Analyzing the relationship between socio - Economic and environmental factors for building an integrated system supporting agricultural land use planning. A case study in Soc Trang province

investment capacity of each land use type to be arranged. CHAPTER 3. RESEARCH METHODOLOGY 3.1 Method for analyzing the relationship of socio-economic-environment factors affecting agricultural land use These kinds of data were collected for analyzing the relationship of socio- economic-environment factors affecting agricultural land use: Land use statistics and land use maps of 3 districts Long Phu, Tran De and My Xuyen in 2005, 2010 and 2015; Socio-economic statistics from 2010 to 2018; The annual agricultural reports of Long Phu, Tran De and My Xuyen districts from 2015 to 2018. Household survey to collect information on agricultural production, socio- economy and environment affected land use. The number of interviews was determined based on formula Yamane (1967). 𝑛 = 𝑁 1 + 𝑁𝑒2 (1) Where N: total agricultural households; e: sample error. 6 Number of agricultural households of 3 districts in the study area is about 55,000 households, sample error is selected as 6%, and thus, the number of samples n calculated was 276 samples. Considering the number of samples of similar studies by according to the district and province levels of Le Quang Tri et al., (2013); Thai Phu Vinh et al., (2015); Santiphop et al. (2012). The total number of interviews in 3 districts was rounded up to 45 households / LUT. Therefore, the total number of famers for surveying is 315 households. Descriptive statistical methods (Mann, 1995) was used to determine average values and standard deviations for quantitative economic indicator as profits of land use types; Qualitative social indicators include: education level, intensive farming level, production capital, technological and scientific transferring; farmer's risk assessment, infrastructure requirements for production; impact of agricultural land use types as environmental qualitative factor. 3.2 Methods of building the integrated model The integrated model, named ST-IALUP (Soc Trang- Integrated model for supporting Agricultural Land Use Planning), was a combination of various tools where input and output data were well connected. The integrated model provided agricultural land area estimation, agricultural land area optimization software - LandOptimizer, and land use allocation model - ST-LUAM. The principle of integration was shown in Figure 3.2. Figure 3.1 Integrated model ST-IALUP Figure 3.2 showed the connection of components of the integrated model named. Using Monte Carlo method is to estimate the land use areas of limited land use types. Estimated land use areas were exported to a CSV file containing the area of the LUTs in the predicted over years. This data source was used as constraint value in LandOptimizer. This software gives optimized area for each LUT and was connected to the land allocation model, ST-LUAM, which performs land use solution maps. 3.2.1 Agricultural land use area estimating In terms of estimating land use area, this thesis focused on analyzing three types of agricultural production: vegetables, fruit and aquaculture. Cultivating area was estimated based on Monte Carlo simulation method which was applied as the diagram 7 in Figure 3.3. In which, data of area of annual crops, fruit and aquaculture from 2010- 2010 were loaded into the model. Figure 3.2 Estimating the cultivated area using Monte Carlo method Next, historical of cultivated area was analyzed to get frequency of occurrence. This frequency data was normalized. For each LUT, model generate cultivated values of next years by generating a random frequency in the range [0, 1]. This random number was used to get the cultivating area according with frequency where area values were classified. After that, the model checks condition to stop simulation, if it is false the calculated area value will be returned to the list of area values the next simulation cycle. 3.2.2 Method for developing agricultural land use optimization LandOptimizer software was built using the programming language Visual Basic.Net on Windows operating system. The main steps of software development were shown in Figure 3.4. 8 Figure 3.3 Main construction’s steps of LandOptimizer software 3.2.3 Method of building land use allocation model After determined optimal area of each land use type (LUT) per each land unit, the thesis proposed a detailed land use allocation within each land unit. In particular, land use types are arranged into the cells inside a land unit map by the Cellular Automata method and multi-criteria assessment based on natural, socio-economic and environment factors as shown in Figure 3.5. Figure 3.4 Diagram for building land allocation model The economic investment capacity index of cells (IInvest): This index was assigned values from commune group. Communes are classified into three groups depending on the level of achieving New Rural construction Standards (NRS); Then standardize the commune group into 3 values [1; 0.5; 0]. These values were assigned to the commune boundary map. The distance index from a cell to the nearest road (IR) and canal (IC) is calculated by the shortest distance from the position of each cell to the nearest road (canal). In the model, buffer method was created layer by layer from the road (canals) to determine the distance of cells contained by the buffers. Distance values were also normalized to the range [0, 1] according to the maximum distance of a cell to the roads (canals). Ratio of density of land use type in neighborhood of a cell (IDEN_LUT). This index was determined by counting number of neighborhood cells of each land use types divided by 8. IDEN_LUT(i) = number_of_neighborhood _LUT(i)/8 (2) The allocation capability index for a LUT of a cell (Icap_LUT) was determined by the formula (3). A LUT was assigned into a cell when it had highest value of Icap_LUT. In case there are many LUTs with the same value of Icap_LUT, the LUT was randomly selected from these LUTs. Icap_LUT(i) =(WR.IR + WC.IC + WDEN. IDEN_LUT(i) + WI.IInvest )/( WR+WC+ (3) 9 WDEN+WI ) In which: WR, WC, WDEN, WI are weights for IR, IC, IDEN_LUT(i). These values will be identified in calibration process of the model. CHAPTER 4. RESULTS AND DISCUSSION 4.1 Introduction to the study area Soc Trang is located at the southern mouth of Hau River in the Mekong Delta (Mekong Delta), about 60 km from Can Tho city, with geographical coordinates from 9014’28 '' to 9055’30 '' North latitude; 105034’16 '' to 106017’50 '' East longitude. Bordering provinces: Hau Giang, Tra Vinh, Bac Lieu and the East Sea. In 2018, Soc Trang province consists of 11 administrative units (1 city, 2 towns and 8 districts): Soc Trang city, Vinh Chau town, Nga Nam town, districts: Ke Sach, My Tu, Cu Lao Dung, Long Phu, My Xuyen, Thanh Tri, Chau Thanh and Tran De. To keep the objective of the thesis, the study area was selected according to the criteria of contiguous districts with fresh, brackish and saline water ecological characteristics. This helps to survey specific land use for these ecoregions. Based on the selected criteria, the study area consists of 3 districts of My Xuyen, Long Phu, and Tran De in Soc Trang province. In particular, Long Phu is in a fresh water area but risk of being affected by saline intrusion during extreme weather events (such as drought and saline intrusion in 2016); My Xuyen belongs to brackish water region; Tran De is divided into two areas: the saline area at the mouth of the river outside the dike and the area of fresh water inside the dike systems. In order to analyze the changes in rice cultivation area over the years as a basis for predicting the development area for specialized cultivation, fruit and aquaculture, the statistics from year 2010 to 2018 for all 3 districts was shown in Figure 4.1. Specifically, the cultivated area of the districts concentrates only 2 crops in which the Summer-Autumn crop lasts for a large area of cultivation. The Summer-Autumn crop is included in the Summer-Autumn and Spring-Summer crops. (Source: Generated from Statistic year book 2012-2018) Figure 4.1 Cultivated area of agricultural land use of 3 districts from 2010 to 2018 Figure 4.1 shown cultivated area of vegetables, fruit and aquaculture in Long Phu, Tran De and My Xuyen districts from 2010 to 2018. In general, the area of vegetables and crops has fluctuated up and down but in the recent period, there was a 10 tendency increase continuously. In contrast, the area of fruit has the least variation and tends to range from 8,546 ha to 8,938 ha. Particularly, aquaculture area tends to increase continuously from 2010 to 2018. 4.2 Analysis of factors affecting agricultural land use 4.2.1 Selection of types of agricultural land use In this study, the agricultural land use selected for research include the ones for the fresh, saline ecology areas which are rice, crops, aquatic products, and fruit. The selected land use types are representative of the ecological regions of the three districts in order to have a basis for data survey. The basis for selecting LUTs was based on relevant studies in the Mekong Delta region and was being of interest by the Department of Agriculture and Rural Development of Soc Trang (2018). Prospective LUTs in 3 districts include: 3 rice crops; 2 rice crops (Winter-Spring-Summer Fall), 2 rice crops + 1 vegetables, Rice-Shrimp, Vegetables (2-3 crops), Fruit and Shrimp (2-3 seasons). 4.2.2 Analysis of economic factors affecting agricultural land use 4.2.2.1 Profit of LUTs Profit is one of the most concern criteria for the choice of agricultural land use types. Statistical results of the survey showed that most farmers wanted to choose the land use production with high profit. The statistical results described the total profit of the 7 uses are shown in Figure 4.2 in which the most profitable was shrimp farming, the two rice crops had lowest profit. Figure 4.2 Profit of land use types Figure 4.2 showed that there had huge difference in profits, especially between LUT7 (VND 277.23 million) and LUT2 (only about VND 42.42 million). However, in order to be able to implement LUT7, it is necessary to have not only capital but also intensive farming techniques as well as natural suitability conditions. 4.2.2.2 Investment capacity factors The results of the implementation, being recognized by new rural construction 11 (NRC) communes. In new rural construction criteria, there are many important indicators including income of households, poverty rate of communes. Therefore, this thesis uses NRC commune criteria as a qualitative factor affecting the disposition of agricultural land use. Based on these two criteria (income and poverty rate), the economic capacity of the communes is divided into 3 groups: Group 1 was the communes meeting NRC standards; Group 2 was the communes that have not met the NRC standard but have an average per capita income of VND 20-28 million and a poverty rate lest than 6%; Group 3 is the remaining communes (Table 4.1). Table 4.1 Group of communes based on economic capacity Dictrict Group 1 Group 2 Group 3 Long Phu Trường Khánh, Tân Thạnh, Long Phú, Song Phụng, Hậu Thạnh Long Đức, Châu Khánh, Tân Hưng, Phú Hữu Tran De Trung Bình, Lịch Hội Thượng, Thạnh Thới Thuận, Viên Bình Viên An Đại Ân 2, Liêu Tú, Tài Văn, Thạnh Thới An My Xuyen Hòa Tú 1, Hòa Tú 2, Ngọc Tố, Đại Tâm, TT Mỹ Xuyên Ngọc Đông, Gia Hòa 1, Gia Hòa 2 Tham Đôn, Thạnh Phú, Thạnh Quới No of criteria archived 19/19 11-18/19 ≤10/19 Average income per capita / year NRC qualified (greater than 30 Million VND) Not up to standard of NRC (20-28 Million VND) Not up to standard of NR (< 20 Million VND) Poverty rate NRC qualified ≤ 4% NRC qualified (≤ 6%) Not up to standard of NRC (≤ 23%) (Source: Generated from annual report of Long Phu, Tran De and My Xuyen) 4.2.3 Social factors 4.2.3.1 Number of labors The results of the survey for the number of labors for each agricultural land use type per year were shown in Figure 4.3. Among them, the vegetables needed the highest number of working days, followed by shrimp that farmers have to take care all of the year. Two rice crops – vegetable and fruit have the similar number of working days, equivalent to 115 and 121 days per hectare per year. 12 Figure 4.3 Number of workdays for LUTs during the year 4.2.3.2 Relationship between infrastructure and land use The survey results showed that local agricultural production was facing problems such as: (i) water supplied and drainage due to distance from canals, (ii) difficulty in transporting materials and travel due to narrow or unpaved roads, (iii) affecting farming practices due to saline leakage from shrimp ponds to rice fields. Particularly for aquaculture, besides natural conditions such as land and water, people still faced difficulties in production due to lack of electricity or weak operation of equipment, thus affecting production efficiency. Figure 4.4 Infrastructure requirements of LUTS The three LUTs such as two rice crops, three rice crops and rice-shrimp, are the types affected by neighbor land uses. Especially for rice-shrimp, if surrounding households raise shrimp or hold salt water in the pond, then neighboring rice-growing households will not be able to cultivate or achieve low yields. For shrimp and fruit farming, the most priority was strong electric power source to operate machinery, followed by the need to be located near the road as well as the influence neighbor LUTs. In fact, if they want to cultivate aquaculture products, the neighboring households must also cultivate the same style, which will bring about high efficiency. About 20% of people agreed that vegetables and fruits should be located near rivers and canals. However, vegetables and fruit needed to be located near the road due to the 13 farming behaviors of farmers. 4.2.4 Environmental factors affecting agricultural land use 4.2.4.1 Risk of LUTs Besides natural characteristics of land, risk factors in production of land use types that were assessed by people through four levels such as high risk, medium risk, and low risk and non-risk as shown in Figure 4.5. (a) (b) Figure 4.5 Risks of use types (a) and environmental benefits of LUTs (b) Risk factors include more or less uncertainty about productivity, prices and weather risks. Figure 4.5 showed that over 60% of the people assessed the shrimp farming style with highest risks in production. On the contrary, about 50% of people think of using two rice crops or two rice crops – vegetable provided low-risk or no-risk crops. In the case of fruit and vegetables, the average risk was assessed because the risk of farming depends only on the market, and the yield and weather do not usually affect these LUTs. 4.2.4.2 Environmental benefits of LUTs The analysis results show that the use patterns such as two rice crops, two rice crops-vegetable, rice - shrimp were assessed to be good for the environment. In contrast, three rice crops, shrimp and vegetables were evaluated as not good for the environment. At a level that was not good for the environment, vegetables has the highest rate of bad environmental among the LUTs. The shrimp was only assessed to be medium to the environment. The results of these assessments will be used to set goals in the application of agricultural land use optimization. 4.2.5 Summary of main factors affecting agricultural land use The specific socio-economic factors surveyed had different impacts on LUT. For applying these factors in building an integrated model to support land use planning, these factors are ranked based on the statistical results for 2 main purposes. The order of applicability of these factors was presented in Table 4.2. Each factor was considered to serve for one purpose: optimize land use area and allocate agricultural land use. Table 4.2 Summary of the influence of factors on agricultural land use 14 Factors Detailed Affecting LUTs and it orders Applied Economic Profit LUT 7, LUT 6, LUT 5, LUT 4, LUT 3, LUT 1, LUT 2. Optimization Capacity of investment LUT 7, LUT 6, LUT 5, LUT 4, LUT 3, LUT 1, LUT 2 Allocation Social No of labors LUT 5, LUT 7, LUT 3, LUT 6, LUT 1, LUT 4, LUT 2 Optimization Road systems LUT 5, LUT6, LUT 7, LUT 1, LUT 4, LUT3, LUT 2 Allocation Road systems LUT 5, LUT 6, LUT 7, LUT 4, LUT3, LUT 1, LUT 2 Allocation Neighboring LUT LUT 7, LUT 4, LUT 1, LUT 3, LUT 2 Allocation Environment Land suitability Based on Land suitability order Optimization Risk of LUT LUT6, LUT 7, LUT 1, LUT 5, LUT 4, LUT 3, LUT 2 Optimization Benefit of environmnent LUT 2, LUT 4, LUT 3, LUT 1, LUT 6, LUT 5, LUT 7 Optimization Thus, in the optimizing agricultural land use area software, factors such as natural adaptation, profitability, number of labors, risk level, and benefit level of environment were used. These factors were considered as the objectives of the optimization model according to single goal or multiple goals. The priority order of LUTs for each element was different, there are opposite situations. Therefore, the application of optimization model will help balance the impact of these factors on the overall results. For the allocation of agricultural land use, factors including investment capacity, transport infrastructure, canals and requirements of neighboring LUTs. The order of the LUTs was ranked based on the results of the survey analyzed and considered in the land arrangement. The combined results show that LUT 7 (Shrimp) and LUT 6 (Fruit) are prioritized to be located near roads, canals and rivers, where investment was possible, compared to other kind of land use. This feature was considering for two regions: For brackish areas, shrimp was prioritized to be arranged in priority areas, near roads, rivers and canals, in areas with investment potential, then it gradually spreads out, followed by rice - shrimp arranged. For fresh water region, vegetables and fruit are prioritized to be located near roads, canals and rivers and investment able areas, followed by rice - vegetable and rice (LUT 1, LUT 2). 4.3 Building an integrated ST-IALUP model The results in previous session showed the relationship of natural, socio- economic, environment factors in optimizing the area of land use and agricultural land use allocation. A newly integrated model called ST-IALUP (Soc Trang- Integrated model for supporting Agricultural Land Use Planning) was built. This integrated model was built with the following tools: (i) Agricultural production area estimating model used to determine boundary conditions of production areas that need for land use optimization; (ii) A new software to optimize the area of agricultural land use according to constraints related to natural, socio-economic and environment 15 conditions; (iii) Agricultural land use mapping model for LUTs that have been optimized. 4.3.1 Modeling area for agricultural production The input data for model collected that were production area of vegetables, fruit and aquaculture in 3 districts was collected from period 2010-2018. The simulation was repeated 10.000 times to determine the mean and the standard deviation from the replicate simulation data. Figure 4.6 shown the area of the 3 LUTs analyzed. Figure 4.6. Estimated area of vegetables, fruit trees and aquaculture products to 2030 The results of Monte Carlo analysis in Figure 4.6 showed that the area of vegetable land in 2030 is 14.868 ± 894 ha, the area of fruit is 8,799 ± 136 ha and the area of aquaculture land is 16.697 ± 2.540 ha. The mean area of vegetable and fruit did not change much. However, the standard deviation value of vegetables is about 900 ha and aquaculture area was more than 2.500 ha. These values and its range various ranges will be used as constraint values in optimization. 4.3.2 Developing software for agricultural land use optimization Based on the socio-economic factors affecting the land use area, LandOptimizer software was built to optimize land use area for each land mapping unit based on these factors. Source and packaged program have been uploaded in to website: https://github.com/nhthao/LandOptimizer 4.3.2.1 Designing the input for the software For the input data, depending on equation programming options, the input data includes: land unit maps, land use suitability of LUTs, profit, number of labor, environment benefit rate, and limited area for LUTs. The profit was based on land suitability level of LUTs in land units. Profit was standardized directly on LandOptimizer as shown in Figure 4.7. 16 Figure 4.7 Standardize profit on LandOptimizer The output data of LandOptimizer were in Excel format that showed the area of LUTs per land unit. For the most part, each land unit could have different LUTs. The results data sheet was also exported to CSV format as input data for the land use allocation tool. 4.3.2.2 Developing optimization objectives Optimizing with single goal The objective optimization function is set in the case of optimizing an objective such as adaptation or profit optimization. The objective functions in equations (4) and (5) are used in the case of adaptive optimization and profit optimization of land use types for each land unit. Maximizing Land suitability level ∑ ∑ 𝑇𝑁𝑖𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 → 𝑀𝑎𝑥 (4) Maximizing profit ∶ ∑ ∑ 𝐿𝑁𝑖𝑗𝑇𝑁𝑖𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 → 𝑀𝑎𝑥 (5) Multi-objective optimization function: The objectives were maximizing the profit, land suitability, number of local labor used, environmental benefit rate, minimizing risk of LUTs. 𝑤1 ∑ ∑ 𝐿𝑁𝑖𝑗𝑇𝑁𝑖𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 + 𝑤2 ∑ ∑ 𝑀𝑇𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 + 𝑤3 ∑ ∑ 𝐿𝐷𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 − 𝑤4 ∑ ∑ 𝑅𝑅𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 → 𝑀𝑎𝑥 (6) Where : i = 1..n, n: index of land units; j = 1..m, with m the index of LUTs Xij: Area of LUTj in land unit i. LNij: Profit of LUTj. in land unit i (unit: million VND / ha). LDj: the number of working days of LUTj. / ha. 17 MTj: Environmental benefit coefficient of LUTj.. Farmer’s assessments of environmental benefits of LUTs. RRj: Risk coefficient of LUTj. This is the LUTj. productivity risk indicator. The smaller the risk value gave greater contribution to the goal function. Wi: The weight of the objectives. In this study, the assumption of equal- weighted goals is set by default to 1 with the meaning that the goals in the multi- objective function have the same priority. These weights can be adjusted for increasing (moving toward 1) or decreasing (moving towards 0) depending on the priority of the local goals for the local development orientation. Constraint equations Constraint of the total area of LUTs per land use units must be less than or equal to the area of land unit (Inequalities 7). ∑ ∑ 𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 ≤ area of 𝑙𝑎𝑛𝑑 𝑢𝑛𝑖𝑡i (7) The total labor demand of the LUTs cannot exceed the local agricultural labor resources. ∑ ∑ 𝐿𝐷𝑗𝑋𝑖𝑗 𝑚 𝑗=1 𝑛 𝑖=1 ≤ Total working days (8) The minimum of each agricultural product to supply (system of inequation 9) ∑ 𝑋𝑖𝑗𝑁S𝑗 ≥ Minimum production of LUT k NSj: Yield of LUTs that provode the product k j = 1. . l ( LUTs that provide product k) k = 1. . p (type of products) (9) The maximum of each agricultural product to supply ∑ 𝑋𝑖𝑗𝑁S𝑗 ≤ Maximum production of LUT k Total area of LUTj <= Total limited area of LUTj ∑ 𝑋𝑖𝑗 𝑛 𝑖=1 ≤ Limited area of LUTj, j = 1. . m (10) (11) The limited area of specific LUTs was estimated in Section 4.3.1. Users can choose one of the pre-set plans to implement the corresponding integrated objective function. Figure 4.8 showed the options for choosing different goals and constraint options for building optimal model. 18 Figure 4.8 The interface of optimization on LandOpimizer The results were displayed in detail on the results box of the software including the area of LUTs per land unit, total profit, the productions of each product. Compare the results of LandOpimizer and the optimization model had been built on GAMS (Nguyen Hong Thao, 2007) for the case of study in Co Do District, Can Tho City. The results showed that the optimization results of LandOptimizer and the GAMS mathematical model are similar. 4.3.3 Developing agricultura