چكيده لاتين :
Introduction
Accurate quantitative evaluation of runoff problems is essential for proper rehabilitation designs because it can help to minimize the harmful consequences. Given the importance of accurate estimation of runoff in watershed management and design of hydraulic structures, different methods have been proposed for the runoff estimation in the watersheds without measurement station. Regardless of the method, nowadays most researchers have considered the application of GIS (Patil 2008). There are many models, such as WMS, SWAT, EPIC, AGNPS, MIKE 11, MIKE FLOOD, HEC GeoHMS, and Cellular automata, in the water and environmental studies which have been combined well with the GIS. A Cellular automata (CA) is one of the newest ones in this field. It is a mathematical model that can be used for computation and simulation of the systems. In this method, the basin is defined with a network of the rectangular cells, and the interactions between the cells together with the geographic rules that govern the area result in the runoff modeling. This model relies on the GIS and satellite images. Cellular automata model uses various information such as digital elevation model (DEM), land use, hydrologic soil groups, rainfall, slope and etc. It has been applied in several studies, such as Van et al (2007); Rinaldi et al (2007); Cirbus and Podhoranyi (2013); Shao et al (2014), for the runoff estimation. In the present study, the runoff of the Lighvan Chai in East Azarbaijan province, Iran, has been modeled by means of the Cellular Automata. GIS software have chosen Python because of the simplicity, high capability, being object oriented, and being compatible with the GIS environment. Comparison of the results with the observation in Hervi and Lighvan stations by means of the efficiency criteria proves that the results are well accurate.
Besides to the advantages of this method in simplicity and implementation of the realistic rules, this method is good at gaining the runoff data at any point of the basin besides to the exit point. It has been employed for gaining data of runoff time series of six points in the basin. Since the program!!! Is ready, it can be easily used for the runoff estimation in different time and position conditions, i.e. long term basin runoff or runoff due to the given storm precipitation.
Data and Methods
Cellular automata, CA, were first introduced by John Von Neumann and Stanislav Ulam in the 1940s. Cellular automata are discrete in space and time, continuous state and the behavior is speci?ed completely by rules governing local relationships (Cirbus and Podhoranyi, 2013). They are made through an attempt to simplify the often numerically intractable dynamic simulations into a set of simple rules that mirror the intuition and are easy to compute. As an approach for modelling the emergent properties of complex systems, it has the great bene?t of being visually informative about the progress of dynamic events. Basically, CA models use several primary components including the cells arranged in a regular mosaic pattern (raster, grid), transition rules determining the changes in cell properties, neighborhood of the cell, and boundary conditions. These components affect the status of each individual cell in a network in a given time step. In this research, Lighvanchai basin in East Azerbaijan province has been modeled using cellular automata. SCS formula is used to predict the runoff in each cell and D8 algorithm is used to simulate flow direction during the calculation of the surface convergence. The procedures for channel network delineation are based on the D8 model for flow over a terrain surface represented by a grid DEM. In this model a single flow direction in the direction of steepest slope towards one of the eight (cardinal and diagonal) grid cells neighboring is used to represent the flow field. Parameters necessary for model are prepared and modified in order to fit the need for the calculation of surface runoff. Optimal inverse distance weighted average method was applied to create a network of rainfall. The relationship between interpolation accuracy and three critical parameters of IDW has been evaluated: power (a value), a radius of in?uence (search radius) and angle of search neighborhood. A total of 19 rainfall stations and rainfall data between 1982 and 2009 were used in this study. The value of the radius of in?uence, and the control parameter (p) were determined by root mean squared error (Lu and Wong, 2008; Chen et al, 2012) to obtain optimal interpolation data of rainfall. Consequently, a power value of 2 was adopted for IDW in this study. According to the wind direction in region was determined angle of search neighborhood.
Map of hydrological soil groups of Lighvanchai basin was determined by means of soil texture map. Also, Land use map of Lighvan basin was extracted from the satellite images (ETM+ Landsat image on September 11, 2013). Land use and SHG maps are read first and the runoff curve number (CN) map is prepared for the normal conditions then. The soil has moderate moisture and the slope is not greater than 5%. Through a combination of SHG and land use and by SCS tables and then is adjusted for dry antecedent moisture condition and slope using the common relationship (Ponce and Hawkins, 1996; Huang, 2006; Hawkins et al., 2009). After reading the rainfall and the CN map, calculate the runoff depth was calculated using the SCS equation.
Runoff production within each cell was simulated by determining the cell state (water surface elevation) that included both the cell altitude and the water depth. The distribution of water flow among cells was determined by applying CA transition rules based on conservation of energy and continuity equations. As noted above, Due to simplicity, powerful, object-oriented programming language of Python and supported with GIS, from this language is used to implement rules for estimating runoff and its expansion.
Results:
The monthly time series in six points (cell) within Lighvan catchment was computed in this study. Two points of the six points were located in the places of Lighvan and Hervi stations, to be able to compare the results of the cellular automaton model with the observed data at these two stations. Comparison of the results with the observed values in Lighvan and Hervi stations using correlation coefficient performance, Nash-Sutcliffe and root mean square error indicates that the accuracy of the results is desirable. Correlation coefficient and Nash-Sutcliffe values are above 0.95 for Lighvan and Hervi stations which indicate that the accuracy of the obtained runoff from CA model was high.
Having more simple structure, application of more realistic communication rules and earning runoff information for each point of the basin except the output of the basin that is useful for ungauged watersheds are the main advantages of this method. According to the prepared program, runoff can be estimated easily for different circumstances of time and place so the long-term runoff from catchment or simulated instantaneous rainfall.
Conclusion:
In this study, it was primarily purposed to demonstrate the feasibility of using cellular automata for rainfall- runoff simulations in terrain it showed that, even a complex natural process such as runoff, can be estimated to some extent with the assistance of a digital elevation model and related raster layers. It also demonstrated that the grid cell can carry the information which can identify some of the natural processes.
Good agreement between the model output and the empirical measurements revealed that a CA approach can provide realistic results for a complex natural process like runoff.
