Monitoring of tropic forest using Remote Sensing in Xishuangbanna. China

Monitoring of tropic forest using Remote Sensing in Xishuangbanna. China

Lizhixi
Laboratory of remote sensing, Southwest Forestry College, White Dragon Temple, Kunming 650224

Abstract
An attempt to the remote sensing image information on examination over tropic forest dynamics in Xishuanghanna prefecture has been under taken. there are of three parts being included.

Mapping forest vegetation in a scale of 1,200,000,. by applying multisensor remote sensing data.
Estimating the forest timber volume by apping multistage remote sensing data.
Monitoring the forest dynamci change by applying remote sensing data.

By this experiment it is verified that this method is efficient on tropic and subtropics forest monitoring

Introduction
Today there is a serious problem in tropic, subtopic forest areas that the ecological environment is getting worse. the accuracy of monitoring on forestry management must not be neglected. the method of applying remote sensing by combining aerial photography and satellite imagery study has been to be most effective. the aerial photography was first applied in China since 1854 for delineation of forest types and compilation of forest maps. since the launching of Land sat – 1 in 1972, the application of forestry survey and mapping by some of our forestry departments have been brought into efforts. in 1986, the establishment of Land sat Station being set up in Beijing, the MSS (Multi-spectral Scanner) and TM (Thematic Mapper) data have been largely acquired. At present, aerial photographs and savellite imagery are being used for forestry and land use mapping. the purpose of this study is a trail of applying satellite remote sensing data for tropic, subtropics forest dynamic monitoring. which foumed part of World

Wide Fund for Nature (Switzerland) aid programmes, while Gent State University, Belgium and Southwest Forestry College, China, are working in cooperative.

Study Area
The Xshuangbanna Dai autonomous prefecture is situated the south front of Yunnan province (figure 1), where the Southwest border of People’s Republic of China. it is looated between 210 08’ – 220 35’ North latitude and 990 ‘ – 1010 50’ East longitede. Within three counties (Jinghong, Monla, Monhai) included and five nature reserves distributed. the total area is 19690 km2.

The are about 947. of the area of mountains. 6% of valleys and basins. the altitude varies at 475-2429 m. The Xishuanghanna prefecture lies at a transitional zone between the floras of Malaysia, Himalaya and Southwest mountains of China. The result is an exceptional diversities of flora. i.e. rare species, medicinal plants, fast growing timber species, edible and oil-crops etc. while the fauna is equally rich in species. where elephant, gaur, tiger, white cheeked gibbons, peacock, sun bird, peacock pheasant etc. which are characteristic of tropic area of China.

Xishuanghannna is one those well conserved primitive forest regions, with tropic forest covering the low land, a piece of treasured area better natural conditions and, rich resources which is seldom seen in China.

Methodology
There are three items of work of Xishuanghanna tropical forest dynamic monitoring by remote sensing as follows. Forest Vegetation Mapping Using Multisensor Remote Sensing.

A purpose of the study is to examine the utility of multisensor remote sensing data to detect and map tropical endangered forest distribution. Those multisensor remote sensing material are: 1, Spot – 1 SX 16-Feb – 88, frame 260/306, 260/307, and 261/307: 2, Landsat – 5 TM 02 Feb – 88, frame 139/45: 3, Landsat – 3 MSS 03-Feb – 82, frame 139/45: 4, areial panchromatic black and white photographs Mar – 89.

The computer enhanced satellite color composites were produced form digital tapes, using a 1 S model 75 with S-600 software, linked to a micro VAX as host computer in the laboratory of remote sensing, Gent State University, Belgium. Besides, newly flown panchromatic aerial photographic at scales between 1:30,000 1-50,000 were available for most districts.

Different sensor images will provide information for different levels of classification in forest vegetation map. According to tropical forest vegetation classification system, the remote sensing imagery were interpreted, that before some field verification of remote sensing imagery were conducted. the details of interest for this study were forest vegetation type the boundaries were delineated by tracing details on to a transparency super imposed on the Spot XS or Landsat TM composites. The satellite imagery can not be used for the classification of all forest vegetation types. This is the reason that both aerial photographs and satellite images were used in this study.

After interpretation, all details of relevant forest vegetation types, from satellite images and aerial photographs were transferred to a base map (scale 1:200,000) for making the forest vegetation map. The preliminary map was completed and printed. Forest Volume Estimating Using Multistage Remote Sensing

The multistage unequal probability sampling for multistage remote sensing has been used. It is sampling with probability proportional to size (P.P.S). For this reason, we first make use of the satellite imagery and aerial photographs to interpreted percentage become the probability of each stage. We make it the basis on sample of extract the units of sample with random and unequal probabilities. According to the order of the sequence of multistage, we make the sampling step by step downward. At the last stage, forest volume is estimated. The estimation is expanded step by step upward. As to the specific, there are a few programs as follows.

1, using the Landsat image to interpret preliminary datum of parentage of forest land on primary stage unit.
2, random unequal probability extract for the sample units of primary stage.
3, using the aerial photographs to interpret preliminary datum of percentage of forest volume on secondary units.
4, random unequal probability extraction for the sample units of secondary stage.
5, extraction of the sample unite of third state.
5, extraction of the sample unite of third stage.
6, measuring the ground plots.
7, estimating the population forest volume.

The forest volume is calculated from the statistical model:

Where
V=estimated forest volume
V_ijk = measured volume of sample units of third stage.
P_i = the probability within primary stage sample units
P_ij = the probability of secondary stage sample units j within primary stage sample unit
P_ijk = the probability of third stage sample units k within secondary stage sample units j form primary stage sample units.
N,m, t_ij = the sample size of primary, secondary and third stage respectively.

Forest Dynamic Change Monitoring Using Multitemporal Remote Sensing In the forest dynamic change monitoring, mutitemporal remote sensing data are as follows

1, Aerial photographs: Feb – 65, scale 1:40,000: Jan – 81scale 1:40,000.
2, Landsat imagery , 02-Mar-73, MSS (scale 1,250, 000), 02 Feb – 88, TM (scale 1:80,000).
3, Spot imagery 16- Feb-88, ES (scale 1:50,000). According to multitemporal remote sensing imagery interpretation, the forest area at 1965, 1973, 1981 and 19888, (see Table 1) of Xishuanghanna prefecture have been calculated.

Results and Discussions
Tropical and subtropical forest vegetation types is ranged as follows.

1, Tropical rain forest

(1), Seasonal rain forest
(2) Mountain rain forest

2, Tropical monsoon forest

(1) Simi –evergreen monsoon forest
(2) Deciduous monsoon forest
(3) Limestone hill monsoon forest

3, Southern subtropical evergreen broad-leaf forest
4, Deciduous broad-leaf forest
5, Warn coniferous forest
6, Bamboo forest
7, Shrub forest
8, Herbosa
9, Economic forest

(1) Rubber forest
(2) Tea forest

10, Agricultural land

(1) Water rice field
(2) Mountain rice field

At the same time, we made a map of the forest vegetation of Xishuanghanna, (scale 1:200,000), by applying multisensor remote sensing imagery.

Besides, we have tried the use of image information of multistage remote sensing, by sampling of the multistage unequal probabilities (samping with P.P.S.), for the forest volume estimating. By the interpretating and calculation, the forest, volume of Xishuanghanna is 70,731,400 M when reliability in 85%.

The advantage of this method lie in the fact that the aerial spatial remote sensing information are combined together, and so is the remote sensing and sampling. Thus it fully develops the potential abilities of remote sensing information. By this experiment, which is verified that this method is suitable to the forest resources monitoring of large areas.

Results form this study also showed that about 33.72% of the total area of Xishuanghanna prefecture is covered by many types of tropical and subtropical forest. But, according to the interpretation of sampling of aerial photographs of 1965, the forest area were 46.46% of the whole prefecture. The comparison of the forest area from photo interpretation data (flown in 1985) with satellite imagery (taken in 1988) showed that within the past 23 years has been degenerated as shown in Table 2 But since 1981 it was provisionally concluded that degeneration of forts area seems to have been returned.

Conclusion
Remote sensing techniques is now being widely applied to identify, analyze and monitor forest vegetation. By means of Spot XS and Landsat TM imagery researcher can get more information in a short period, at a lowest cost and in highest precision.

Future studies will be including an attempt for wide life habitat monitoring and endangered forest management.

Xishuanghanna has varied landscapes with multiple species of plants and animals, scientific research and tourism which are of tremendous economical value. To protect such resources of rare plants and animals, we are building the natural reserves of Xishuanghanna. In order to strengthen the construction of Xishanghanna and natural reserves, arduous efforts need to be made for a long tome.

Acknowledgements
This paper presents works not only there are of the international scientific cooperation, but also involved the study of author previous work. All studr was funded by World Wide Fund for Nature, and the Forestry Ministry of China. In drafting the plan of study, we have got valuable help from the Forestry Ministry of China, the Forestry Department of Yunnan Provine, and Dr. John R. mackinnon, Dr. Christopher Elliott, Mrs. Pasoale Moehrle of the World Wide Fund for Nature, and Prof. Yang Yuan Chang of Southwest Forest College. The assistance of Prof. Roland E. Goossens. Mr. Robert R.De Wulf. Mr.Bruno verbist from Gent University, and the graduate students Yang Chunjian, Fan Jainrong, Zhou Rulian from Southwest Forestry College, provided company and help. And the paper has Chang. I am very grateful.

Reference

De Wulf. R., Goossens, R., Gerard, F., Vebist, B., De Rovjer, B., Borry, F. and Li Zhixi, 1990. Forest type classification for degradation assessment in he Mengyang nature reserve. P.R., of China. Symposium of international photographic and remote sensing (Canada). Heller, R.C., 1985. Remote sensing with multispectral, multitemporal, and multistage sampling. Reports on training course on application of new remote sensing techniques to forest resources surveys. Institute of forest inventory and planning, the P.R. of China. 180-187 P.

Li Zhixi et al, 1982. Forest area monitoring in Xishuangbanna. Forest inventory and planning, Central Southwestn of China. No: 1,17-19 P. (in Chinese).

Li Zhixi et al, 1985. A preliminary study on the earth’s surface landscape classification by computer in Xishuanghanna. ACTA phytoecologica et geobotanica. Vol 9: no. 3,231-234.

Li Zhixi et al 1986. Forest resources inventory with multistage remote sensing in Xishuanghanna. Proceeding in Beijing international sympcsium on remote sensing. 162-166.

Langley, P.G., Aldrich, R.C. and Heller R.C., 1969. Multistage sampling of forst resources by using space photography. Proceeding 2nd Ann. Earth resources aircraft program status review. NASA-JSC, Houston. TX. 19. 1-19.21.

Peorasukdi Adisonprasert. 1985. Forest inventory in Thailand using remote sensing techniques. Inventorying and monitoring endangered forests. IUFRO Conference. 109-133.

Singh K.D., 1985 Review of recent FAO contribution to inventorying and monitoring endangered forests. IUFRO Conference. 55-59.

Sader A.S., Waide R.B. et al., 1989. Tropical forest iomas and successional age class relationships to a vegetation index derived from Landsat TM data. Remote sensing of Environment. (U.S). Vol: 28. nol 1-3 143-156.

Xu Yongchun, Jaing Hangqao, Yang Yuanchang. Guo Yingqun, et a., 1985. Investigation report of Xishuanghanna nature reserve. Yunnan Press. (in Chinese).

Table 1 Forested Area in 1965-1973, 1981, 1988

Table 2 Forested Area Changed During 1965-1988