Page 41 - AJWEP-22-4

Spatiotemporal variability and climate forcing mechanisms

The periodic analysis results of the EWED has higher TSR levels, showed the smallest decreasing
and the annual TSR in NWC from 1961 to 2019 trend (−7.86 W·m ·a ).
−1
−2
(Figure 3A and B, Table 1) showed that, on this
timescale, the EWED and annual TSR exhibited 3.3. Climate-driving mechanisms of wind and SER
significant positive and negative phase alternations To understand the climate-driving mechanisms of WER
during cycles of 25 – 40 years and 23 – 44 years, and SER in NWC on an interannual scale, existing
respectively. Meanwhile, the wavelet variance results climate indicators were selected for correlation analysis.
indicate that the EWED and the TSR in NWC have The results (Table 2) showed that, in the 2-time series
significant main oscillation periods of 29 and 30 years, (before and after the abrupt change), WER and climate
respectively. This suggests that the periodic variation of factors in NWC exhibited opposite trends. Before
EWED in NWC is faster than that of TSR. The influence the abrupt change, the correlation between the ASC
of climate change and other factors on the interdecadal and WER was high. After the change, WER showed
variation of TSR may be limited, whereas the EWED relatively strong correlations with cloud cover and
demonstrated a higher periodic frequency. relative humidity. In contrast, the correlation between
The table also shows that both WER and SER in SER and climatic factors exhibited a pattern opposite
NWC are decreasing year by year, with stable periodic to that of WER. Before the mutation, SER was more
changes. Overall, over the past half-century, WER and strongly correlated with relative humidity, while after
SER in NWC have exhibited a significant downward the mutation; SER correlation with the ASC became
trend, accompanied by a pronounced main cycle relatively high. Based on relevant research, these
change, indicating potential future changes in WER and results suggest that the driving influence of climate
SER in the region. These findings highlight the need factors on the WER and SER may vary due to feedback
for further simulation analyses incorporating additional mechanisms at the water–soil–air interface in the region,
influencing factors. influenced by global warming, which in turn alters the
dominant climatic control factors.
3.2. Spatial variation characteristics of wind and In addition, due to the spatial heterogeneity of
SER topography and altitude in the northwest region, there
In this study, the multi-year average values of EWED were significant differences in climate distribution
and annual TSR in NWC from 1961 to 2019 were and characteristics. As a result, WER and SER in the
calculated. Based on the inverse distance weighting northwest region were affected by the interaction of
model in ArcGIS 10.8 software, spatial distribution multiple climatic factors. Regression analysis among
maps of EWED and annual TSR for NWC were these factors may lead to multicollinearity issues,
generated (Figure 4). The results in Figure 4A and B thereby reducing the explanatory power of individual
show that EWED in NWC generally increases from climatic factors on the interannual variation of WER and
south to north. In most areas, EWED ranged between SER. To address this, the RF model, based on the ranger
0 and 100 W·m , while in Xinjiang and northern Inner package in R, was used to calculate the importance
−2
Mongolia, it exceeded 100 W·m . The region with the of various climate factors on WER and SER at the
−2
smallest decreasing trend in EWED (−0.26 W·m ·a ) interannual scale. The dominant influencing factors
−1
−2
is located in the southern part of the northwest region, were then identified based on their importance rankings
including southern Qinghai, Gansu, and Shaanxi (Figure 5). Figure 5A-D show that the R values of the
2
provinces. In contrast, Xinjiang and northern Inner model results exceeded 0.6, indicating a high model fit
Mongolia (regions with high EWED) also exhibited the and a strong correlation between WER, SER, and the
most significant decreasing trend (−1.44 W·m ·a ). selected climate variables.
−2
−1
The results in Figure 4C and D indicate that the spatial The results of the RF model (Figure 5) showed
distribution patterns of annual TSR and EWED in NWC significant differences in the main climate-controlling
differed. TSR generally decreases from the central factors between WER and SER on the interannual
region toward the northwest and southeast. In most scale. Before 1991, the main climate-controlling factor
areas, annual TSR ranged from 5,650 to 6800 MJ·m , influencing changes in WER was the North Atlantic
−2
while in southern Shaanxi, it fell below 5,000 MJ·m . oscillation (NAO; explaining 10.43%; p<0.01),
−2
The region with the most significant decrease in annual while after 1991, the dominant factor shifted to cloud
TSR (−28.12 W·m ·a ) was in southwestern Qinghai, fraction, which explained 11.84% (p<0.01). Over the
−2
−1
whereas the northern part of Gansu province, which past half-century, changes in WER in NWC have also

Volume 22 Issue 4 (2025) 33 doi: 10.36922/AJWEP025190147