Moslem Darvishi; Abouzar Ramezani
Abstract
Extended Abstract Introduction Due todecreased rainfall and increased groundwater harvesting, our country faces drought. With drastic decline of water levelin lakes and hydroelectric reservoirs, water scarcity is deeply felt. Thus, managers and officials shall find new ways of decreasing waterconsumption ...
Read More
Extended Abstract Introduction Due todecreased rainfall and increased groundwater harvesting, our country faces drought. With drastic decline of water levelin lakes and hydroelectric reservoirs, water scarcity is deeply felt. Thus, managers and officials shall find new ways of decreasing waterconsumption and overcome this crisis. Due to the rising global temperatures and reportsof the World Wildlife Fund, water scarcitycrisis will dominate most countries of the world, especially in Europe and Asia in the next ten years (Sengupta, 2018). Therefore, advanced water management principles shall be applied to decrease water consumption in the agricultural sector and maintain water security. Iran is among the top five countries of the world in terms of having vast irrigated land (Bruinsma, 2017), which shows that in many parts of the country agricultural lands are irrigated. Thus, the country’s water resources reach a critical stage, and because of limited resources, no more water can be provided for agriculture. The present study primarily seeks to optimize crop cultivation using two approaches: first, reduce water consumption and increase farmers’ income and second, reduce water consumption and meet domestic demand. In order to achieve this goal, first, the type of crops and area under cultivation were determined using remote sensing and satellite imagery. Then,spatial information system was used for data analysisand optimization of crop cultivation. Materials & Methods Remotely sensed images were used to collect data about the area under cultivationin agricultural patches and crop type. Those images were then analyzed using remote sensing techniques.According to pixel-based classification ofmultitemporal satellite images using training data, a croplabel is assigned to each pixelin this method. Moreover, borders of each agricultural land are extracted from pan-chromatic images of the region with higher spatial resolution. Finally, fitting the results of pixel-based classification with the extracted bordersof each agricultural land,a final croplabel is determinedfor the total area of the agricultural landbased on the majority labels. In order to optimize the problem, two objective functions (relationships 1 and 2) are defined in which income maximization and water consumption minimization are considered. Typically, location and allocation problems include objective and constraints functionswhich are maximized or minimized based on the goal of the problem. Linear programming is used to solve the problem. Linear programming is a classical optimization method whichdevelop a deterministic algorithm tosolve the problem. This method can only be used when the relationships between variables are linear. In other words, the relationship between variables shall be perfectly proportional and directin this method. (1) (1) (2) Result &Discussion The study area consists of 198 hectares of agricultural land in vicinity of GolangTapeh village of Asadabad city. The city covers an area of 1195 km2 and constitutes 6.1% of Hamadan province. It is located between 34° 37› to34°50 ‹northern latitude and 47°9› to 47°51›eastern latitude. Its average height is 1607 meters above sea level. The city is bounded in northwest with the province of Kordestan,in west with the province of Kermanshah, in southeast with Tuyserkancity and in the northeast withBaharcity. Assad Abad consists of three plains and a mountainside, but since it mostly consists of fertile plains, it can be considered as a flat area (Fig. 1). Fig1: Case study area Figure 2 shows the results of pixel-basedclassificationusing neural network method. In this method, network is trained using ground data. After training the network on the basis of ground truth estimator data, the estimation accuracy is about 88%. Fig. 2: The results ofclassification using neural network Following the calculation of the area under cultivation in agricultural lands and the type of crops, optimization is investigated using two scenarios (Figure 3). In the first scenario, reduction of water consumption and increased farmers’ income and in the second scenario,meeting domestic demandsto prevent capital outflow is considered. Fig3: Crop type and boundaries of agricultural lands In the first scenario, our priority is to reduce water consumption and increase farmers’ income. In this scenario, the goal is to select the type of crops according to the modeling constraints so that the crop type and water consumption are optimized. Figure 4 shows the proposed crop type. Fig4: The results of thefirst scenario Conclusion The present study used a combination of remote sensing and spatial information system to find a solution for optimization ofcultivation pattern through two different scenarios. First, land boundaries and types of crops were determinedusing pan-chromatic images and artificial intelligence. Then, two objective functions were developed to minimize water consumption and maximize income. Also, constraints such as crop type, periodicity constraints and domestic demand were modeled. Considering two objective functions, an algorithm was presented to optimize the cultivation pattern and the results were implemented in this algorithm. Results indicated that the difference between the first scenario which seeks to minimize water consumption and maximize farmers’ income and the second scenario which seeks tominimize water consumption and maximizedomestically demanded crops is relatively small. In both scenarios, the water use rate inAsadabad plain have decreased by about 1000 m3. In other words, in both scenarios there was a 50% reduction in water consumption. Moreover, if priority is given to meeting domestic demand, water consumption increase by about 3% and income decrease by about 3%. In future studies, owners of each agricultural land can be determined and each farmer’s incomecan be considered to further optimize crop cultivation.
