Applying object-oriented processing techniques for investigating land use/land cover changes and predicting future changes (Case study: Miandoab, Iran)

Abstract

Assessing the dynamic changes in land use/land cover in the management and sustainability of natural ecosystems is important. This study aims to investigate the ability of object-oriented processing of satellite images for land cover detection and prediction of future changes using CA Markov model in the township of Miandoab. For this purpose, land cover maps were obtained for the years 1984, 1994, 2004, 2014, and 2018 using the threshold setting method in eCognition 9.1 software and Landsat satellite images of TM, + ETM and OLI. The CA-Markov model was used to predict future changes for 2032. The accuracy of the forecast model was obtained by matching the predicted map of 2018 with its real value. This value represents the validity of modeling results. The results showed that during the period of 34 years (1984-2018), gardens and human-made lands have been improved significantly, so the area of the gardens grew45km between 1984-2018 and human-made lands increased from 6.21 km2 in 1984 to 27.55 km2 in 2018. In recent years, compared to previous years, the water bodies have diminished, and if the management and resources planning are not altered in the region, this trend of change will have irreparable consequences soon.