contribution of bush fires to the greenhouse effect. The determination of burned biomass has been made by overlaying fire maps with those of the biomass production produced by CSE, which totals the amount of vegetation biomass available from the previous growing season. Copies of this study are available from the United Nations Environment Programme (Inventaire des sources d’émission des gaz à effet de serre au Sénégal. Ministère de l’Environnement et de la Protection de la Nature. 1994).
The importance of bush fires in the process of global change merits the establishment of sub-regional cooperative programmes involving institutions that monitor and measure the phenomenon of bush fires. CSE already has extensive experience in monitoring of bush fires and vegetation production and could bring its expertise to bear on such an initiative, which will ultimately reinforce programmes directed at the global level.
From: Racine Kane Address: Bushfire Monitoring Unit Centre de Suivi Ecologique (CSE) B.P. 15352 Dakar, SENEGAL