Page 231 - AJWEP-22-6

FVC and climate in Yarkand Basin

investigate the pixel-by-pixel correlation between FVC past behavior), respectively. Furthermore, as the value
and climatic factors (temperature and precipitation) of H approaches 1, the persistence strength of the VHI
in the Yarkand River Basin from 2000 to 2023. The series increases; conversely, as it approaches 0, the anti-
formula of the correlation coefficient was shown in persistence intensity strengthens.
Equation III.

n ( x )( y) 3. Results
x y
r i1 i i (III) 3.1. Characteristics of the spatial distribution of FVC
n ( x ) 2 n ( y y) 2
x
i1 i i1 i The Yarkand River Basin flows from north to south and
passes through six counties and cities: Bachu County,
where r is the correlation coefficient of x (FVC) and Tumushuke City, Makit County, Shache County, Zepu
y (temperature or precipitation), and x and y are the County, and Yecheng County. The FVC in the basin
mean values of variables x and y. Based on the calculated shows noticeable spatial differences, as shown in Figure
results combined with the significant p-value, the 2. Overall, it presents a long and narrow vegetation
correlation was categorized into five classes: highly belt running from northeast to southwest, as well as a
significant positive correlation (r ≥ 0.5, p<0.01), vegetation belt spanning from northwest to southeast,
significant positive correlation (0.25 ≤ r < 0.5, p<0.05), respectively. The high and medium-high FVC is mainly
non-significant correlation (|| < 0.25, p≥0.05), and distributed in the central part of Bachu County, the
significant negative correlation (−0.5 < r ≤ −0.25, middle area between the north and south of Tumushuke
p<0.05), and highly significant negative correlation City, the long and narrow western area of Makit County,
(r ≤ −0.5, p<0.01). 30,31 the eastern part of Shache County, the vast northern and
central regions of Zepu County, the northwestern area
2.3.5. Persistence analysis connecting Yecheng County and Zepu County, and the
The Hurst exponent, which can be used to predict mountainous region in the central part.
future trends based on vegetation health index (VHI) FVC and its interannual fluctuation were analyzed
time series, is typically calculated using the rescaled across county-level administrative units of the Yarkand
range (R/S) analysis method. It characterizes the long- River Basin (Tables 1 and 2). The results show that the
term memory of the VHI time series, ranging from 0 proportion of low coverage in Bachu and Makit is the
to 1. The Hurst exponent is estimated through several highest (close to 80%), and the high fluctuation changes
computational relationships (Equations IV to VIII). are significant (55.9% in Bachu and 72.63% in Makit),
R c H (IV) indicating that the FVCs in these two counties are poor
S and have low stability. The proportions of medium-
high and high coverage in Zepu and Tumushuke were
the highest (61.01% in total for Zepu and 40.51% in
R τ = max 1 τ ≤≤t X t , τ −min 1 τ ≤≤t X t , τ (V) total for Tumushuke), among which the high fluctuation
change in Zepu was the lowest (25.41%), indicating
1
1
S t1 FVC FVC 2 2 ( 12 , , n , ) (VI) that its FVC was good and relatively stable. Within

t
the entire area, the proportion of low coverage reached

was 55.95%, indicating that the overall FVC level in the

X t, t FVC FVC ( 1 t ) (VII) 65.75%, and the proportion of high fluctuation variation

t
t1
study area was relatively low and exhibited significant
FVC 1 FVC 12 , , n , ) (VIII) interannual variation.
From 2000 to 2023, the FVC coverage and volatility
(
t1 t
in the Yarkand River Basin varied significantly among
where H represents the Hurst exponent, t is the time different land use types (Tables 3 and 4). The high
variable, τ is the time lag, and c is a constant. Values of H coverage of agricultural land accounted for 51.75%,
within the ranges of 0 < H < 0.5, H = 0.5, and 0.5 < H < 1 and the medium-high coverage of forest amounted to
indicate that the VHI time series exhibits anti-persistence 60.81%, indicating that irrigation and natural conditions
(i.e., future trends are opposite to past behavior), random promote the growth of vegetation. Low coverage in
walk behavior (i.e., future changes are uncorrelated with bare land accounted for 95.58% and high fluctuation
the past), and persistence (i.e., future trends align with accounted for 74.51%, indicating high environmental
Volume 22 Issue 6 (2025) 225 doi: 10.36922/AJWEP025350269