Elham Yusefiroobiat; Fatemeh Jahanishakib
Abstract
Introduction: Nowadays, detection and modeling land use changes using satellite imagery is a useful tool for understanding future environmental changes associated with human activities. Monitoring these changes will help us understand the development process in the past and future patterns. Land-cover ...
Read More
Introduction: Nowadays, detection and modeling land use changes using satellite imagery is a useful tool for understanding future environmental changes associated with human activities. Monitoring these changes will help us understand the development process in the past and future patterns. Land-cover change models are important tools for analyzing the causes and consequences of shaping and expanding land uses for a better understanding of the performance of land cover systems and management and identifying sensitive areas. But applying the predicted patterns requires validation and correction of cases that the model can’t predict. In this research, using satellite imagery processing and Cellular Automated Markov chain (CA-Markov) model, agriculture and urban land use changes of Birjand watershed were modeled and predicted in 1404. Material and methods: In the present study, first land use changes were revealed and modeled using Landsat 7 in 2000, and Landsat 8 in 2014. Then, using the CA-Markov Model, land use changes in 2014 were predicted and modeled. To validate the modeling method, the consistency and inconsistency between the predicted map and the classified map were estimated on different kappa (Kstandard, Kno, Klocation, and KlocationStrata) coefficients. Validation of the changes in 2024 was predicted with high relative validity. Finally, by identifying the main drivers of developments, four scenarios of development were developed. A probable scenario based on population growth and the required area was selected among them. Results and discussion: The results of this research showed the detection, validation, prediction and correction of the model by scenario analysis. The increase in agricultural and urban lands will be 0.525 and 18.9 km2, respectively. Validating with an accuracy of over 98%, the simulation allowed prediction of future land use changes in 2024. From different scenarios, the probable scenario with an occurrence probability of 70% of the forecasted changes (scenario 3) resulting from the CA-Markov was selected according to the documentations and experts' opinions. Also, a comparison of two maps in different units resulted in a trend that by increasing the comparison units and coarse grain, the amount of the disagreement would go further towards the agreement. It is noted here that the Klocation in the cell, KlocationStrata, and Kno had the same numbers, and different from the Kstandard. Conclusion: According to the results, the spatial dimension of urban development in the north of the city was correctly identified. At the same time, the level of agricultural and urban-rural changes was less predicted. In the case of agriculture land use, this lower prediction was due to the construction of urban sewage treatment and in the case of urban land use, this difference can also be attributed to different urban growth in different periods. Also, despite the credibility and accuracy of prediction, some of the main drivers of development have no predictability by the model in the future. Therefore, it is suggested that research in predicting changes, in addition to validating the modeling approach, not only satisfy the final results, but also modify the results of the model by taking into account development drivers.
