Page 230 - AJWEP-22-6

Qin, et al.

ground validation, we conducted field measurements respectively. Combined with the dynamic adjustment
24
at selected sites (e.g., Yecheng, Shache, and Zepu thresholds for land use types (woodland, grassland,
counties) using a handheld NDVI device (GreenSeeker bare land, farmland, etc.), the FVC was categorized
RT200, Trimble Inc., United States of America), which into five levels: low coverage (0 ≤ FVC < 0.2),
showed good agreement between remote sensing data medium-low coverage (0.2 ≤ FVC < 0.4), medium
and ground-measured NDVI values, with a correlation coverage (0.4 ≤ FVC < 0.6), medium-high coverage
coefficient (R ) exceeding 0.85. This validation (0.6 ≤ FVC < 0.8), and high coverage (0.8 ≤ FVC ≤ 1). 25
2
enhances the reliability of the MODIS NDVI dataset for
ecological analysis in the study area. 2.3.2. CV
In this study, the CV was used to quantify the
2.2.3. Meteorological data interannual fluctuations of FVC in the Yarkand River
In this study, the China Meteorological Administration Basin from 2000 to 2023, assessing vegetation stability
(CMA) daily dataset was utilized to obtain temperature (Equation II).
and precipitation data for the Yarkand River Basin from
2000 to 2023. Data were collected from approximately CV FVC FVC (II)
2,400 meteorological stations across China using the FVC
Vaisala HMP155A (Vaisala Oyj, Finland) temperature Where σ is the standard deviation, and FVC is
FVC
and humidity sensor for temperature measurements, and the mean FVC over the 24 years. A smaller CV
FVC
the Tipping Bucket Rain Gauge RG13-H (Vaisala Oyj, indicates lower fluctuation. Combining land use types
Finland) for precipitation measurements, with a spatial (forest land, grassland, bare land, farmland, etc.) and
resolution of 1 km. To align with the 250 m resolution of county/city classifications, spatial differences in FVC
the MODIS NDVI data used for FVC analysis, the CMA fluctuation were analyzed. FVC fluctuation was
data were resampled using ArcGIS software (ArcGIS categorized into five levels: low fluctuation change
10.8, ESRI, United States of America). By overlaying (CV ≤ 0.1), lower fluctuation change (0.1 < CV ≤ 0.15),
FVC
land use type maps (forest land, grassland, bare land, medium fluctuation change (0.15 < CV ≤ 0.2), higher
FVC
farmland, and others), we generated annual average fluctuation change (0.2 < CV ≤ 0.3), and high
FVC
temperatures and total precipitation for each land type. fluctuation change (CV > 0.3). 26,27
FVC
In addition, by integrating the administrative boundaries FVC
of counties and cities, we extracted annual average 2.3.3. Sen+Mann–Kendall trend analysis
temperatures and precipitation for each administrative This study employed the Sen+Mann–Kendall trend
unit to assess the spatial heterogeneity of the regional analysis to assess the spatiotemporal trends in FVC in
climate. The annual temperature and precipitation data the Yarkand River Basin from 2000 to 2023. The Mann-
for the entire region were calculated as average values Kendall test determined the presence of a significant
to reveal the overall trend of climate change. A public trend by comparing the order relationships between data
access link for the CMA dataset is available at http:// points, without relying on the specific distribution of the
data.cma.cn. data, making it suitable for vegetation studies in arid

regions. The Theil-Sen slope quantified the magnitude
2.3. Research methodology of change, and spatial heterogeneity was analyzed by
2.3.1. Pixel dichotomy combining land use types (forest land, grassland, bare
In this study, the FVC of the Yarkand River Basin for land, farmland, etc.) and county/city classifications.
2000–2023 was estimated using the pixel dichotomy The Mann-Kendall test was used to assess significance.
method, which decomposes the image spectra into Trends were categorized into five levels: highly
vegetative and non-vegetative components to calculate significant increase (S > 0, p<0.01), significant increase
the FVC (Equation I). (S > 0, 0.01 ≤ p<0.05), no significant change (p≥0.05),

NDVI NDVI�� significant decrease (S < 0, 0.01 ≤ p<0.05), and highly
FVC min (I) 28,29
NDVI max �� NDVI min significant decrease (S < 0, p<0.01).
Where NDVI is the image element NDVI value, 2.3.4. Pearson correlation analysis
NDVI and NDVI are the 95% and 5% confidence This study employed Pearson correlation coefficient
min
max
values of the regional NDVI gray scale distribution, analysis using the Matrix Laboratory (MATLAB,
reflecting pure vegetation and bare soil characteristics, MathWorks, Inc., United States of America) platform to

Volume 22 Issue 6 (2025) 224 doi: 10.36922/AJWEP025350269