Predicting Land Use Change Using Markov- Cellular Automata Model in Badavar-Nurabad Watershed, Lorestan

Accurate detection of changes land use in Accurate and timely, Basis for a better understanding of the relationships and interactions of human and natural phenomena to manage and provides better use of resources. Principal land use management requires accurate and timely information in the form of a map. Regarding the widespread and unsustainable changes in land use, including the destruction of natural resources in recent years, Investigating how landslide changes during time periods are essential for satellite imagery. Since conservation of natural resources requires monitoring and continuous monitoring of an area, Land-use change models are now used to identify and predict land-change trends and land degradation one of the most widely used models in predicting land use change is the Auto-Markov cell model. the aim of present study is to monitor land use changes in the past years and predict changes in the coming years in Badavar-Nurabad watershed in the Lorestan province with an area of 71600 hectares.

The Markov chain method analyzes a pair of land cover images and outputs a transition probability matrix, a transition area matrix, and a set of conditional probability images. The transition probability matrix shows the probability that one land-use class will change to the others . The transition area matrix tells the number of pixels that are expected to change from one class to the others over the specified period (Ahadnejad 2010). Automatic cells are models in which adjacent and continuous cells, such as cells that may include a quadrilateral network, change their state or attributes through simple application of simple rules. CA models can be based on cells that are defined in several dimensions. The rules for changing the state of a cell from one mode to another can be either a combination of growth or decrease, such as a change to a developed cell or without development. This change is the source of the change that occurs in the adjacent cell. Neighborhood usually occurs in adjacent cells or in cells that are close together(Ghorbani et al, 2013).  In order to detect land use changes in the studied area, TM , ETM+  and OLI satellite images of Landsat were used during three time periods of 1991, 2004 and 2016. After applying geometric and atmospheric corrections to images, the land use map for each year was prepared using the maximum probability method. The Kappa coefficient for the classified images of 1991, 2004 and 2016was 0.81, 0.85 And 0.90 obtained. Then, to model land use changes using the Auto-Markov cell model for 2028 horizons, First, in the Idrisi Selva software using Markov chain, the map was selected as input from the years 1991 and 2004, the 12-year prediction of the changes was considered by 2016 to determine the likelihood of a change in application. Then, using the CA-Markov method, the data from the Markov chain and the map of 2016 were used as input data for the automated-Markov cell method.

Assessment of the match between simulated and actual map of 2016 with 0.97 kappa index showed that this model is an appropriate model for simulating of land use change. The results from monitoring satellite imagery that in 1991 to 2016, the extent of residential areas, land is Dry farming, garden and irrigated farming land added in front of vast pastures, shrubbery and other is reduced. After verifying the model's accuracy, a 2028 map was prepared to predict the changes over the coming years. Well as the results show that the vast pastures of the forecast is reduced in the amount of 659.89hectares and 395.47 hectares will be added to the extent of irrigated farming.

The results of the Auto-Markov cell model showed that if the current trend continues, the size of the ranges will decrease sharply. Comparison of simulated map of 2016 by model and actual map with Kappa index showed that Auto-Markov cell model is a suitable model for predicting land use change and can accurately assess the future status of land use and vegetation to predict. Therefore, it is suggested protective measures and make appropriate management decisions to control non-normative changes continue to apply more than ever.