Spatial distribution of Persian Oak decline using a combination of geostatistical techniques and remote sensing (Case study: Barm plain, Fars province)

Subject Areas : Agriculture, rangeland, watershed and forestry

Raziyeh Bagheri ¹ , Yousef Erfanifard ²

1 - MSc. Student of Desert Management, Faculty of Agriculture, Shiraz University, Iran
2 - Assoc. Prof. Department of Remote Sensing and Geographical Information System, Faculty of Geography, University of Tehran, Iran

Received: 2019-11-04 Accepted : 2020-03-21 Published : 2020-03-20

Keywords: RapidEye image, Decline map, Persian oak, geostatistic, Spatial variation,

Abstract :

Zagros arid woodlands are among the most important and valuable areas of Iran΄s natural resources that due to the dieback of trees in recent years, it seems necessary to manage and rehabilitate this vegetation. This research was aimed to study the spatial distribution map dieback of Persian oak trees (Quercus brantii Lindl), analyze and describe the spatial distribution using a combination of geostatistical techniques and remote sensing in Barm plain, Fars province. First, the RapidEye satellite image was classified into two categories of healthy and dried trees with the supervised classified algorithm including maximum likelihood. The overall accuracy and Kappa coefficients were 80% and 73%, respectively. The data were then collected in circular sample plots of 2000 m2 (with a radius of 25.24 m) based on a 300×300 meter network in a randomized manner. After preparing the point map, the percentage of drying of the classified image was determined by simple, ordinary, and universal Kriging  interpolation method, which were evaluated using three models: Exponential, spherical, Gaussian. After evaluating the data using the cross-evaluation results, the most accurate fitting was shown by the simple Kriging method with the exponential model (mean estimation error of 0.023). Dieback map was obtained with classes of zero to 10, 10-20, 20-30, 30-40 and more than 40%. The largest area was related to class 20% to 30% with 493.9 ha and the smallest area was for zero to 10%, with 70.46 ha. The present study showed that it is possible to obtain maps of the spatial distribution of Persian oak dieback and recognize the focal points using geostatistical techniques and remote sensing.