Predict possible change in land use by using satellite imagery and CA- Markov model

Background and objectives: Soil is one of the main non-renewable natural resources that today its destruction is considered one of the most severe problems all over the world. In recent decades, rapid and unsustainable changes and land use due to the urban development activities and increasing population created a great deal of modifications in land cover and land use and has been increasing the environmental degradation including soil degradation. Therefore, reviewing these changes through satellite images, evaluating and forecasting their potentialities via modeling can help managers and planners to make more effective decisions. The aim of this study was to assess the changes in land use during the period 1992 to 2015 via using satellite images to calculate the rate of land use changes to each other and predict possible changes in land use in the years 2020, 2025, 2030 and 2035, using cellular automata - Markov model (CA-Markov) in Joupar plain, Kerman province.
Materials and Methods: In order to prepare land use plans the three periods of Landsat satellite images including Landsat 5 satellite TM (1992), Landsat 7 ETM + (2000) and Landsat 8 satellite OLI (2015) were used in this study. To prepare land use maps through satellite images, initially the mentioned images were exposed to primary pre-processing such as geometric and atmospheric corrections. In addition, via providing training samples the satellite images were classified and their accuracy were evaluated using Idrisi imagery software through maximum probability algorithm. The developed land use maps of different periods were transited to CA-Markov model in order to produce transition probability matrix. Ultimately, the transition probability matrix was produced that shows the likelihood of transition of one land use to others. Then the chain analysis of cellular automata – Markov on the basis of land use plans and transition probability matrix in CA-Markov model with an emphasis on land use changes were expected in 2020, 2025, 2030 and 2035 were implemented in Idrisi software with various numbers of repetitions and steps. Based on the survey results, changes in land use and the level of current land use changes calculated, compared and evaluated and the future land use changes were predicted.
Results and Conclusion: The results of the detection of changes in the first period (1992-2000) revealed the highest increase in land area which was attributed to the use of pasture, grassland, irrigated agriculture and orchard and the highest decrease in land area was related to bed stream. In the second period (2000-2015) the greatest increase in land area was associated to the use of irrigated agriculture, orchards and bed stream and the greatest reduction was in pasture and grassland use. The results obtained from the prediction of future user changes of the region based on CA-Markov showed decreasing levels of land use attributed to orchard and irrigated agriculture and increasing levels of land use associated to pasture, grassland and bed stream comparing to 2015. Also the results obtained from the prediction of the findings regarding the years 2020, 2025, 2030 and 2035 revealed a reduction in land use related to bed stream, pasture and grassland due to the lack of rainfall and temperature rise and this will lead in the destruction of vegetation cover as wellas the more soil degradation. Also, due to the lack of rainfall, the recent droughts and previous studies we can conclude that the approach of CA-Markov model is more compatible with the conditions of the region.