Journal of Undergraduate Research
 Volume 2, Issue 8 - May 2001

Land-Cover and Land-Use Change within Sisaket, Thailand from November 1990 to March 2000

Krithi Karanth

ABSTRACT

Thailand, like many of its East Asian counterparts, has undergone rapid economic growth in the last few decades, but income-growth disparities among regions have emerged. Environmental variability is thought to be responsible in part for these regional disparities. Environmental variability can be expressed both temporally and spatially by measuring changes of land cover over time, which can be an indicator of either efficient or poorly managed natural resource uses depending on the nature of the changes. This study examines the changes in land-cover in the province of Sisaket in the northeastern region of Thailand, as defined by both conversions from one land-cover class to another, and changes of the Normalized Difference Vegetation Index calculated from Landsat 5 and Landsat 7 data. Sisaket is one of the poorest regions in Thailand, had among the lowest household income growth from 1975 to 1997, and has undergone significant land conversion, most of which were conversions from forests to agriculture, degradation of the southern-forested areas, and mean NDVI decreased from 0.25 in November 1990 to -0.10 in March 2000. I hypothesize that these changes will be correlated with economic factors such as such as access to credit unions, land area owned, amount of money spent on fertilizer and agricultural indices like crops grown, seasons and types of fertilizer used. There might also be correlations with rainfall patterns and climatic phenomena like El Nino.

INTRODUCTION

Thailand covers a total land area of 321 million rai (1 rai = 0.133ha). Initially, all land belonged to the king, but in 1954 the Land Code established the legal system of land rights (Feder, 1988). Agricultural households dropped from 73.9% in 1960 to 62.9% in 1970 and to 55.6% in 1980 as a result of more people following non-agricultural pursuits (Wyatt, 1984). In 1960, the economy was heavily biased towards agriculture, forestry, and mining. In the 1960's the government instituted economic policies that supported and promoted industrial development. The 1970's saw a shift towards a more export-oriented industry that resulted in a booming, highly diversified manufacturing sector. Nonetheless, agriculture continues to be a major economic sector, and Thailand is one of the few steady rice exporting countries in the world. Rice yields have been greatly improved with the onset of the Green Revolution. At the same time farmers continued to diversify their crops and alternatives such as tapioca became a major export in the late 1970's (Wyatt, 1984).

In spite of diversifying its economy, in the mid 1970's Thailand was caught in a worldwide slump in prices of primary commodities as well as manufacturing goods (Wyatt, 1984). Inflation occurred and this resulted in a slower rate of development that affected both Thai farmers and workers. Irrigation projects and agricultural extension efforts also hastened the Green Revolution. Improved communication and development of the economic infrastructure supported crop diversification and improved marketing of agricultural surpluses (Wyatt, 1984). At the same time, the growth in rural income was not equally shared among Thailand's farmers. Rapid demographic change brought on by population growth and the exhaustion of the supply of new lands led to a decrease in the average farm size and an increase in agricultural tenancy.

Historically, as with many other developing nations, Thailand has been faced with the problem of illegal occupation and use of land by many farmers (Nicol, 1982). In the 1970's much of Thailand's population was agricultural and much of its non-agricultural population was dependent on agriculturally based activity for a livelihood. Increase in productivity was primarily achieved by expanding land areas under production. But increased pressure by growing populations is making such expansions harder and exerting pressure on the agricultural sector and the forest reserve areas in the province.

Forests nationwide have decreased from 70% in the early 1900's to 30% in 1984 of total land area, and about one-fifth of that is occupied by squatters (Feder, 1988). The Forest Reserve Act of 1964 failed to stop agricultural land uses, and squatters were granted temporary and restricted cultivation rights in 1981. As a consequence of the restrictions, squatters continue to occupy forest reserves, and may cause land use and subsequent land cover changes in the forested regions.

Land cover changes are known to be consequences of socio-economic conditions and activities, but the specific patterns and their mechanisms have changed over time. Fortunately, an archive of satellite remote sensing data that can be used to study land-cover change over time is available, beginning in 1972 and continuing to the present day. I analyzed remotely sensed data to determine decadal-scale changes of land-cover and land-use in Sisaket, which have resulted from a combination of historic, environmental and socio-economic factors. The province of Sisaket is an appropriate study site as it represents one of the poorest regions in Thailand that has also undergone significant land conversion. I conducted both land-cover classification analyses and calculation of the normalized difference vegetation index (NDVI). Land-cover changes were also measured and mapped using NDVI (measured by arithmetic combinations of reflectances of two or more spectral bands.

DESCRIPTION OF STUDY AREA

Sisaket is in the northeastern part of Thailand along a rolling plateau and shares its southern border with Cambodia (Figure 1). The major land cover types in the province are agriculture and forests. Agricultural production takes place on land classified by the land code for agricultural purposes and sometimes on non-agricultural lands. Agricultural cultivation in Sisaket takes place predominantly by rain-fed water control. The soils in this region are known for their poor fertility (Michael W. Binford, personal communication and unpublished data). The rainy season extends from April to October and has average rainfall of about 1,300 mm/year. The major crops grown in the northeastern region, including Sisaket, are rice and cassava; the minor crops are kenaf, sugarcane, corn and bamboo (Nicol 1982).

Figure 1. Map of Thailand with Sisaket inset.

MATERIALS

The materials used for this project were scenes of Sisaket from 1990 November Landsat 5 (Path: 127, Row: 49) and 2000 March Landsat 7 (Path: 127, Rows: 49, 50). I used the software programs ERDAS Imagine 8.4, ENVI 3.2 and Arc View 3.2 for all data processing.

METHODS

The two scenes were normalized using histogram matching to eliminate atmospheric effects. I used an unsupervised classification method since the land covers were unknown to me. Land-cover classification results in a map of classes on the Earth surface with distinct patterns of spectral reflectance. The classification into five, seven, and twelve land-cover classes was conducted using the Isodata algorithm. The five and seven land-cover classes were unsatisfactory as they were unable to distinguish between spectral reflectance of open water and that of dense forests.

Calculating NDVI with Landsat 5 TM and Landsat 7 ETM data.

The Normalized Difference Vegetation Index is an index that was developed to monitor vegetation. NDVI is a quasi-continuous field that is calculated as a normalized difference between the reflectances of two biologically meaningful bands of the electromagnetic spectrum. Actively photosynthesizing leaves absorb the red wavelengths (Landsat TM band 3) as a source of energy for photosynthesis. Leaves reflect the short-wave infrared (Landsat TM band 4), so the difference between the two is proportional to the amount of photosynthesis. NDVI responds to changes in biomass and chlorophyll content, and hence serves as a useful measure or primary productivity.

Figure 2. Reflectance Spectra of Vegetation.
Modified from Jensen, J 2000. Remote Sensing of the Environment. Prentice-Hall.
Courtesy: Sharlynn Sweeney, Department of Environmental Engineering

RESULTS AND DISCUSSION

The land cover classification using five and seven classes were useful for distinguishing between agricultural fields and forested regions and shows that there have been changes in the province especially in the forested regions (Figure 3). It failed to account for the spectral differences between open water and dense forests.

Figure 3. Isodata Classification using seven classes.

The land-cover classification using twelve classes was more effective distinguishing between the forested, open water and agricultural areas within the province. This classification also helped identify three types of forest found in the province (Figure 4).

Figure 4. Isodata Classification using twelve classes.

 

These twelve classes were then aggregated to form five basic types of land cover classes &endash; open water, agricultural fields, and 3 types of forests. (Figure 5) This method was helpful in identifying areas where there has been change in land cover, distinguishing between open water, agricultural fields and 3 types of forests.

Figure 5. Isodata Classification using twelve classes clumped to five classes.

NDVI was initially calculated for the entire province and then just the southern forested regions given below in (Figure 6 and Table 1).

Figure 6. NDVI Classification

Table 1 NDVI Values for the entire province and southern forested tambons
NDVI	Mean	Std Dev
1990 November Composite	0.25	0.16
2000 March Composite	-0.10	0.10
1990 November Forested Region	0.52	0.11
2000 March Forested Region	0.12	0.09 These values indicate a decrease in primary production in Sisaket generally from November 1990 to March 2000. Some of this difference is probably due to the differences in seasons between the two years. The November 1990 scene is at the end of the growing season and the March 2000 scene is at the end of the dry season. NDVI values could be influenced by the types of crops that are grown in March versus those crops that are grown in November and it may also be influenced by the crops' growth stages. The differences in NDVI in only the southern-forested regions from the two years in same region clearly indicate that primary production in the region has decreased. The differences may be, at least in part, a consequence of leaf-less trees in March vs. late-season leaves in November leading to higher NDVI in the November scene, or this might be a result of deforestation that might have occurred in the southern parts of the province. Figure 7. Selected Tambons in Sisaket The NDVI data were subsetted to examine NDVI values for 22 individual tambons in the province (Figure 7). Eleven of these tambons are ones in which socio-economic surveys, including soil surveys, were conducted in 1997, 1999, and will be repeated in 2001 (R. Townsend and M.W. Binford, personal communication) and the remaining eleven were selected using stratified random sampling. The stratification was based on forested versus non-forested tambons so as to ensure that a few forested tambons would be included in the analysis. The change in NDVI from 1990 to 2000 was calculated and is presented in Table 2 and Figure 8.   Table 2 NDVI values for the 22 selected tambons Tambon NDVI 1990 NDVI 2000 NDVI % Age Change in Mean Std Dev Mean Std Dev Diff NDVI Jig Sang Thong 0.11 0.10 0.03 0.05 0.08 73.35 Muang Can 0.13 0.89 0.04 0.04 0.09 65.97 Kung 0.08 0.08 0.05 0.05 0.03 38.54 Rung Reng 0.20 0.11 0.05 0.04 0.16 77.81 Huay Tik Shoo 0.31 0.12 0.09 0.04 0.22 70.91 Jai Dee 0.26 0.10 0.06 0.03 0.20 76.36 Prue Yai 0.28 0.11 0.08 0.04 0.20 71.75 Buk Dong 0.48 0.14 0.20 0.07 0.28 57.96 Wiang Nhua 0.32 0.15 0.11 0.04 0.21 65.24 Pi Bung 0.24 0.11 0.06 0.03 0.18 73.93 Som Poy 0.14 0.13 0.07 0.05 0.06 46.92 Nhong Ka 0.15 0.15 0.06 0.05 0.08 58.16 Nhong Mee 0.12 0.12 0.06 0.04 0.07 53.55 Hua Chang 0.17 0.09 0.04 0.04 0.13 75.68 An equality of variance test based on the F statistic was carried out (Table 3) at the a = .01 level. The results indicate at the 99% confidence level we can reject the null hypothesis that the two means are equal and support the alternative hypothesis that the mean NDVI values in November 1990 are higher than in March 2000. Table 3 F-Test Sample for Variances Results NDVI 1990 NDVI 2000 Mean 0.225609 0.077119 Variance 0.011395 0.001696 Observations 22 22 df 21 21 F 6.72006   F Critical one tail 2.857377   Figure 8. Change in NVDI in the 22 selected tambons. This was followed by a regression analysis (Figure 9) of NDVI in November 1990 and March 2000. The points above the line represent tambons where change in primary production was higher than expected and those points below represent those where primary production was lower than expected. Maximum absolute change in NDVI occurred in the two southern-forested tambons. These are known to be among the poorer tambons in Sisaket. The least absolute change in NDVI occurred in the northernmost tambons known to be economically better off (Figure 8). But in terms of proportional changes the average percentage change in NDVI was around 65% for most of the 22 tambons. In Figure 9 we see a spatial pattern of the northern-richer tambons having lower decrease in NDVI values to those southern-poorer tambons having greater decrease in NDVI. Figure 9. Regression Analysis of Primary Productivity in Sisaket as measured by NVDI. CONCLUSION Based on land cover classification and calculation of NDVI, it is clear that there have been changes in land-cover and land-use in the province of Sisaket from 1990 to 2000. The tambons did not change evenly. Primary production decreased in the entire province, but the southern tambons showed a greater decline and there is some indication that the ensuing spatial pattern may be directly related to economic growth patterns within these tambons. I intend to link remotely sensed data on land cover and land use in Sisaket with appropriate socio-economic data being measured in the province currently. This should result in an ecological-economic model that captures both spatial and temporal aspects of natural and human induced changes that govern land cover and land use change within the province of Sisaket. REFERENCES Jensen, J.R. 2000. Remote Sensing of the Environment: An Earth Resource Perspective. Prentice-Hall. Upper Saddle River, NJ 544 pages. Jensen, J.R 1996. Introductory Digital Image Processing: A Remote Sensing Perspective. Prentice-Hall. Upper Saddle River, NJ 316 pages. Research Systems, 1999. The Environment for Visualizing Images. Better Solutions Consulting Limited Liability Company (BSC) 864 pages. Feder, G. 1988. Land Policies and Farm Productivity in Thailand. John Hopkins University Press. Baltimore, MD 165 pages. Nicol, K.J. 1982. Agricultural Planning in Thailand. Iowa State University Press. Ames, Iowa 326 pages. Wyatt, D.K. 1984. Thailand- A Short History. Yale University Press. New Haven, CT 351 pages. Liverman, D., Moran, E.F., Rindfuss, R.R, and Stern, P.C, Editors. 1998. People and Pixels. National Academic Press. Washington, D.C. 244 pages. --top-- Back to the Journal of Undergraduate Research College of Liberal Arts and Sciences | University Scholars Program | University of Florida | © University of Florida, Gainesville, FL 32611; (352) 846-2032.