Investigation of classification of support vector machine on the estimation of the deciduous broad-leaved forests area in the north of Iran

Mohammadreza Tabesh; Seyed Armin Hashemi; Sahar Tabibian; Mohsen Abbasipour; Maghsood Hosseini

Investigation of classification of support vector machine on the estimation of the deciduous broad-leaved forests area in the north of Iran

2024

Mohammadreza Tabesh | Seyed Armin Hashemi | Sahar Tabibian | Mohsen Abbasipour | Maghsood Hosseini

During the last decades, Caspian forests have been attacked by human interference. Easy access, abundance and diversity of valuable forest products have led to the increase in population density, the creation of new residential areas and the activity of deforestation. Revealing the changes is one of the basic methods in the management and evaluation of natural resources. The purpose of this study is to estimate the forest area using support vector machine classification model in Landsat 8 satellite images in Shafarood forests of Gilan province. The results of the image classification in the support vector machine method showed that the area of the forest in 2010 was equal to 12104.64 hectares, which has reached 9478.69 hectares in 2020, that is, its area has decreased by 2625.95 hectares due to changes in its use and its conversion to residential use and poor rangeland. With its growth in the years 2010 to 2020, residential use has changed from 1385.1 to 2542.35, that is, the area has increased by 1157.25 hectares. The changes in pasture use have also changed from 2707.74 to 4478.17 hectares, i.e. 1770.43. The reason for the increase in this could be that many parts of forest use have turned into poor rangeland. According to the Kappa coefficient and the overall accuracy obtained by the support vector machine classification in 2010 and 2020, for 2010, 95.76 and 91.75% and 99.72 and 99.62% were obtained in 2020. The results showed that the support vector machine method had a higher accuracy in the separation of uses.

Afficher plus [+]