The integration of image data into GIS is one of those great ideas whose time has come. Furthermore, remote sensing is often the most cost-effective source of information for updating a GIS and it is a valuable source of current land use/land cover data. Remote sensing techniques has been utilized successfully in certain areas of application, including agriculture and related fields, especially in the developed countries were agricultural patterns are well defined and methodologies developed. The areas of applications in agriculture have been the identification or classification of crops, inventory of crop acreage, forecasting of crop yield, soil survey, design and operation of irrigation projects, and assessment of flood damage. Soil has an easily distinguishable characteristic reflectance pattern in the visible, near-infrared and mid-infrared wavelengths. The characteristic of soil reflectance pattern is easily distinguishable from green vegetation. Leaf biomass for most crops is highly correlated with the leaf area index (LAI). Change of LAI during vegetative stages of crops is closely related with growth. Remote sensing and crop growth simulation models have become increasingly recognized as potential tools for growth monitoring and yield estimation. Some cost benefit studies indicate that satellite remote sensing can provide cheaper and faster information. Economic efficiency of remote sensing data applications can be expressed both directly as reduction of the cost and indirectly by an increase in the quality, reliability, and details of information. Malaysian agriculture should attempt to operationalize remote sensing and geographic information techniques (GIS) intensively for future planning and control of the sustainable agriculture development. Remotely sensed data, when complemented by existing and supporting GIS, could improve management decision in Malaysian agriculture for the next millenium.

Keywords: Remote sensing, Geographic information system, Precision farming.
 
 
 
 

Agriculture is highly dynamic in nature because of the changing phenomenon of agricultural crops which is further complicated by the interaction of the crops with the environment. If accurate results are to be expected, then proper interpretation of remotely sensed data integrated into a Geographic Information System (GIS), knowledge of agricultural crop phenology of the area or region, and ground verification during the time of collection of remotely sensed data have to be collectively taken into consideration.

The primary considerations for ground verification in agriculture are aspects of crops, soils, insects, pest and diseases, meteorological factors, etc. Remotely sensed data (spectral values or measurements) can be a source of useful information in many agricultural applications. Some of the most promising agricultural applications include: (1) crop identification and area estimation, (2) crop condition assessment, (3) yield forecast and estimation, (4) rangeland surveys, (5) soil survey and mapping, and (6) water resource surveys and mapping for the purpose of water supply and irrigation, etc. (Nualchawee, 1984). With a high resolution remote sensing (RS) data, farm managers can obtain details and specific crop and soil information over large areas to assist in decision making, but the greatest agricultural benefits of remote sensing is expected to be inventories performed over large areas over a certain required time in temporal analysis.

One of the main constrains of crop identification on satellite imagery is the relatively low spatial resolution. Besides the small fields sizes, tropical countries often face problems due to high cloud coverage (Reichert, 1984). Recently, remotely sensed data have been integrated in GIS databases, such as to facilitate temporal analysis for resources monitoring. Furthermore, remote sensing is often the most cost-effective source of information for updating a GIS and it is a valuable source of current land use/land cover data (Silapathong and Blasco, 1992). The greatest benefit of using SOYGRO with remote sensing and GIS technologies is the ability to detect spatial patterns of yield during individual growing season (Carbone et al. 1996). Upon establishing a maximum yield for a given site, is necessary to determine maximum economic yield (MEY) by selecting those management practices and input levels which will provide the highest profitability to the producer (Pretty and Sanders, 1985). This is not necessarily maximum yield, but may be to maximize financial advantage while operating within environmental constrains. Remote sensing data can be used to map and quantify crop responses to variable soil and weather condition and farm management. Remotely sensed spectral information, combined with simulation models and fields measurements are the main components towards a precision farming decision support system (Booltink and Verhagen, 1997).
 
 

2. APPLICATION OF REMOTE SENSING AND GEOGRAPHIC

Precision Agriculture (PA) or Precision farming (PF) is a technology already available for sustainable agriculture production which enables farm management based on the small scale spatial variability of soil and crop parameters in the fields. For precision agriculture to be successfully implemented, efficient, and accurate methods for accessing the spatial variability of soil, and relating it to quantity and quality of yield components, must be developed and utilized (McBratney and Pringle, 1997).

Precision agriculture deals with the question of where and when. Many, but not all of those questions are related to soil conditions in space and time. For growing most crop in the world, we deal with the number of questions that relate to soil conditions. Examples are when to: (a) sow or plant, (b) fertilize, (c) apply biocide to combat pest and deseases, (d) allow soil traffic, and (e) apply tillage (Bouma, 1997). Precision farming is a tool for increasing the economic efficiency of agricultural production and a way of better accounting for ecological requirements. In order to account for heterogeneity of agriculture field soils, the temporal and spatial variability patterns of their soils properties have to be known (Domsch and Wendroth, 1997). For example, a Californian farmer Ken Deaver of Plymouth, in Amador county, will check the weather outlook and decide which rows of winegrapes to be harvested. With a few clicks of the keyboard, and from comfort of his home office, he will view color-coded maps of his property that can give advance warning of pest infestations, over irrigated soil conditions or under fertilized regions. The GIS is expected to save substantial amounts of money by producing maps
 
 

that show where and when to pick crops, enhancing the very tedious and time consuming logistics of moving crews and equipment at night (Lang, 1997).

The aim of PA/PF in relation to cropping is to match "resources application and agronomic practices with soil attributes and crop requirements as they vary across a site" (McBratney and Pringle, 1997). Once the maximum yield for a given site has been established, it will be necessary to determine the MEY by selecting those management practices and input levels which will provide the highest profitability to the producer (Pretty and Sanders, 1985).

Remote sensing techniques and GIS have been utilized successfully in certain areas of application, including agriculture and related fields, especially in the developed countries where agricultural patterns were well-defined and methodologies developed. Proper and effective management decisions which ultimately led to economic gain and environmentally ground farming decision seems to be the results of Precision Agriculture Practices in Muda Agriculture Development Authority (MADA), Kedah, Malaysia (Kamaruzaman and Dayang, 1998). The areas of applications in agriculture have been the identification or classification of crops, inventory of crop acreage, forecasting of crop yield, soil survey, design and operation of irrigation projects, and assessment of flood damage (Nualchawee, 1984). Remote sensing data can be used to map and quantify crop responses to variable soil and weather conditions and farm management. Global Positioning System (GPS) for positioning and navigation in the field and GIS is required for the computation of geo-coded field data. Precision farming requires accurate maps of plant nutrients in the soil within fields (Oliver et al. 1997). GIS is a great tool to rapidly demonstrate or to communicate research results to user in a user-friendly form (Bouma, 1989).
 
 

2.1 Application RS and GIS in Land Use Inventory

Up-to-date information of major land use, acreage, and distribution of crops is a basic need in agriculture throughout the world. In both the developed and developing countries, these data are essential for efficient management of agricultural resources and sustainability. Planning for future growth is greatly enhanced based on the information on the present extent, location, and productivity of land used for different purposes is needed for analysis. Competition for use of land is currently attracting much attention.

Manual (traditional) mapping methods take a relatively long time and high cost. Pressing needs for land use inventory in many of the developed countries and some of developing counties have already prompted the use of remote sensing and GIS to provide up-to-date information (Luney and Dill Jr., 1970). Recent advances in remote sensing and GIS technology has become very cost effective, due to the following reasons: (a) satellite images are sufficiently accurate and reliable, (b) changes over time can be identified, (c) computers have the power to rapidly process large quantities of data, and (d) object-oriented GIS provide enormous flexibility in storing and analyzing any type of data, providing decision support modeling for effective management (Buchan, 1997). Vegetation reflectance involves four different and complementary points of view which have led to distinct works: (a) the reflectance of plant parts (single organs-pigments), (b) reflectance of sets (canopies), (c) nature and state of the plants, and (d) structure and texture of the set (Goillot, 1980). There are numerous factors, either internal to the plant or external, coming from the environmental conditions, which have an influence on the specific spectral reflectance. For instance, using radiometric measurements in the field, calculated reflectance values of various land cover types such as coconut, oil palm and rice in Sungai Besar area, as showed in Figure 1 (Azhar,1993)
 

In the visible spectrum, coconut and oil palm reflectance show relatively low values, with a slight increase for rice. However, in the near infrared region spectrum, rice reflectance lower than coconut and oil palm. When the radiometric measurements were taken, rice was in the mature stage, whereas coconut and oil palm trees were found healthy and green. Healthy green vegetation is characterized by very high reflectance and very low absorption in the infrared region (Lillesand and Kiefer, 1987).

The new generation of GIS that integrated satellite images with maps data means that this technology can be successfully used for remotely monitoring land use cover. According to Buchan (1997) when image analysis and GIS are combined into one package, it can offer a very efficient and cost-effective solution. Satellite images are used to identify what is growing, while the GIS component is used to measure area, categorize it, and locate its position on the earth?s surface to provide complete record of the site. European Space Agency (ESA, 1997) reported that analysis of European Remote Sensing Satellite-Synthetic Aperture Radar (ERS-SAR) unsupervised isodata has a classification accuracy of 66.3 %, whereas a SPOT-based accuracy is estimated at 85 % (Table 1).
 
 

Table 1. Comparison of Early Surface Estimates for The Chartres (F) Test Site

Using ERS and SPOT (ESA, 1997).
 
 

The images were derived from a classification utilizing data from Synthetic Aperture Radar (SAR) instrument onboard a European Remote Sensing Satellite (ERS-2). Preliminary results obtained in early spring 1997 showed four distinct classes from the interpretation of four images acquired in the time frame between October 1996 and February 1997 (Figure 2).

2.2 Application of RS and GIS in Soil Characteristics
 
 

Most remote sensing techniques use radiation which shows only a shallow penetration upon interaction with soil, rock and plant materials. By using these techniques, it is only possible to obtain direct information about the surface of soils and rocks or about vegetation covering of soil. Field-works is necessary to estimate the properties of the three-dimensional soil profile (Mulders, 1987). Remote sensing in general is provide an inexpensive and valuable tool to detect long term variability within fields, it also contributes to a more efficient soil sampling design. Satellite images provide a good source of information for large areas such as farms larger than 1,000 ha. Experience has shown that many earth surface features of interest can be identified, mapped, and studied on the basis of their spectral characteristics (Grenzdorffer, 1997). Soil can be delineated on a reconnaissance level from Landsat data, especially in semi-arid areas, where soil is not covered by vegetation throughout the year. Soil erosion due

to improper land use, soil degradation and soil salinity can also be delineated from
 
 

spaceborne remote sensing data. Even more important is the assessment of soil moisture, being one of the prime factors for agricultural production (Reichert, 1984).

Soil reflectance properties depend on numerous soil characteristics such as mineral composition, texture (particle size), structure (surface roughness), percentage of organic matter, and moisture contents (Bunnik, 1981, Lillesand and Kiefer, 1987). These factors are complex, variable, and interrelated. Mineral composition, organic matter and moisture content are the main factors governing spectral absorption of radiation. Azhar (1993) reported that there are differences in reflectance of three types of soil i.e peat, paddy and forest soils in Sungai Besar, Selangor (Figure 3).

The curves for forest and paddy soils show higher reflectance values as compared to the peat soil. The high organic matter and soil moisture content of the peat soil must have caused influence the low reflectance of the peat soils. Field soil reflectance is reduced, particularly in the visible portion of the spectrum, when organic matter, iron oxide, or moisture content is high. The near infrared and middle infrared region of the spectrum were also affected. Soil has an easily distinguishable characteristic reflectance pattern in the visible, near-infrared and mid-infrared wavelengths. Soil reflectance pattern is generally linear with increasing reflectance as wavelengths increase from visible to mid-infrared (Todd and Hoffer, 1998). The characteristic of soil reflectance pattern is easily distinguishable from green vegetation. Green vegetation reflectance is low for visible bands (particularly red), with a sharp increase in reflectance in the near-infrared portion of the spectrum. Reflectance is low in the mid-infrared regions associated with water absorption (Lillesand and Kiefer, 1987).

Some of the soil induced effects on vegetation indices have been attributed to additional near infrared (NIR) irradiance underneath and in-between canopies due to NIR scattering and transmission properties of canopy, with intermediate canopies displaying the largest effects (Huete, 1988). Canopy scattering is small with low vegetation cover while the soil signal is small with high vegetation cover. Soil reflects some of the scattered and transmitted NIR flux back toward the sensor , depending on the soil reflectance properties. The red to near infrared feature space shows a positive correlation between the increase of leaf area index and the near infrared/red reflectance ratio, both for a dry and a wet surface (Bunnik, 1981).

From satellite imagery, managers can see soil patterns that help explain why they see differences in yield or quality, even they can see small differences in crop vigor, before they can see it with naked eye (Arvik, 1997). With a GIS, stratified soil samples can be collected and could generate a soil nutrient gradient model. Figure 4 showed a significant soil change on the upper left hand corner of the image in indicated by a box. Top soil was re-graded to accommodate the change in irrigation flow direction. The disturbance of the top soil affects the plant vigor over the entire plot (Pelzmann Jr. 1997). Different land use or different vigor in the same land use usually indicate the differences in soil characteristics.
 

2.3 Application of RS and GIS for Crop Performance Monitoring

Spectral curve for healthy green vegetation almost always manifest the peak and valley configuration. The valleys in the visible portion of the spectrum are dictated by the pigments in plant leaves. Chlorophyll, for example, strongly absorb energy in the wavelength bands centered at about 0.45 and 0.67 m m. Hence, our eyes observe healthy

vegetation as green in color because of the very high absorption of blue and red energy by plant leaves and the high reflection of green energy. The chlorophyll plant features a typical spectral reflectance the general aspect of which, for a healthy plant, in the range from 0.4 to 2.6 m m. High absorptance in the ultra violet and the blue, reduced in the green, high in red, and very low in near infrared. The very abrupt increase in reflectance of near 0.7 m m and the fairly abrupt decrease near 1.5 m m are present for all mature, healthy green leave (Goillot, 1980). Difference kind of crop plantations and either difference stages of growth in one crop plantation, will be difference reflection in image
 

performance due to difference in spectral reflectance. Figure 5 shows spectral reflectance curves for different types of vegetation (Azhar, 1993).

Figure 5. Spectral Reflectance Curves for Various Vegetation
in Raub (Azhar, 1993)
 

By comparing a series of such records, it is possible to built up a history on the development of the vegetation in a region and to generate statistics on the rate of growth or decline, patterns of growth, ownership, etc. (Figure 6).

Green vegetation reflectance is low for visible bands (particularly red), with a sharp increase in reflectance in the near infrared portion of the spectrum. Reflectance is low in the mid-infrared regions associated with water absorption. Physical vegetation properties vary with plant species, environmental stress, and phenology (Todd and Hoffer, 1998). Pigmentation and moisture content change as a plant senesces. A loss of chlorophyll pigmentation produces higher visible reflectance, particularly in red region of the spectrum. Identification of healthy and diseased sugar beets (affected by nematodes) were accurately possible and different degrees of infestation have also been distinguished. Several other studies have proved the feasibility of applying remote sensing to disease detection, mainly for forestry crops (Riechert, 1984). Spectral reflectivity characteristics of damaged leaf of soybean by Spodoptera litura were showed in Figure 8 (Yamato et al. 1991). Spectral reflectivity of percentage of damage leaf areas 87.6 % in near infrared range, was lower by 55 % than spectral reflectivity of in test leaf.

As showed in Figure 9, information will be important for an early planning replanting costs and identifying potential problem areas. Differences of color in the image, indicate difference performance of crop, as vigour, ages of crop and homogeneties growth of plant. In the same age of crop, red colors indicate better vigor and healthy than blue colors.

Compared to traditional methods, such as experimental harvests and other types of point observations in the field, remote sensing has clear additional values such as the non-destructive collecting crop date, it supplies a full spatial cover, and relatively low cost, especially if compared to laborious field sampling. When applied on large scale and using standardized, commercially viable sensors, cost can be reduced significantly.
 
 

During the last 10 years, several mathematical models simulating canopy reflectance have been developed to investigate the influence of changing crop properties and changing illumination and observation conditions. A direct advantage of such models is the possibility to improve the understanding of the role of the different crop parameters in the process of multi-spectral reflectance. Vegetation canopy reflectance models are subdivided into deterministic and statistical models. The deterministic approach predicts reflectance values which represent the expectation value of the actual statistical distribution of independent reflectance values. By means of statistical models, the probability distribution function of reflectance values is obtained. Both type of models are based on the assumption of indefinite horizontal extension of the canopy, the subdivision in layers and the use of optical and structural parameters as input data.
 
 

2.4 Application of RS and GIS in yield forecasting

Prediction and estimation of yield is closely related to the capability of identifying crop species and certain agronomic variables such as maturity, density, vigor, and disease
 
 

which can be used as yield indicators (Nualchawee, 1984). The use of remotely sensed data in crop acreage estimation has been demonstrated by various researchers in different parts of the world (Saha and Jonna, 1994). Satellite data are complementary to data from GIS, Global Positioning System (GPS), yields monitor, and penciled notes on the back of envelopes. Data from all sources should be brought together to give the grower the best opportunity to maximize yield and quality (Arvik, 1997).

Yield is influenced by a large number of factors such as crop genotype, soil characteristics, cultural practices, weather conditions and biotic influences, such as weeds, diseases and pests. Two approaches adopted for yield modeling using remotely sensed data, firstly is remotely sensed data or derived parameters that are directly related to yield, and the others is remotely sensed data that are used to estimate some of the biometrics parameters, which in turn are input parameters to a yield model. Geographic information system can handle, manipulate and analyze data from different sources and coordinate systems, scales and formats (Navalgund, 1994).

Remote sensing and crop growth simulation models have become increasingly recognized as potential tools for growth monitoring and yield estimation (Bauman, 1992). In the last few years, attention has been paid towards using satellite remote sensing data in crop estimation surveys, in view of its advantages over traditional procedures in terms of cost effectiveness and timeliness in the availability of information over larger areas. The existing procedures derive post-harvest yield estimates through crop cutting experiments (CCE) over large administrative units such as a district (Murthy et al. 1994). The comparison of satellite estimated yield and the actual yield obtained CCE in 1993-1994 Rabi Season is presented in Table 2. The adaptability of yield model developed with 1992-1993 Rabi data to predict the yield of 1993-1994 yield has thus been validated.

Murthy et al. (1995) reported the validity of crop yield models with satellite derived normalized difference vegetation index (NDVI) is determined by the strength of association between the two variables including in the model. Hence, it is essential to understand the correlation existing between yield and NDVI at different phonological stages of the crop for selecting an appropriate date of pass to include in the model

(Table 3).
 
 

Table 2. Validity of yield estimates in 1993-1994 Rabi Season

(Murthy et al. 1994).
 
 

Table 3. Correlation between yield and NDVI at different stages of

The correlation coefficients were found to be statistically significant with the NDVI of April 16^th and April 27^th which represent towards a heading phase of standing paddy crop. The correlation become weaker with NDVI moving away from heading i.e., at panicle initiation and ripening. It is appropriate to select the satellite data representing either a heading or just before heading phase. However, in practice, a single date may not represent the same physiological process all over the study area due to differential crop calendars adopted by farmers. As a process result, the coefficient correlation is not applicable to the entire study area. In general, such problems can be overcome through time composition of multi-date satellite data representation (Murthy et al. 1995; Nualchawee,1984). Seeni Mohd. et al. (1994) reported that the NDVI values obtain from a combination bands 3 and 4 of the Landsat-5 TM showed better correlation with yield as compared to the RVI values obtained from the combination of bands 3 and 4 and combination of bands 3 and 5. NDVI is the ratio between the difference in the infrared and red bands and the sum of these bands, i.e.,

(IR - RED)

(IR + RED)
 
 

The infrared bands used are band 4 (0.76-0.9 m m) and band 5 (1.55-1.75 m m), whilst the red band is band 3 (0.63-0.69 m m) of the Landsat-5 TM data. The paddy yield estimate using the NDVI gave 30 % higher estimate of yield when compared to the field estimates. The yield estimation from satellite data can be calculated as follows:
 
 

YIELD (kg/ha) = 34.30 x (NDVIsat - 0.1836) - 237.85
 
 

The average yield in the study area has been found to be about 6,750 kg/ha by using the equation above. However, the yield data from the Department of Agriculture indicated an approximate value of 5,175 kg/ha for the same year (Seeni Mohd. et al. 1994).

During the mid 1960s, Mark Systems Inc. developed techniques for forecasting the yield of wheat, rice and sugar cane using high altitude aerial photography. Their technique is to acquire historical yield data and information about crop phenology, local agronomic conditions, and climatological conditions, and then subtract yield reducing such as disease, low crop density due to moisture or nutrient deficiencies , or lodging. Using Landsat time series coverage, Earth Satellite Corp. developed a production model for spring wheat. In Williams Country, North Dakota test area, the production estimates for a sample of 42 fields and for the whole test site were corrected to within 0.1 % and 6.2 %, respectively (Nualchawee, 1984). The use of satellite remote sensing data for agriculture crop (mainly wheat and rice) studies at the regional level has been under investigation in India since 1986 (Saha and Jonna, 1994). Paddy accounts for more than 40 % of the total annual food grain production in India. Digital supervised classification of Landsat MSS data has been used for the identification and district level acreage estimation of kharif paddy. A two stage stratified sampling approach and supervised digital classification of Landsat MSS, Landsat TM, and IRS LISS-I data gave better estimates of paddy crop acreage in large study areas such as a group of districts or state. Various studies using ground based experiments have shown a good relationship between crop yield and spectral vegetation indices. Spectral vegetation indices (SVIs) are normally derived by taking into ratio or normalized difference of the reflectance in the near infrared and visible red bands of linear band combination. Digital analysis for paddy acreage estimation consisted of three major steps as district boundary mask generation, training signature generation, and supervised maximum likelihood classification (Saha and Jonna, 1994). The district boundary was first traced from a topographic maps. After digitizing ground control points (GCPs), their latitude and longitude coordinate were obtained. Image coordinates of the GCPs, along with latitudes and longitudes (map coordinates) were used to obtain the second order map image affine transformation. The districts boundary was then digitized to obtain the latitude and longitude of each vertex of polygon. This boundary was later overlaid on an image, which was then classified to get the district crop acreage. Ground truth sites for paddy and other classes like scrub, barren land, plantation/forest, fallow land and water bodies were identified from the image displayed on a Pericolour monitor, with help of ground truth information.

The experiment established that the technology developed for large area crop inventory experiment (LACIE) met the performance goals for wheat production inventory. The 1977 Soviet final production estimate released in January 1978 was 92 million metric tons and the LACIE final estimate was 91.4 million metric tons, a difference within 1 % (Bryan, 1980). LACIE was mainly to demonstrate the feasibility of applying spaceborned remote sensing techniques for harvest yield production of wheat in USA. Landsat data of sample plots were evaluated in combination with historical yield statistics and current weather data. Hence, the great plant of the US were subdivided into 12 zones and yield prediction model was developed for each of these zones. The results were found to meet the 90/90 criterion, that means an accuracy of

90 % of the harvest yield was achieved in 9 out of 10 years with the one exeption of 1974, which was a very dry year (Reichert, 1984).

3.0 ECONOMIC IMPACT OF RS AND GIS
 
 

The benefits from using RS and GIS technology depend on the level of success of its application for solving a concrete task. In general, these benefits can be divided into for categories such as scientific, technological, methodological, and economic efficiency (Badarch, 1990).

The scientific efficiency of remotely sensed data also includes obtaining new facts for corroboration and quantitative clarification of previously known, qualitatively studied data. Technological efficiency means increasing of the work productivity (mainly the most expensive field job), making norms for fieldwork and speeding up of natural resources mapping, reducing the fieldwork volume, shortening the time necessary for territorial surveys and reducing the number of personnel engaged in natural resources surveys. Methodological efficiency means increasing the accuracy and detail of spatial research of natural resources and also of observing widespread and dynamic processes and phenomena. Finally, economic efficiency of remote sensing data applications to natural resources can be expressed both directly (in the reduction of the cost of mapping) and indirectly (by an increase in the quality, reliability, detail, and information of the results).

The economic benefits of using remote sensing is very difficult to estimate. The direct and indirect cost benefit ratio in Thailand varies from 1:1 to more than 1:10, as shown in Table 4 (Vibulsresth and Ruangsiri, 1990). The national survey of land resources with remote sensing cost only 0.04 yuan/km2 compared with conventional methods of more than 1.00 yuan/km2 (Hua, 1990).

The cost of remote sensing data is dependent on the cost for the entire system, consisting of data acquisition, processing and analysis components. This can be broken down into major cost items as follows: satellite development and operation, ground receiving stations and preprocessing facilities, and data processing, analysis and presentation to user community (Nanayakkara, 1990). Satellite remote sensing system produce the data provided by ground systems and airborne surveys is much more intensive (detailed) and specific in nature and is project oriented.
 
 

Table 4. Remote sensing application project economic analysis

(Vibulsresth and Ruangsiri, 1990)
 
 
 

Item	Name of project	Area (sq.km)	Time (year)	RS. Application cost (Bath)	Conventional cost (Bath)	B/C ratio
1	Evaluation of landuse: Its change and impact on Khaoyai National Park	5,859	½	13,500 (2.3 bath/sq.km.)	1,757,700 (300 bath/sq.km)	1:130
2	Geomorphological units as seen from the landsat imagery southeast coast of Thailand	28,000	1	69,500 (2.5 bath/sq.km.)	25.200,000 (900 bath/sq.km)	1:360
3	The studies and assessment of flood damage in Southern Thailand.	300	1	355,200 (1,184 bath/sq.km)	3,750,000 (12,500 bath/sq.km)	1:10
4	Application of GIS for agriculture development in Amphoe Phatana Nikhom and Chai Badan, Lop Buri.	225	1	320,000 (1,422 bath/sq.km)	2,812,500 (12,500 bath/sq.km)	1:9
5	Base map revision using high resolution satellite data.	16,848	1	126,000 (7.5 bath/sq.km)	21,060,000 (1,250 bath/sq.km)	1:167
6	The application of satellite imagery for water quality study in the Inner Gulf of Thailand.	10,000	1	34,000 (3.4 bath bath/sq.km)	300,00 (30 bath/sq.km)	1:9
7	Mapping and monitoring of settlement changes in areas with environmental limitation using RS data.	2,500	1	63,800 (25.5 bath/sq.km)	3,125,000 (1,250 bath/sq.km)	1:49
8	Feasibility study on an estimate of fruit trees cultivated area through remote sensing method in the export oriented province.	28,022	1	92,800 (3.3 bath/sq.km)	23,342,326 (833 bath/sq.km)	1:252
9	Metropolitan and it?s vicinities using high resolution satellite data for development planning	3,920	1	181,000 (46 bath/sq.km)	9,800,000 (2,500 bath/sq.km)	1:54
10	A study on natural resources landuse and infrastructur of Bang Pa Khong watersheed by RS data.	21,000	1	165,000 (7.86 bath/sq.km)	13,125,000 (625 bath/sq.km)	1:79

Note: 1. 25 bath = 1 US$ (approx)

2. Selected from 108 NRCT projects
 
 

Topographic data of a high accuracy is provided, and the metric quality meets accuracy standards specific to the scale of mapping. Satellite imagery provides the most cost-effective method of gathering information from repeated observation of croplands and, in some cases, is the only way that such information can be obtain economically. By selecting the resolution of imagery based on the type of information needed, crop vigor and production efficiency across wide areas and over many fields can be monitored from a single location. Santos (1990) highlighted that estimations of the cost of these activities should consider the time factor, i.e., the timeliness and relevance of the data in order to grasp the real advantages of remote sensing methodologies over the more conventional means of resource assessment, monitoring and mapping (Table 5).
 
 

Table 5. Comparison of forest inventory methods (Santos, 1990)
 
 
 

Criteria	Manual Landsat Interpretation	Digital Landsat analysis	Multi-Stage  Remote sensing	Aerial photo interpretation
A Financing  B. Man-days	US $1,050  40 man-days	US $1,290  32 man-days	US $163,060  72 man-days	US $346,578  100 man-days
Equipment needed	Light table, additive colour viewer, lenses	Computerized analyser	Computerized analyser, stereoscope, zoom ransfer- scope, planimeter	Stereoscope, zoom transfer-scope
Scale of final output maps	1:250,000	1:250,000	1:250,000	1:50,000

 
 

4.0 RS AND GIS IMPLEMENTATION IN MALAYSIAN AGRICULTURE

Like most other developing countries, Malaysia places immense emphasis in the management and exploitation of its natural resources and environment. The National Resource and Environmental Management Program (NAREM) was initiated in 1996 under the aegis of the Economic Planning Unit (EPU), Prime Minister?s Department with the objective of collecting, inventorying, archiving and retrieving various natural resources data for rational planning of national development projects. For this purpose, natural resource maps of agriculture, forestry and geology, among others, are being revised every eight years by the government agencies concerned, by using aerial photographs and ground-survey techniques.

Aerial photo surveys at a scale of 1:25,000 and ground surveys using base maps of 1:25,000 to 1:63,360 were conducted to acquire natural resources data. Due to the unfavourable climate conditions of the country (heavy rain and thick clouds), the acquisition of aerial photographs has been costly and uncertain in various parts of Malaysia. Satellite remote sensing techniques were introduced in Malaysia during the early eighties under the auspices of the National Remote Sensing Program (NRSP), initially implemented by EPU but later by the Malaysian Centre for Remote Sensing (MACRES) of the Ministry of Science, Technology and the Environment. The primary objective of the NAREM is to established and operate a comprehensive system of spatial database on national resources and the environment to support integrated national development planning using RS and GIS technologies. Secondary objective is to strengthen capacity building and human resource development at national level to facilitate operationalisation of the programme (Mahmood and Loh, 1999).

Prime Minister Department (1996, In Mahmood and Loh, 1999), reported, in the seventh Malaysian Plan 1996-2000, IT has been trusted into the forefront of national development focusing on increasing the spectrum of IT usage particularly covering education, business, social economic development and R&D intensification. RS and GIS fall within the IT domain. Integrated development planning necessitates the input of reliable and timely information, which can be provided by satellite remote sensing , cost effectively.

Peninsular Malaysia covers an area of about 132,000 km2. In order to have full coverage of the peninsular, a total of 15,750 aerial photos at 1:25,000 scale was needed to be secured through aerial survey conducted by the Department of Survey and Mapping, Malaysia (Mahmood and Fook, 1990). From these areas, 35.25 % were covered by agricultural land. Crop hectareage by categories in Malaysia was 4,653,091 ha and can be categoried as follows: industrial crops occupy the largest area (4,347,951 ha), followed by fruits crops is 260,882 ha, vegetables crops is 260,882 ha, cash crops is 8,808 ha, spices /rempah is 2,614 ha, and other crops 2,638 ha (Agriculture Department of Malaysia , 1998).

A case study on the use of NAREM database for interactive agriculture land use planning in the North West Selangor Integrated Agriculture Development Project (IADP) was reported by Loh, et al. (1999). The operational model adopted resides on three important component as Agro-ecological Zoning (AEZ), Agro-suitability Zoning (ASZ), and Interactive Multiple Goal Linear Programming (IMGLP). In Kemubu Agriculture Development Authority (KADA) Kelantan, Kamaruzaman and Nasri (1999) reported that soil fertility status in Ladang Merdeka, was classified as follows: total nitrogen varied from 4.55 to 21.63 ppm, phosphorous is 6.13 to 17.51 ppm, and potassium is 62.12 to 93.27 ppm. Whereas, soil pH varied from 4.94 to 5.59. Meanwhile, in another study at UPM Pasture Farm Selangor, the total nitrogen varied from 13.8 to 320.6 ppm, phosphorous is 3.96 to 12.9 ppm, and potassium is 1.39 to 4.60 ppm, whereas soil pH varied from 3.40 to 4.72 (Kamaruzaman et al. 1999)

Based on the area extent of agricultural uses, RS/GIS should be implemented for planning and controlling of the agricultural development. Integration of technology remotely sensed data, when complemented by existing and supporting information technology like GPS, could aid management decisions. The above mentioned points are of great value to developing countries like Malaysia whose economics rely greatly on the agricultural production. Remote sensing applications in agriculture will take a large share of the field in the future.
 
 

5. CONCLUSION

Agriculture is highly dynamic in nature because of the changing phenomenon of agricultural crops which is further complicated by the interaction of the crops with the environment. If accurate results are to be expected, the proper integration of remotely sense data and GIS, knowledge on the agricultural crop phenology of the area or region, and ground truth information during the time of collection remotely sensed data have to be taken collectively into consideration.

Being Malaysia an agriculture based country, operationalization of remote sensing/GIS for planning and controlling of the agriculture development should have a good future. When complemented by existing and other supporting information technology like GPS, management decisions in agricultural development and productivity will be more realistic and accurate.