Multi-temporal Analysis of the VEGETATION Data for Land Cover Assessment Monitoring in Indochina

Transcription

1 Multi-temporal Analysis of the VEGETATION Data for Land Cover Assessment Monitoring in Indochina Chandra Giri, Tin Aung Moe & Surendra Shrestha UNEP Environment Assessment Programme for Asia and the Pacific Asian Institute of Technology P.O. Box 4, Klongluang, Pathumthani 12120, Thailand & Josef Aschbacher Directorate General Joint Research Centre (DG/JRC) Space Applications Institute (SAI) I ISPRA (Va) Italy Abstract Land use/land cover changes are occurring at an unprecedented rate and scale in Indochina. Accurate and reliable data, however, have not been available in the past. The current study aims at improving this situation. SPOT VEGETATION and NOAA AVHRR data were used to assess the usefulness of the data for accurate delineation and demarcation of major land cover types in the region. Land cover maps of 1985/86 and 1992/93 were prepared using NOAA AVHRR and a land cover map of 2000 was prepared using VEGETATION data. Hot Spot areas were identified using these data sources. Detailed investigation of identified Hot Spot areas were performed using high spatial resolution satellite data such Lansat and SPOT, the main purpose of which was to identify driving forces responsible for the changes. The paper presents a synopsis of this exercise focusing on the usefulness of VEGETATION data and its comparative advantage over NOAA AVHRR data. The ultimate purpose is to integrate the use of VEGETATION data into a regular assessment and monitoring operation of land cover types in Asia. Background Increasing population growth coupled with unsustainable resource use is a major factor affecting sustainable development in Indochina. Land use/land cover change particularly that of forest 1

2 cover change has been occurring at an unprecedented rate and scale threatening all major ecosystems of the region. The underlying causes of forest degradation, forest fragmentation and deforestation are multifarious. Shifting cultivation, expansion of agricultural lands, forest fire and illegal logging are some of the prominent factors responsible for the change (Giri et al. 1999). With the rapid economic growth, land use/land cover change is occurring in the forestland, agricultural land (through intensification), wetlands, and in urban areas. The increased awareness of environmental issue and the need to strive for sustainable management of natural resources has focused attention on the need to map and monitor such changes on a regular basis. The available datasets are however, outdated and are available in incompatible formats. Production of timely and reliable information on land cover has a significant contribution towards State of the Environment reporting. Moreover, such information is needed to better understand the functioning of vast and varied ecosystem of the region. Generation of standardized and harmonized land cover information using the same data sources, similar analytical tools, and land cover classes covering the whole Indochina is an urgent need (Giri & Shrestha, 1995). Land cover, which cuts across several sectors of resource management such as agriculture, forestry, hydrology etc., plays a significant role in global and regional ecosystem assessment. Such information is vital for formulating sound management strategies at different level of management. This is in conformity with Chapter 40 of Agenda 21 that underlines the importance of improved availability and presentation of information on all aspects of environment and development for decision making towards sustainable management. With the launch of SPOT VEGETATION satellite, the possibility of such a regular and timely assessment and monitoring of land use/land cover types of the region has increased. This project was selected during the second call for investigation of the VEGETATION preparatory programme, implemented jointly by Joint Research Centre (JRC), Ispra, Italy and UNEP Environment Assessment Programme for Asia and the Pacific (UNEP/EAP.AP), Bangkok, Thailand. Experimental Objectives The broad goal of the project is to assess the usefulness of VEGETATION data for land cover assessment and monitoring, develop methodology, and to integrate VEGETATION data sets in the on-going land cover assessment and monitoring project of UNEP/EAP.AP. The followings are the detailed experimental objectives of the project. Multi-temporal analysis of the VEGETATION data for the land cover assessment and monitoring of Indochina; Identification and investigation of hot spot ; Development of methodological guidelines for use with VEGETATION data; and 2

3 Study Area Integration of VEGETATION data in the ongoing land cover assessment and monitoring project of UNEP/EAP.AP. The term Indochina in a broader sense encompasses Cambodia, Lao P.D.R., Myanmar, Thailand and Vietnam. In a narrower sense it refers only to Cambodia, Laos, and Vietnam. The present study area covers a broader part of Indochina (Fig. 1). Fig. 1. Approximate location of the study area The physiography of Indochina is highly diversified ranging from mountains to flats to delta. A series of parallel ranges of mountains, with north-south trend, occupy a large proportion of Indochina (Fig. 2). The highest mountain is in Myanmar with an elevation of 6,000 meters. Indochina experiences a moist tropical and sub-tropical climate. Population density is high except Lao P.D.R. 3

4 Methodological Approach Fig. 2. Digital elevation model of the study area NOAA AVHRR HRPT data of 1985/86 and 1992/93 covering both summer (March August) and harvest season (October February) were acquired. Similarly, VEGETATION data of harvest season was also acquired. All the datasets were resampled to 1 km and geo-referenced accepting a Root Mean Square Error (RMS) of 1 pixel (1 km). For detailed investigation of hot spot areas Landsat and SPOT data were used. Both visual and digital classification techniques were deployed. Unsupervised classifier was used followed by iterative labeling during the digital classification. The interpretation was supported by a vast amount of ancillary data already available at the GRID (Global Resource Information Database) database. Primary data collected during the field visits were also used during the interpretation (Fig. 3). 4

5 METHODOLOGY Data Secondary Field data collection Data acquisition Pre-processing Unsupervised classification Iterative labeling Land cover map Accuracy assessment GIS analysis Reporting on LCA & M GIS database NDVI analysis UNEP/EAP-AP The best scene (in terms of cloud coverage and data quality) selected from the summer season data was analyzed first. Alternative scenes were used to fill in the gaps. Information obtained from NDVI composting was incorporated in the final land cover map. The methodological approach is presented in Fig. 4. Summer Season VEGETATION DATA T1.TN Best Scene Alternative Scenes Cluster Analysis Iterative labeling Land Cover Map Harvest Season Data NDVI Compositing NDVI Categorization Fig. 4. Methodological Approach 5

6 Results and Discussion The qualitative assessment of land cover was carried out during the period of 1985/86, 1992/93 and year The land cover produced using VEGETATION data has been presented in Fig. 5. 6

7 Eleven major land cover types were delineated using both NOAA and VEGETATION data viz: evergreen forest, deciduous forest, mixed forest, mangrove forest, croplands, shrublands, savannas, wetlands (swamps), barren lands and water bodies. Harvest season data was found to be useful in discriminating forest and non-forest areas whereas summer season data provided complimentary information of evergreen and deciduous forest. Land cover types in Indochina are found in a mosaic of forest and agricultural lands that are often less the size of the spatial resolution of the satellite data used. Because of this, forest areas in northern Laos and northeastern part of Myanmar have been overestimated. Shifting cultivated areas in many areas were not delineated accurately. VEGETATION data under such circumstances was found to be very useful that provides greater details compared to NOAA AVHRR. Land Cover Change Based on the visual interpretation of classified land cover maps of 1985/86, 1992/93 and 2000, it was observed that major causes of deforestation are: conversion of forests to agricultural lands, shifting cultivation, forest fire, and logging. Change patterns of these major land cover types are presented in Fig. 6. Closed forest Open forest Agricultural land Barren land Shrub land Swamp forest Fig. 6. Land Cover Change Pattern 7

8 In Cambodia, the magnitude of land cover change is not very significant. However, in the northeastern part of the country, moist mixed deciduous forest was found to be converted into mosaic of forest and scrubland. Similarly, increasing signs of upland cultivation and shifting cultivation were also noticed. In Lao P.D.R., forest degradation was visible primarily because of the increased pressure from the shifting cultivators. With such coarse resolution data, it was not possible to ascertain the effects of on-going logging practices in both countries. Shrinking of forest coverage in the Northern Region and Central Plateau and rapid reduction of marshland to give way for agriculture throughout the country were observed in Vietnam. The land cover change of Myanmar could be attributed to degradation of evergreen forest into a lesser form of vegetation (i.e., scrubland) along the hills. Errors associated with geometric and mis-classification of the VEGETATION and AVHRR data affects the overall accuracy of the change analysis. Unavailability of cloud-free pixels, resulted data gaps causing problems in comparing time series data. In order to arrive at more precise monitoring results, a thorough understanding of the periodicity and environmental adaptations in relation to changing habitat condition are necessary. Incorporation of relevant thematic information through GIS manipulation is also indispensable on that effect. Hotspot Areas Investigation With the analysis of (NOAA) and (NOAA) and 2000 (VEGETATION) data three "hot spot" areas were identified: Oudomxay province in Northern Laos, Mekong Delta in Southern Vietnam and Loei province in Northern Thailand. The prime objective of this exercise was to investigate in detail the pattern and magnitude of land use/land cover change and identify the driving forces responsible the changes. The output of this exercise intends to serve as an early warning system for preventive measures in such areas exhibiting major land transformation or active deforestation. Major Findings of Hot Spot Areas analysis of Oudomxay, Lao P.D.R. An approximate area of 60 x 60 km was investigated using a high spatial resolution SPOT data. Various stages of shifting cultivation practices from land preparation (burnt and exposed soil) up to fallow fields covered by thick bushes were identified both in the satellite image and on the ground. Shifting cultivation plot size ranges from 2 to 650 ha were distributed in the watershed of two major river systems, Nam Tha and Nam Beng. Proximity to these rivers and a road along the Nam Beng that connects Xai and Pak Beng districts provided a strong market pull for small and large villages to be attached with these natural features and available infrastructure in exchanging their goods and services. Satellite image interpretation approximates that 42% of study area is subjected to human exploitation under various form of agricultural practices from lowland to upland, and the 8

9 remaining 32% are still covered by evergreen forest. Paddy cultivation is common along the rivers. Combining the location of the village with GIS database suggests that their influence on their surrounding ecosystem in utilizing and looking for new agricultural plots creeps far beyond their backyard. The presence of plots where soil is exposed or recently burnt site shows that human activity is still in effect despite the absence or remote distances of their villages with respect to that area. The rolling topography in a middle elevation gradient harbors the large proportion of remaining forest in the watershed where upland cultivators showed their strong preference to satisfy the existing demand for available and fertile grounds. The intensity of occupation of some areas, and the considerable size of the agricultural plots, indicated a high pressure on the land and low yields per hectare. This combination most certainly will cause a long-term decline in the productive potential of the uplands for both agriculture and forestry. Major Findings of Hot Spot Areas analysis of Mekong Delta, Vietnam. Preliminary investigation of coarse spatial resolution satellite data indicated that land use/land cover changes are occurring in the delta both across time and space. These changes are both temporal and seasonal. Knowing this, an area of 60 km x 60 km was investigated in detail using high spatial resolution SPOT multi-spectral satellite data. The delta is characterized by one of the most productive rice growing area coupled with highest population density of Vietnam. This is also one of the major economic development regions of the country. Much of the forest areas have already destroyed during the famous Vietnamese war and economic reconciliation after the war. Few scattered and degraded forests mainly of Malaleuca and mangroves are left, suffering from intense pressure from humans. Basically, two major types of land use/land cover changes are occurring in the delta, namely natural and man-made. Natural factors responsible for the change are flood and tide. Man-made factors on the other hand include, changes in agricultural practices, conversion of land, destruction of mangrove and Malaleuca forests and urban growth. Agricultural activities are changing in terms their intensities and management regimes attributed by change in crops (mainly from rice to cash crops such as, sugarcane, aquaculture, etc.) and change in cropping patterns (from one crop to multiple crops per year). With artificial reclamation, agricultural production has increased significantly but in some cases at the expense of aquatic animals and mangrove forests. Some efforts are being placed to conserve the environmental base but a holistic approach to manage complex and interwoven environmental problems in the Delta is a must. Regular assessment and monitoring of land use/land cover types in the Delta helps identifying driving forces responsible for the change that can be utilized for modeling future land use/land 9

10 cover trajectories. The multi-temporal analysis of 1 km resolution AVHRR data of Vietnam provided an indication of change in the Delta that led us for the detailed investigation using high resolution satellite data. The macro-scale assessment of the country focusing on the land cover changes render a direct information link in identifying factors responsible for the changes that can be used for decision making. Major Findings of Hot Spot Areas analysis of Loei Province, Thailand An area of 60 km x 60 km was investigated in detail using high spatial resolution SPOT multi-spectral satellite data. Landsat MSS dated December 19, 1985 was acquired and analyzed for change analysis. The Loei province is characterized by one of the most forest based resources of Thailand, consisting of 20 reserve forest areas, 5 national parks and one wildlife sanctuary. Satellite data interpretation confirmed that the dominant land use/land cover type in the province is forest that includes closed forest, degraded forest, hill forest and plantations. Shifting cultivation is a common phenomenon being practiced on the old alluvium terraces and fans for cultivating upland rice, maize, groundnut, beans and cassavas. Transplanted rice is the predominant crop in the river basin and in the valleys. Part of the cultivated areas is devoted towards cultivation of cash crops such as, soybean, sugarcane, tobacco, maize, peanuts, vegetables and fruit trees. Conversion of forestland to other land use is the main type of land use/land cover change observed in the province. The underlying causes for the forest cover change are multifarious including shifting cultivation, illegal logging, forest fire, expansion of agricultural land due to intense population pressure, etc. Agricultural activities are changing in terms of change in management regimes and change in cropping pattern (mainly from rice to cash crops such as sugarcane, cassava, rubber, etc.). Migration from other provinces to this relatively less populated province has created additional stress on the fragile natural resources of the province. The deforestation risk model map was prepared that provides areas likely to experience severe pressure due to shifting cultivation, illegal felling and forest encroachment. The model, however, does not include soil and rainfall data, two important parameters for vegetation distribution. "Scoring" and "Weighing" process was performed using filed information, which demands further refinement. Conclusion and Recommendations VEGETATION and NOAA AVHRR data were found to be very useful for assessing major land cover types in Indochina. These datasets are useful for monitoring land cover transformations at a wider scale. Eleven land cover types were delineated using both NOAA and VEGETATION data and the results obtained from them are comparable. Methodological approach designed and used for NOAA/VEGETATION analysis was found extremely useful for assessing and monitoring major land cover types in Indochina. VEGETATION data provided a grater detail in the areas having a mosaic of forest and agricultural lands. VEGETATION data will be acquired and used in the on-going land cover assessment and monitoring project of UNEP/EAP.AP. 10

11 Time series analysis of NOAA and VEGETATION data were used to identify hot spot areas. The use of high-resolution satellite data such as Landsat and SPOT were found to be of very important in investigating hot spot areas in detail. Up-to-date information on the present status of land cover variables and their transformation through time could be a valuable guide for formulating appropriate policies and effective implementation of programs for natural resource allocation and a wider aspects of its management. It is recommended to use satellite data covering a whole year to capture seasonal variation. It is also recommended to monitor the land use/land cover types of the sub-region in every 5 years using coarse spatial resolution satellite data such as NOAA and/or VEGETATION. References Giri, C. P., R.S. Ofren, D. Pradhan, E. Kratzschmar & S. Shrestha, (1998). Land Use/Land Cover Change in Southeast Asia. United Nations Environment Programme for Asia and the Pacific, Bangkok, Thailand. Giri, C.P. and S. Shrestha, (1995). Land Cover Assessment and Monitoring at UNEP/EAP-AP: A RS and GIS Approach, paper presented at the International Symposium on Vegetation Monitoring August, 1995, Tokyo. 11