GISdevelopment.net ---> AARS ---> ACRS 1990 ---> Poster Session

An application of Remote Sensing techniques in forestry Hazard Monitoring and Damage Estimation

Li Liuyu, Kou Wenzheng, Cao Xufeng, Long Jing
Academy of orest Inventory and Planning Ministry of Forestry,
Beijing, China


Abstract
This article describes the techniques and experiences of using Meterological Satellite, Thematic Mapper (TM) and Spot data to monitor forest fire and assess its damages caused by fire and wind hazards; lists advantages and disadvantages in using the data and technical fields to be solved, and explores the developing trends of such methods; presents actual application status and results by examples of the destructive forest fire occurred in Daxinganling and wind hazard in Changbai Mountain in 1987 and 1986 separately in China, and clearly indicates the possibilities of using such information and methods.

Introduction
It is known that forests are important biological resources with multiple benefits in econo ics, society and ecology, and play a key role in keeping the global ecological balance. It is our fundamental policy to reserve, develop and use our forest resources in China, Afforestation Has been actively carried out in large scale and much has been done in preserving existing forest resources, which have contributed greatly to keeping the global ecological balances. However, due to the long growing circle, forests are constantly attacked by diferent kinds of hazards under the effects of natural stresses and human activities. In order to effectively protect existing forest resources, active and effective methods must be adopted to monitor the occurrences and spreading of various forest hazards and assess te damages accordingly so as to timely control its occurrences and development, to summarize the experiences and the minimize the losses. Over a decade, based on our traditional hazard monitoring method, we have applied remote sensing techniques to accomplish the monitoring mission.

Application of Remote Sensing techniques to Monitoring forest fire
Fires are the worst hazard for forest resources. It not only causes the losses of forest and property, but also severely destroys ecological environment. In order to implement the forest fire control policy of early suppression, we have actively developed and applied the forest fire monitoring techniques using meteorological satellite data since 1985, it had played an important role in fire fighting in the destructive forest fire in Daxinganling which happened on May 6th of 1987. Using Meterological Satellite information to monitor forest fire not only has features of large ground coverage and timely warming of fire occurrences (especially in inaccessible forest areas in which other monitoring means can not be fully established), but also obtains a whole set of data on the occurrences and expanding of forest fires, which provide accurate and timely in formation in organizing fire suppression and spreading. Meteorological Satellite data applied in forest fire monitoring has demonstrated advantages of high sensitiveness, time-effectiveness and low cost.

During the research of using meterological Satellite data to monitor forest fires, we have mainly explored the following technical areas.
  1. Time-effectiveness of Meteorological Satellite Information in Forest fire Monitoring

    Forest fire is characterized by sudden occurrence, and fire control agencies want to have warning information as early as possible. Existing NOAA satellite passes China's territory every four hours. Since the fire occurring possibility is low in cloudy and rainy weather, cloud cover can both bring a substantial impact on forest fire monitoring. Statistics made from over a hundred cases of occurrences and expanding of forest fire have showed that the time span by using Meteorological Satellite to monitor forest fire is 2-4 hours and its accuracy is a little higher than that of air patrolling, and lower than that of human patrol within the visibility on ground. Therefore it is concluded that the time-effectiveness is acceptable by using NOAA-2, which is operating, in space in forest fire monitoring. After the destructive forest fire in Daxinganling in 1987, another ten small fires happened within the region, of which four were detected by Meteorological Satellite ahead of ground methods six were ahead of air plane and two were in the same time as air plane and ground patrols.

  2. Sensitiveness of meteorological Satellite to forest fires

    Theoretically the temperature resolving power of sensors in Meteorological Satellite is 0.1k, and spatial resolution is 1100m. Temperature resolving power can totally meet the requirement of forest fire monitoring, and spatial resolution is much lower, however if the two are combined, and feature extracting techniques are adopted, the requirement can also be met Taking the destructive forest fire in Daxinganling as an example again, all burning sites that were over one hectare in the ten events were detected by Meteorological Satellite. Although burning areas were small, the effect of released heat energy were far beyond the actual burned area resulting in temperature raises in relatively large areas and were reflected in Metrological Satellite imagery. Evidently, to fully develop temperature sensitiveness of Meteorological Satellite imagery. Evident, to fully develop temperature sensitiveness of Meteorological Satellite to better forest fire monitoring, attention must be paid to the functions of middle inferred band, and adoption of proper synthesis of three bands and feature stretching method. Our research has indicated that forest fire about half hectare in size can be detected by using Meteorological Satellite.

  3. Meteorological Satellite Information Integrated with GIS to Increase Correct Interpretation and Precise Location of Forest Fires.

    Due to the displacement under the Satellite, frequently large distortion is created in observation areas, which causes difficulties to observe and correctly locate forest fires. Although the application of Meter logical Satellite is to monitor and assess forest fires, location techniques are still very important.

    In order to solve locating problem we have incorporated Geographic information system (GIS) into fire monitoring system by Meteorological Satellite data. Functions of GIS are: (a). Broadly locate and judge fire hazards areas, and provide timely information to the responsible organizations; (b) Set up geographic coordinates, identify fire occurrence site and provide basic data to fire fighting agencies; (c) mark out burned spots, increase the ratio of correct interpretation of fire sites, and to avoid misinterpretation problems that would indicate normal fire as fire hazard. Without such information, residential areas and brick kilns are mis-interpretated as fire hazards; and (d). Determine fire categories, and differentiate forest fire from wild fire in grassland.

    By applying GIS, fire-locating accuracy by Meterological Satellite is incresed from 0.1 (latitude and longitude) to plus and minus 0.5 kilometer.

  4. recording Complete Course of Fire Occurrence and Spreading Servicing researches on Forest Fire

    One of the important advantages of using Meterological Satellite to monitor forest fire is that it can record whole process from happening to expanding . It can trace back to the status before fire occurrence upon the request, which provide overall and actual information for forest fire research. For example, during the destructive forest fire in Daxinganling, we had collected all data of four passes eac day by Meterological Satellite from 1stof May to 30th of June of 1987. We conducted multiple researches and analysis by these data integrated with other information, such as decides the time of and research ofire occurrence with relation to the environmental factors, and found out the internal rule of suddenly occurred forest fire. Meterological Satellite data provide a clear evident on starting time of the destructive forest fire in Daxinganling because fire spots were found on the imagery which was caus\ght at tow o'clock in the afternoon on May 6th, and no spots were found on imagery that was caught back to May 1st; so the fire occurring time was determined without error, which facilitate the extraction of other related data (meterological data) to analyze the courses. Researches on fire stretching pattern, spreading speed, expanding trends were conducted which laid a better foundation for further understanding of fire hazard.

    Generally, application of Meterological Satellite data in forest fire monitoring has created a wide research area, and it is a potential and valuable work but it has to be further developed.
Assessment of forest disastrous losses by Remote Sensing data
Wide application of remote, sensing techniques has provided and effective means to assess in a timely fashion forest losses caused by natural disasters, and understand different kinds of disasters. Remotely sensed data used in evaluating forest disasters losses are acquired from SPOT, TM and Meterological Satellite. The types of disasters we have evaluated are forest and wind damages.
  1. Assessment of Burned Acreage by Meterological Satellite Data

    Using Meterological Satellite Data to assess burned acrease after forest fire is charattrized by high speed, low cost, but precession is low.

    After a forest fire it is an urgent need to know the burned acreage for rehabilitation project in order to work out a proposal for rescue construction. Although the traditional assessing method, such as ground survey, can provide reliable data, it has high cost and slow in speed. Using Meterological Satellite data can provide such information immediately after fire occurrence. For example, the destructive forest fire in Daxinganling was completely suppressed on June 6th. By using Meterological Satellite data we estimated the burned acreage was 1.32 million hectares with a accuracy 96% of the precise statistics.

    A simple method to use Meteorological Satellite to evaluated forest fire losses is to count pixals, which are multiplied by pixels areas. In large burned region, the statistical precision of disastrous estimation by this method is over 90% and high statistical bias may occur in small areas (under five square kilometers). A relatively precise method to calculate fire damage is to integrates Meterological Satellite data with GIS which locates fire boundaries and estimates the acreage.

    No matter what method is adopted, attention must be paid to te estimating operation which should be conducted after reduction of abnormal heat effects. A primary point is to judge burned areas by changes of grey scale in imagery because the low spatial resolution of the Meterological Satellite can easily cause an over estimation of burned areas.

  2. Estimation of Forest Fire Losses by SPOT and TM Data.

    Relatively speaking, using SPOT and TM data to estimate forest fire losses can provide accurate results; due to influence of passing circle, it has relatively low time-effectiveness. For the evaluation of fire losses in large areas, aerial survey of ground inventory has high economic benefits and time-effectiveness, but for small areas, the cost is a little high. Using Spot and TM data evaluate forest fire losses can obtain the following results: (a) Estimating burned acreage; for example, the acreage of the destructive forest fire in Daxinganling was estimated at 1.265 million hectare which is nearly the same as the result of ground survey; (b). Grading forest fire losses; due to the uneven burning, there is a significant difference in gray scale of spectral reflectance, thus the losses can be classified into 3-5 grades accordingly. In the case of Daxinganling, four grades were identified, and a volume loss in each garage was better estimated by integrating with ground data; (c). Analyzing the correlation between species, landform and loss grade, identifying fire assistant-species, stands and structure, researching on the advantages and disadvantages of different landform to the fire spreading in order to provide data for opening fire lines scientifically and setting up fighting multiple date to find out actual losses and provide trends for natural rehabilitation by species. In the course of using SPOT and TM data to estimate forest fire losses wee tried to solve following techniques; (a) Deciding optimal synthetic bands in accordance with spectral features; (b). Analyzing relationship between loss status and geographic factors in detail by integrating with GISa (c). Developing techniques for grading fire losses by comarison with forest vegetation distrubution before the fire through full use of complex tecniques of mutiple phase information; (d). Establishing better models that are similar to ground truth survey.

  3. Forest Wind Damage Assessment by TM Data

    In 1986, virgin forest of changbai mountain nature Reserve in Jilin Province was attacked by byg burricane and about tenthousant hecteres of the virgin forests were blown down and 1.85 million cubic meters of growig stock were lost. Due to inconvenient transportation, it was difficult to conduct ground survey. In order to immediately determine the wind damage, TM data was used to estimate the losses and satisfactory results were achieved, which had strongly assisted further proposals. Main subjects of using MT data to assess wind damage are to provide fast and accurate information about losses of acreage and growing stock by combining fewer ground collected information, and landform and relief and transportation status as well.

    During the damage estimation, following technical problems had been solved: (a). Extracting wind damage feature. There is a significant difference in color tone between features of vegetation in wind damaged site and that before the damage; multiple band was synthesized by segmented linear stretching processes; (b). Techniques of damaged area grading. Integrated with GIS, damaged areas were delineated and then were graded, and further were filtered by medium values: (c). Method to estimate growing stock losses of forest trees. Using fewer angle gauge plots on ground, correlation relation was established with imagery, so the growing stock losses was calculated.
Conclusion
Using remote sensing data to monitor forest fire, and to estimate forest disastrous losses is a potential, practicable and economic technical system. Although we have made certain achievement in this area, a lot of more work has to be done on future research and development.