Document Type : Research Paper
Authors
1 Ph.D. Candidate of Rangeland Science, Faculty of Natural Resources and Geosciences, Shahr-e-Kord University, Iran,
2 Associate Professor, Faculty of Natural Resources and Geosciences, Shahr-e-Kord University, Iran
3 Assistant Professor, Faculty of Natural Resources, Kurdistan University, Sanandaj, Iran
Abstract
The use of satellite data is one of the proper methods, which makes studying ecosystems less costly. This research was carried out to determine the correlation among the vegetation cover of dominant species in three sites with the NDVI index. For this purpose, a study was conducted on three different vegetation types. In each region, a range was determined for sampling. Then, within this range and in the horizontal direction, 30 sampling units of 30 x 30 m were selected along three 900-meter transects in a random-systematic manner. In each sampling unit, nested quadrates with dimensions of 1 × 1, 2 × 2 m were placed. In order to prevent geometric errors, the 900-meter ranges were 60 meters apart. Within each sampling unit, geographic coordinates were recorded with GPS. The number of individuals and canopy cover percentage of dominant species were recorded separately in the plots. Then, the correlation of canopy cover percentage with NDVI index was obtained by processing Landsat 8 images. The distribution pattern was also determined using species density data. The results showed that in all species, the correlation coefficient of NDVI index was higher in the plots with higher area. Also, the correlation coefficient with one quadrat increased towards five quadrats. Due to the lack of high correlation between the total canopy cover percentage of a quadrat and the NDVI index, the use of one quadrat inside the pixel is not recommended in any way. Selecting the type of sampling depends on distribution pattern and species size as well as access to facilities and correlation coefficient acceptability. For species with clumped distribution, more quadrats are needed with proper distribution inside the pixel on the ground, so that sampling could be a good representation of the total pixel. In uniform distribution, fewer samples are needed since the whole pixel is the same in terms of plant growth.
Keywords
)