Exploring land use options for agricultural development in Bac Kan Province, Vietnam This work will be presented by Dr. Pham Quang Ha ( Vietnam LUPAS Coordinator at the AEZ-LADA Workshop, Bangkok, Thailand, 10-14 November 2003) LUPAS project was carried out to exploring land use options for agricultural development at provincial level in using Backan province as a case study in period 2000-2002. Different scenarios were formulated under linkage between natural resources evaluation, socio-economic study and multiple goal linear programming for food security and environmental protection. This approach is now developed and adjusted for district levels with different agro-ecological zones both in North and in South Vietnam by NISF team and further research is enforced with the collaboration of the project entilled “Systems Research for Integrated Resource Management and Land Use Analysis in East and Southeast Asia” (IRMLA) INCO-Dev, CEC, Brussels coordinated by Alterra, Wageningen. The follows names were key member of VIETNAM-IRRI LUPAS team: BUI Tan Yen1, KAM Suan Pheng2, PHAM Quang Ha1, CHU Thai Hoanh2, BUI Huy Hien1, CASTELLA Jean-Christophe3, HO Quang Duc1, VU Dinh Tuan1, VU Nguyen4, CAO Ky Son1 1 National Institute for Soils and Fertilizers (NISF), Chem, Hanoi, Vietnam International Rice Research Institute (IRRI), Los Banos, Laguna, Philippines 3 Institut de Recherche pour le Development (IRD, France) – IRRI, Hanoi, Vietnam 4 Vietnam Agricultural Science Institute (VASI), Ha Noi, Vietnam 2 ABSTRACT A study was carried out in the poor, mountainous Bac Kan Province in the northern part of Vietnam to explore land use options and analyze tradeoffs between different development objectives. The methodology is based on a GIS-linked multiple goal linear programming (MGLP) modeling tool, the Land Use Planning and Analysis Systems (LUPAS). The LUPAS methodology comprises three main components: (i) evaluation of the resources required for agricultural development, (ii) estimation of inputs and outputs for production activities, and (iii) optimization of land and resource allocation subjected to policy views and development objectives. This paper describes the use of regional-level soils information in developing the knowledge base for the modeling. In the first step of qualitative land evaluation, timeinvariant terrain and soil properties were used for analyzing slope-soil-crop suitability. Because of the complex topography of the province, we used the raster GIS data structure for more efficient handling of the highly heterogeneous spatial variables. The slope-soil-crop suitability assessed at grid cell level for 18 crop types was used to delineate land units (LUs) within commune boundaries. The second step takes into account dynamic climatic factors that influence crop-season suitability at each LU. Thirdly, each LU was evaluated for its suitability for selected land use types (LUTs), which are combinations of crop-seasons. Data on agricultural production collected at commune level were used to quantify inputs and outputs associated with existing and Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 1 promising LUTs for each LU. The input-output tables were used in optimization modeling. The model computes the optimal allocation of resources (land, water, labor, capital) to achieve user-specified development objectives, under certain assumptions about the levels of resource availability and technology use. The results indicate which LUTs are selected, how many resources are needed, and the production levels and incomes expected. The model results are mapped to show the spatial pattern of the selected LUTs. The outputs provide a rational basis for geographical targeting of different land use types within a particular region. Key words: Land use planning, resource evaluation, soil-crop suitability, optimization model, GIS INTRODUCTION Bac Kan, a province of 459,000 ha in the upper part of the Red River basin, is one of the least developed and poorest provinces in Vietnam (Figure 1). The majority of the population (83%) is rural and dependent on agriculture. Increasing population and greater market integration places higher pressure on the use of the land and its resources. Because of the mountainous terrain, only about 5% of the land area is considered suitable for agricultural production. In reality, increasing areas of sloping land are being used for food production, giving rise to concerns about long-term sustainability of the natural resource base to support the demands for food production and income generation. The province faces the challenge of identifying opportunities and optimizing the use of limited land and water resources for crop, livestock and forestry production. Land use planning needs to take into account not only of the biophysical conditions but also the socioeconomic circumstances and policy environment. The province needs to consider different possible options under various development objectives, and understand the trade-offs among different scenarios of resource allocation to meet these objectives. In this paper, we describe the use of a land use planning and analysis tool for exploring land use options for Bac Kan province, and illustrate the potential of this tool to assist policy makers in decision support for resource allocation to meet certain development objectives. MATERIALS AND METHODS The Land Use Planning and Analysis system (LUPAS) is a methodology developed for exploring options for land and resource use allocation, using a combination of resource evaluation methods with interactive modeling using multi-goal linear programming approach (Roetter et al. 1998). LUPAS is linked with geographical information systems (GIS) to provide an explicit spatial dimension as well as facilitate integration and analysis of biophysical and socioeconomic data. LUPAS is organized to three main components (Figure 2): (i) evaluation of the resources required for agricultural development, (ii) estimation of inputs and outputs for production activities, Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 2 and (iii) optimization of land and resource allocation subjected to policy views and development objectives. 1. Resource evaluation Resource evaluation is fundamental to the LUPAS methodology. It is carried out to assess if the existing or intended use is appropriate or can be supported by the resource base, or conversely for what kinds of uses the land is suited or has potential for. It is necessary to define or delineate spatial units of land for resource evaluation. Qualitative resource evaluation is carried out in Component (i) based on biophysical conditions. This is followed by a more quantitative analysis in Component (ii), taking into consideration quantities of resources and outputs, and the socio-economics of production. 1.1 Qualitative evaluation The main task in qualitative evaluation is to match the qualities of the land with the requirements of the intended land use type. The land qualities are influenced largely by environmental factors such as climate, topography, soils, hydrology and constrained to some extent by present land cover/land use. All these factors vary across space and time. Particularly given the rugged terrain of Bac Kan province, the spatial variability of soil and hydrology is very high, and the limited arable land is also highly fragmented. These conditions lead to a high diversity of existing and promising land use types (LUTs). In determining the requirements of the LUTs, we need to take into consideration Different crop types, e.g. rice, maize, citrus trees. Crops grown in different seasons, i.e. in spring, summer and winter Combinations of crops, i.e. the number and kinds of crops grown in a year. When matching the LUT requirements with the land qualities we need to take into consideration not only spatial but also temporal variability of the resource base. The temporal variation is influenced by climatic and hydrological factors. Because of the complex topography of the province, we use the raster GIS data structure to capture this spatial heterogeneity rather than attempt to delineate large homogeneous land units (LUs) in the vector mode (Kam and Hoanh, 1998). The qualitative resource evaluation was carried out in a number of steps, as described briefly below. The methodological details are described in Kam, et al., (2000). We first considered the time-invariant factors, i.e. match the terrain and soil conditions against the requirements of particular crop types (e.g. maize, rice, potato, etc.). Thus the slope-soil-crop suitability was determined for selected crop types at each cell in the raster representation of the province. Grid cells having similar slope-soil-crop suitability for the selected crop types were then clustered into land units (LUs). The LUs were defined within land administrative units, in this case the commune, so that socioeconomic data collected at commune level would be incorporated at the quantitative evaluation stage. In the next steps of the qualitative evaluation, the dynamic factors are taken into account by: Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 3 a) matching, for each LU, the climatic (seasonal temperature and, for non-irrigated conditions, rainfall) characteristics with the requirements of the crop-season to determine crop-season suitability, e.g. a LU may be more suitable for summer rice than for spring rice because of temperature constraints; and b) determining, for each LU, the combined suitability of LUTs, which are specific combinations of crop-seasons. A LUT may consist of only one crop (e.g. summer upland rice only, or a perennial crop), or have combinations of two or three crops grown in different seasons of the year. For example a two-crop LUT may comprise summer rice followed by winter vegetables; a three-crop LUT may be spring rice followed by summer maize and winter potato. The approach that we developed for this study constitutes a modification of the FAO land evaluation procedure (FAO, 1976; 1993), whereby the LU is delineated first, taking into consideration climatic, terrain and soil factors, and then suitability evaluation is done for crop types of interest. Our approach overcomes a number of limitations of the FAO approach, particularly in dealing with multiple crops in different seasons and with high spatial heterogeneity of environmental factors. 1.2 Quantitative evaluation At this stage of resource evaluation, we determined quantities and values of main- and by-products that can be obtained, assuming certain levels of technology used, as well as corresponding levels and costs of inputs (including agricultural chemicals, labor, water and transportation). Data and information on inputs and outputs for each crop type at low and high technology levels were collected from field surveys and various research findings, and from consultation with local experts. Using these data, input/output tables were constructed for all LUs for existing and promising LUTs. These tables, which encapsulate quantified outcomes of the resource evaluation, are fed into the LUPAS optimization model. 2. Land use Optimization In the land use optimization process, the following steps are implemented: (a) determine the land use objective functions, (b) determine resource constraints to land use, and (c) find optimal land use options for user-defined objectives by using a multiple goal linear programming model. In step (a), we translated development objectives into objective functions for the model. Some development objectives, as articulated by key stakeholders in Bac Kan province (including provincial government authorities, line agencies concerned with resource management such as provincial and district agriculture departments), are as listed below Increase agricultural production through intensification and diversification; Increase regional income not only from agriculture, but also forestry sector; Minimize encroachment of agricultural activities into forest land; and Intensify agriculture in the lowlands, thereby relieving pressure and minimize soil degradation in the fragile sloping lands. In step (b), we determined the underlying assumptions about resource availability, e.g. constraints/limits on the quanta or sharing of land, water and labor Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 4 resources and of financial capital. The assumptions made were based on information gathered about existing situations (e.g. fiscal policies, institutional arrangements) and future development plans (e.g. for infrastructure development, investment schemes). Each, or combinations, of these assumptions when applied to run the optimization model in step (c), would generate different scenarios of resource allocation that optimizes (i.e. maximizes or minimizes) the objective function of interest. Each scenario result gives the selected LUT(s) assigned to each LU for the optimal solution. Since the results are computed for each LU, it is possible to display the results in the form of maps by linking the optimization model with GIS. Post-optimization analysis is carried out to interpret the results of the scenarios modeled, and to determine the implications of each scenario. 3. The use of terrain, soil and climatic data in resource evaluation A recent land use map of Bac Kan province at 1:100.000 scale (GDLA, 1998) was used to mask out areas that are physically and legally unavailable for agriculture such as rocky mountains, water bodies, settlements, and forest areas designated for special purposes (national parks, protection forests and special purpose forests). All subsequent analysis was carried out only for areas that are not masked, which constitute 56% of the total for the province. A digital elevation model of Bac Kan province (Figure 1) was used to generate a slope map, which was used in conjunction with a rasterized soil map (NISF, 1973) to determine, for each 250-m grid cell, the suitability for 18 crop types based on slope and soil requirements. The weighted linear combination method (Sys, et al., 1993) was used to match the slope and soil requirements for each crop type. The weight assigned to each factor for each crop type reflects the relative importance of the factor in influencing crop performance and yield. A total of 2,506 LUs were formed within 122 communes, based on slope-soil-crop suitability. Gridded surfaces of average monthly temperature and rainfall were generated by spatial interpolation of weather data obtained from 12 stations within and surrounding Bac Kan province. Areas presently under irrigation were estimated from existing lowland rice area. These data were used to determine the suitability, at each LU, for 27 crop-seasons, which were in turn used to determine the combined suitability for 42 LUTs. Of these 42 existing and promising LUTs, 10 are three-crop, 18 are two-crop and 14 are one-crop LUTs (of which 6 involve annual crops and 8 involve non-annual crops). RESULTS & DISCUSSION Resource evaluation Much of Bac Kan province is composed of mountain ranges and plateaus (Figure 1). The valleys are narrow and occur in pockets; the broadest valley runs in a general northwest-to-south direction down the center of the province. These valleys are associated with narrow strips of alluvial soils (Figure 1). The gentlest slopes are associated with the valley bottoms and the plateaus (the most extensive one is located in the southwestern corner of the province in Cho Don district). Under present land use Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 5 conditions, cropping is most intensive in the narrow valleys where the soils are thicker and more fertile and water availability is good. Figures 3(a) to 3(e) map out the suitability for selected crop-seasons. Larger areas are suited for summer rice than for spring rice because of the more favorable temperature and rainfall conditions in the summer months for rice cultivation. The limited LUs suited for spring rice are confined to the valley floors where temperatures are warmer and irrigation water is available in the early part of the year. Comparatively, more areas are suited for maize and sweet potato than for rice. The spatial distribution of areas suited for spring maize differs partly from those suited for spring tobacco and winter sweet potato; the difference is most marked at the eastern and western parts of the province. The suitability rating for winter sweet potato is generally higher for most of the suitable LUs than that for spring tobacco. Figures 4(a) to 4(c) map out the results of the qualitative land evaluation, based on LUs that are suited for one-, two- and three-crop LUTs respectively, the map legend indicating the number of different LUTs that are suitable at each LU. There are only limited LUs that are suitable for three-crop LUTs, and these tend to be confined to the valleys where irrigation is possible (Figure 4a). The LUs in the northeastern corner (shaded orange) are suited for a smaller number three-crop LUTs because of the colder temperatures (dropping from 20 to 11oC) and lower rainfall (ranging from 20 - 50mm per month) there during the winter months of October to January. Not surprisingly, there is a much high number of LUs that are suited for one-crop LUTs, and the LUs that are suited for a greater variety of one-crop LUTs (up to a maximum of 7) are those in the valleys and the plateau areas (Figure 4c). Figure 4b shows that LUs suited for high numbers of two-crop LUTs (shades of green) show a distinct Y-shaped spatial pattern traversing the center of the province. The western arm and the stem of this Y-shape correspond with the most contiguous and broadest valley described above. The eastern arm corresponds with northeastern corner of Ngan Son district. There also appears to be a distinct difference between the eastern and western mountainous parts of the province separated by the Y-shape. The LUs in the eastern part are suited for higher number of two-crop LUTs (as shown by the pinkish magenta and yellow colors) compared with the western highlands and plateau (bluish magenta and red colors). The plateau area in the southwestern corner in Cho Don district is suited for a low number of two-crop LUTs despite the gentler slopes and more favorable winter temperatures compared with the northeastern corner. The results of the qualitative resource evaluation are intermediate outputs of the LUPAS methodology; nevertheless they are informative for assessing the potential of different areas in the province for particular land use types. Interpretation of model results The MGLP model generates the optimal allocation of resources (land, water, labor and capital) to achieve user-specified land use objectives, under certain assumptions about the resource availability and technology application. The results indicate which LUTs are selected in each LU, how much resources are needed, and the production and incomes expected. Since the population is irregularly distributed in this mountainous province, two scenarios in labor availability were considered: (i) labor force in each commune is only used for land use in its territory; and (ii) labor force can Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 6 be shared among communes. The possibility of labor sharing among communes is determined based on the accessibility with the help of GIS. Figure 5 shows the trade-offs between two objectives of food production and annual regional income. The curves lift upwards from (a) to (f) with the liberation of more resource constraints. When land is the only constraint, the optimal curve (f) is highest, while with all constraints in land, capital, water and labor without sharing, the optimal curve (a) is lowest. The current situation (white star) represents the present focus on food production by local people at present. The distances between the curves (a), (b) and (d) clearly indicate that availability of capital and water are two major constraints. The change in direction of curves from (a), (b) and (c) to (d), (e) and (f) shows that when water and capital constraints are removed, food production increases faster than regional income. A large gap between curves (d) and (e) compared with minor gap between curves (b) and (c) indicates that labor sharing would have significant effects only if water and capital constraints are overcome. Figure 6 shows the main LUTs under two extreme scenarios when land resource is the only constraint: maximize regional income without requirement of food production, i.e. point I in curve (f), and maximize food production without paying attention to regional income, i.e. point F in curve (f). In the first scenario, suitable fruit trees and two- or three cash crops are selected, while in the second, obviously food crops as rice and maize predominate. CONCLUSIONS Conventional soil-crop suitability analysis stops at determining if particular crops can be grown on certain soil types. In this study we take suitability analysis further by considering not only the season for growing the crop but also the combination of crops to be grown in the three main planting seasons in the year (i.e. the land use types). In addition, the quantification of inputs and outputs for different LUTs at different technology levels takes into consideration socio-economic factors. Finally, the use of the information gathered in an interactive land use planning tool extends the utility of suitability analysis and land evaluation for exploring optimal allocation of resources under different development objectives. Applied to a geographical area where the driving factors for land use are highly variable in space and time, we have had to develop methodologies for data processing and analysis that would capture this heterogeneity. The results of the qualitative resource evaluation show a distinct spatial association of LUs suited for more intensive cultivation with areas of distinct terrain, soil and hydrological characteristics. The model results also show that removal of the most serious constraint, i.e. capital, would substantially improve the capacity to increase food production and regional income. Under the present status of limited capital within the province, improving water availability would enable higher levels of food production and regional income to be attained. In the absence of capital constraint, labor would emerge as the next most constraining factor; hence sharing of labor resources among communes would further increase food production and regional income. Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 7 REFERENCES FAO, 1976. A framework for land evaluation. FAO Soils Bulletin No. 32, Rome, Italy, 72 pp. FAO, 1993. Guidelines for land use planning. Interdepartmental Working Group on Land Use Planning. FAO, Rome, Italy, 140 pp. General Department of Land Administration (GDLA). 1998. Land use map of Bac Kan Province, scale 1:100,000. Institute of Land Survey and Planning - General Department of Land Administration, Ha Noi, 1998. Kam, S.P. and C. T. Hoanh. 1998. Implementing GIS in a decision support system for analyzing the balance between rice supply and demand. Paper presented at the GISDECO conference in Pretoria, South Africa, October 1998. 12 pp. Kam, S.P, B.T.Yen and C.T. Hoanh. 2000. Land evaluation for optimizing land and resource use for agricultural production in a heterogeneous mountainous environment - a case study of Bac Kan province, Vietnam. IRRI.GISDECO,2000. National Institute for Soils and Fertilizers (NISF). 1973. Soil map of Bac Thai Province, scale 1:100,000. Soil Research Group - National Institute for Soils and Fertilizers in collaboration with Soil Research Group of Department of Agriculture - Bac Thai Province (1963 - 1967). Agricultural Publishing House, Published 1973. Roetter, R. and C.T. Hoanh. 1999. Exploring land use options under multiple goals in support of natural resource management at sub-national level. In: NN Kinh, PS Teng, CT Hoanh and JC Castella (eds), Towards an Ecoregional Approach for Natural Resource Management in the Red River Basin of Vietnam. Ministry of Agriculture and Rural Development, Hanoi, Vietnam and International Rice Research Institute, Los Baños, Philippines, 29-57. Sys, C., E. Van Ranst, J. Debaveye and F. Beernaert. 1993. Land Evaluation Part III Crop requirement. Belgium General Administration for Development Cooperation. Agricultural Publications No. 7. Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 8 N W Bac Kan province Vietnam E S Ba Be Ngan Son Bach Thong Cho Don Na Ri Bac Kan Soil types Lo¹ i ®Êt Ký hiÖu Ld FH FQa FQj FQk FQp FQq FQs FQv Fa Fj Fk Fp Fq Fs Fv Ao,Hå J Ld Ldk Lf Luc Luk P P' Pb Pi Elevation, m Cho Moi Figure 1: Location, digital elevation map (superimposed with district boundaries) and soil map of Bac Kan province. 2. Resource evaluation Delineation of land units, resource assessment, land suitability analysis & yield estimation Constraints concerning resources: area, water, labour & capital 3. Land use optimization Input/output tables of production activities Optimization Model using MGLP Land use options & Goal achievements Data Maps Objective functions Policy views 1. Scenario construction Figure 2. The GIS-based Land Use Planning and Analysis System (LUPAS) S1: Highly suitable S2: Suitable S3: Marginally suitable N: Not suitable a. Spring Rice b. Summer Rice c. Spring Maize d. Spring Tobacco e. Winter Sweet Potato Figure 3. Suitability for selected crop-seasons Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 9 a. Three-crop LUTs b. Two-crop LUTs c. One-crop LUTs Figure 4. Suitability for one-, two- and three-crop land use types (legend shows the number of 1-, 2- and 3-crop LUTs that are suitable at each land unit) Tradeoff between Food production & Regional income 800 I Only land constraint 700 f Regional income (bil VND) Land & sharing labor 600 d e Land & labor w/o sharing 500 Land, capital & sharing labor b 400 a Land, capital & labor w/o sharing c F 300 Land, capital, water, labor w/o sharing 200 Current situation of food production & regional income Current 100 0 50 100 150 200 250 300 350 Food production (thousand tons) Figure 5: Optimal lines of food production and regional income in different scenarios GOAL: Maximize regional Income CONSTRAINT: Land resource Lut_21 Legend 0_Apricot_0 0_Citrus_0 0_SoybM_MaizeW MungbS_RiceM_0 SoybS_RiceM_PotatoW VegS_VegM_VegW GOAL: Maximize food CONSTRAINT: Land resource Legend Lut_11 RiceS_RiceM_MaizeW RiceS_RiceM_PotatoW RiceS_RiceM_VegW MaizeS_RiceM_MaizeW MaizeS_RiceM_PotatoW RiceS_RiceM_0 MaizeS_RiceM_0 0_RiceM_SPoW Figure 6. Maps of model outputs for two scenarios Vietnam Lupas paper at AEZ-LADA Workshop, Bangkok, Thailand, 10-14 Nov.2003/PQha 10