Spatiotemporal Evolution of the Aridity Index and Its Latitudinal Patterns in the Lancang River Basin, China

Abstract

Under the context of global climate change, aridity responses exhibit significant differences across various latitudinal zones, and quantifying the dependency relationship between aridity and latitudinal zones is of great importance for differentiated water resource management. The Lancang River Basin in China spans 13 latitudinal zones with distinct altitudinal gradients, making it crucial to analyze the relationship between long-term aridity variation patterns and latitude for understanding basin hydrological response mechanisms. This study adopted the United Nations Environment Programme (UNEP) aridity index definition and utilized publicly available high-resolution datasets to divide the Chinese Lancang River Basin into 26 regions at 0.5° N intervals. The spatiotemporal evolution characteristics of the aridity index at interannual and seasonal scales from 1940 to 2022 were analyzed, and the trends of aridity index changes and their relationship with latitude were quantified. Results indicate: (1) The spring aridity index increased significantly (trend rate of 0.015/10a, Z = 2.39), driving an overall basin-wide humidification trend. (2) The aridity index exhibited significant spatial and seasonal differences with latitude: southern regions (south of 24.75° N) showed negative correlations, northern regions (north of 30.5° N) showed positive correlations, while central regions displayed distinct seasonal transitions and spatial differentiation characteristics bounded by 27.25° N. (3) The rate of aridity index change in regions north of 27.25° N was significantly higher than in southern regions (p < 0.001). This study reveals the latitudinal patterns of AI changes in the Lancang River Basin, providing guidance for developing adaptive water resource allocation strategies under climate change scenarios.

1. Introduction

Over the past several decades, research has demonstrated that global warming rates are correlated with elevation, a phenomenon known as elevation-dependent warming (EDW). Scholars have proposed extending the EDW concept to elevation-dependent climate change, encompassing not only temperature variations but also other key climate variables to better describe the elevation-dependent nature of critical climatic parameter changes [1,2]. However, the rates of change in climate variables with elevation are not entirely consistent globally and may exhibit significant differences owing to latitudinal variations [3]. Latitude serves as a key factor in modulating the vertical distribution patterns of climatic variables and significantly influences their rates of change. It directly affects the regional energy balance and water vapor transport processes by influencing solar incident angles, seasonal variation amplitudes, and atmospheric circulation patterns, thereby determining the climate characteristics of different latitudinal zones. Research has indicated that different latitudes exhibit varying magnitudes of precipitation change, with latitude even influencing whether precipitation increases or decreases [3]. Therefore, the aridity index (AI), which comprehensively reflects the degree of dryness and wetness, will inevitably undergo changes in response to global climate change; however, the relationship between its rate of change and elevation and latitude remains unclear.

Precipitation change rates and temperature change rates have close relationships with elevation, although research results from different regions are inconsistent. Regarding precipitation patterns, among the world’s major mountain regions, the Tibetan Plateau, the Greater Alps, and the subtropical Andes exhibit increased moisture with elevation, while the Rocky Mountains demonstrate aridification with increasing altitude [4]. In China’s arid regions, precipitation increasing trends are significantly amplified with elevation, confirming the existence of elevation-dependent wetting (EDWE) [5]. Similar phenomena have been observed in the Tibetan Plateau [6], Morocco [7], and Tunisia [8]. However, this elevation dependency varies significantly across different regions. In the Indian Himalayas, precipitation decreases with increasing elevation [9], and the same pattern occurs in the southeastern Tibetan Plateau [10]. Research on temperature changes has indicated that global warming trends are more pronounced in high-latitude regions [11,12], with mountain temperatures showing greater variability at high elevations [13]. In northeastern India, variations in precipitation and temperature are strongly dependent on elevation [14]. Similarly, in the Tibetan Plateau and surrounding regions, annual mean temperature continues to increase with elevation [15].

Furthermore, latitude is also a crucial factor influencing climate variable variations. Research has indicated that the rates of temperature and precipitation change with elevation are not constant but are influenced by latitude [3]. Analysis of precipitation changes in Iran’s Alborz Mountains revealed that latitude is the primary factor influencing precipitation gradients [16]. Concurrently, China’s regional precipitation changes exhibit a pattern characterized by increases in mid-high latitude regions and decreases in tropical and subtropical areas [17]. Based on the above research, both precipitation and temperature changes exhibit significant elevation gradient effects and latitudinal zonality characteristics. Under global warming conditions, these trends may be amplified by increasing elevations [18]. As an indicator that comprehensively reflects regional hydrothermal conditions, the spatial distribution pattern and long-term changes in AI may also be influenced by elevation and latitude.

The AI serves as a climate indicator characterizing regional dry-wet conditions [19,20] and has been extensively applied in previous studies [21,22]. Since the early 20th century, research on AIs has made significant progress [20,23], leading to the development of various aridity indicators [23,24,25]. Among these indicators, the calculation method for AI proposed by the United Nations Environment Programme (UNEP, 1997) [AI = potential evapotranspiration (PRE)/precipitation (PET)] has become the most widely used approach owing to its scientific validity and robustness [26,27,28,29]. Compared with indicators that rely solely on precipitation or temperature, the PRE/PET ratio method comprehensively considers both precipitation and evaporation, providing a more complete reflection of actual regional dry-wet conditions.

The Lancang River Basin (LRB), as an important international river in China, spans 13 latitudinal zones [30,31] and covers 6 climatic zones: frigid, cold temperate, temperate, warm temperate, subtropical, and tropical [32]. The basin extends in a north–south direction, with elevation increasing with latitude, making it an ideal region for studying the spatiotemporal patterns and long-term variability of AI in relation to latitude. This basin is not only crucial for the ecological environment and water resource security of southwestern China, but as the upper reaches of the Mekong River, its hydro-climatic changes also affect water resource utilization and management in down-stream Southeast Asian countries [33]. Previous studies have shown that the spatial distribution of precipitation in the LRB exhibits significant latitudinal differences, whereas longitudinal differences are insignificant [34]. In recent years, annual precipitation in this region has increased at a rate of 24.8 mm/decade, while the warming rate is approximately 0.6 °C/decade, significantly higher than the global average warming levels during the same period [30]. The spatiotemporal patterns of AI in the LRB may change under the influence of continuously rising temperatures and gradually changing precipitation. Some researchers have conducted studies on meteorological elements and drought in the LRB, including drought [35,36], precipitation [37,38,39], and evaporation [32]. Most of the existing studies have primarily focused on individual meteorological factors (e.g., precipitation or temperature). However, due to the limitations of station observations in terms of representativeness in complex terrains and the restricted length of time series, it remains difficult to comprehensively reveal the spatiotemporal evolution of hydroclimatic conditions at the basin scale. In particular, investigations concerning the aridity index (AI)—which can serve as an integrated indicator reflecting regional water–energy balance—and its latitudinal variation patterns are still insufficient. In this study, we employed a long-term, high-resolution dataset (1940–2022) released by the National Earth System Science Data Center of China. The Lancang River Basin was divided into 26 subregions along a 0.5° N latitudinal interval from south to north. On the basis of analyzing the spatiotemporal distribution of annual and seasonal AI across the entire basin and its subregions during 1940–2022, we further examined the relationships between AI (as well as its trends) and latitude. This study aims to: (1) analyze the spatiotemporal evolution characteristics of the aridity index in the Lancang River Basin at interannual and seasonal scales; (2) quantify the variation trends of the aridity index across different latitudinal zones and their spatial heterogeneity; (3) reveal the quantitative relationship between aridity index changes and latitude, along with their spatial pattern characteristics; and (4) provide a scientific basis for developing differentiated water resource management and adaptive allocation strategies in the Lancang River Basin under the context of climate change.

2. Materials and Methods

2.1. Study Area and Data

The LRB is a major international river system in Southwest China, originating from the Tibetan Plateau. Upon exiting Chinese territory, it becomes the Mekong River (MRB) and ranks among the world’s longest rivers. The LRB is located between 93°86′ E–101°84′ E longitude and 21°30′ N–32°40′ N latitude, with a total length of 2160 km within Chinese territory and a watershed area of approximately 174,000 km² [30]. The watershed exhibits complex and diverse terrain with a maximum relative elevation difference approaching 5000 m (Figure 1), displaying significant vertical geomorphological variations [40]. The multi-year average precipitation in the watershed is 962.3 mm [37], with annual average potential evapotranspiration ranging from 900 to 1300 mm [32]. The multi-year average temperature is 11.79 °C [31], with evident spatiotemporal variations in climate change [41]. Land cover exhibits distinct vertical zonation: the upstream region is dominated by temperate alpine meadows and coniferous forests, the midstream area consists of temperate/subtropical shrublands and seasonal forests, while the downstream region is characterized by tropical evergreen broad-leaved forests and rainforests [42]. Watershed land use is primarily composed of forestland and grassland, accounting for more than 85% of the total area, with Oxisols being the predominant soil type [43]. In the context of global warming, the LRB, one of the most climate-sensitive regions globally, has attracted significant attention from relevant domestic and international organizations and governments regarding water resource utilization and management [44]. Considering the unique north–south distribution characteristics of the Lancang River Basin and the elevation-latitude relationship, this study adopted a 0.5° latitudinal zoning method for analysis. This approach can effectively capture the elevational gradient effects within the basin while avoiding the potential uncertainties that may arise from point-by-point analysis of gridded data.

This study utilized the 1940–2022 monthly precipitation and evaporation datasets for China published by the National Earth System Science Data Center of China (https://www.geodata.cn, accessed on 30 September 2024) with a spatial resolution of 0.0083333° (approximately 1 km). These datasets were generated by combining CRU’s global 0.5° climate data with WorlClim’s high-resolution climatological datasets using the delta spatial downscaling method. The CRU data cover the entire Earth, except for Antarctica, at a spatial resolution of 0.5° × 0.5° [45,46]. Since its first release in 2000, the CRU data have been successfully applied in numerous hydro-climatic studies worldwide [45,47,48], and compared with other datasets, CRU data show better consistency with meteorological station observations [49].

Compared with the original CRU data, the monthly precipitation and evaporation datasets for China published by the National Earth System Science Data Center of China (https://www.geodata.cn, accessed on 30 September 2024) demonstrate a 25.7% reduction in average precipitation error. Furthermore, these datasets incorporate topographic effects, distance to the coast, and satellite-derived covariates from the World Clim dataset, enabling a more accurate representation of topographic influences on the climate [50]. In this dataset, potential evapotranspiration was calculated using the Hargreaves potential evapotranspiration formula, which is considered the optimal method for calculating PET [51]. This method relies solely on three monthly climatic parameters: mean, maximum, and minimum temperatures [52]. Furthermore, while soil type and land cover may potentially influence AI calculations through evapotranspiration processes, our study design minimizes these impacts through: (1) standardized UNEP AI methodology suitable for large-scale assessments, (2) basin-scale analysis that smooths local heterogeneity, (3) precipitation’s dominant role in AI variations, and (4) use of meteorologically driven PET following FAO-56 standards [53]. The monthly classification adopted in this study was as follows: winter (December, January, February), spring (March, April, May), summer (June, July, August), and autumn (September, October, November). All data processing, statistical analysis, and visualization in this study were conducted using Python (3.12.0, Python Software Foundation, Wilmington, DE, USA) and related scientific computing libraries, including NumPy (1.26.4) for numerical computations, Pandas (2.2.3) for data manipulation, SciPy (1.13.1) for statistical calculations, and Matplotlib (3.9.2) (all open-source libraries available at https://pypi.org/, accessed on 30 September 2024) for data visualization. Microsoft Excel (2021, Microsoft Corporation, Redmond, WA, USA) was also utilized for preliminary data organization and quality control procedures.

2.2. Method

2.2.1. Aridity Index

The UNEP (1997) aridity index is defined as the ratio of precipitation (PRE [mm]) to potential evapotranspiration (PET [mm]) (Equation (1)).

$$\mathrm{AI}={\displaystyle \frac{PRE}{PET}}$$

(1)

A higher AI value indicates a more humid climate, whereas a lower AI value indicates drier conditions. The climatic classification based on AI is presented in

Table 1. UNEP Aridity Index (AI) classification.

Aridity Index (mm/mm)	Category/Climate
<0.05	Hyperarid
0.05–0.2	Arid
0.2–0.5	Semiarid
0.5–0.65	Dry subhumid
0.65–1	Wet subhumid
>1	Humid

.

2.2.2. Linear Tendency Rate

The temporal trends of AI in the LRB were evaluated using linear regression. The regression equation is as follows:

$$X_t=a_0+a_{1}t$$

(2)

where t represents the time series (1940–2022) and $X_t$ represents the AI (ratio of precipitation to potential evapotranspiration, P/PET) estimated via linear regression. The regression coefficient $a_1 = d{X}_t/dt$ represents the rate of temporal change in AI, with the climatic tendency rate of AI defined as $b = a_1 \times 10$ (per decade change rate). A positive value suggests a trend toward increased wetness, whereas a negative value indicates a shift toward heightened aridity.

2.2.3. Detection of Trends

The M–K test was employed to analyze trends in AI across the LRB. Some scholars consider the nonparametric Mann–Kendall (MK) test [54,55] to be the most appropriate method for analyzing climate data series changes, as it can effectively identify long-term trends in climate variables [56]. The M–K test statistic S is calculated as follows:

$$S = \sum _{i=1}^{n-1}\sum _{j=i+1}^n\mathrm{sgn}\left(x_j-x_i\right)$$

(3)

where $\mathrm{sgn}\left(x_j-x_i\right)$ is the sign function, which equals −1, 0, and 1 when the argument is negative, zero, and positive, respectively. When n ≥ 8, the statistic $S$ approaches a normal distribution with zero mean, and the variance is given in Equation (4).

$$Var(S) = \left[n(n-1)(2n+5)-\sum _{k=1}^m{t}_k\left(t_k-1\right)\left(2t_k+5\right)\right]/18$$

(4)

where $t_k$ is the number of data points in the $k$ group of tied values, m is the total number of tied groups, and n is the sample size. The M–K test statistic $Z$ was calculated using Equation (5).

$$Z = \left\{\begin{array}{c}(S-1)/\sqrt{Var(S)}, \mathrm{if} S>0\\ 0, \mathrm{if} S=0\\ (S+1)/\sqrt{Var(S)}, \mathrm{if} S<0\end{array}\right.$$

(5)

Positive and negative Z-values represent increasing and decreasing trends, respectively, indicating that the basin has become wetter or drier. When $\left|Z\right|>Z_{1-\alpha /2}$ is used, the null hypothesis of no trend is rejected, where $Z_{1-\alpha /2}$ is the critical value of the standard normal distribution. At significance levels of 5 and 1%, the $Z_{1-\alpha /2}$ is 1.96 and 2.576, respectively.

2.2.4. Change Point Test

The Pettitt test [57] is a non-parametric statistical method that serves as an important tool for detecting change points in a time series and has been widely applied in hydro-climatic variability studies. Its basic principle is to assess the distributional differences before and after a certain time point in a sequence by constructing rank statistics. For a time series of length, the Pettitt statistic is defined as:

$$U_{t,n}=\sum _{i=1}^t\sum _{j=t+1}^n\mathrm{sgn}\left(X_i-X_j\right)$$

(6)

where $\mathrm{sgn}(x)$ represents the sign function, which takes the value of 1 when x > 0, 0 when x = 0, and −1 when x < 0. The test statistics are calculated as follows:

$$K_n=\underset{1\le t\le n}{\max }\left|U_{t,n}\right|$$

(7)

When the $p$-value is less than the given significance level, a change point is considered to exist in the statistic $K_n$ at the corresponding time point.

2.2.5. Multiple Regression Approach

Multiple regression is commonly used to explain how multiple independent variables influence a dependent variable and to analyze the degree of influence of independent variables on the dependent variable [58]. We employed this method to conduct a sensitivity analysis of AI, using standardized regression coefficients to characterize the contribution of different independent variables to AI. The formula used is as follows:

$$y=a{x}_1+b{x}_2+\epsilon$$

(8)

where $y$, $x_1$ and $x_2$ represent the standardized AI, precipitation, and evaporation data, respectively. The standardization process eliminated the influence of different measurement units among the variables. a and b are standardized regression coefficients, and ε is the residual term.

3. Results

3.1. AI Spatial Patterns

Figure 2a–e illustrates the spatial distribution characteristics of AI across the basin. The LRB exhibited distinct spatial and seasonal variations in AI, with different climatic types prevailing during different seasons. Figure 2a presents the spatial distribution of annual AI, revealing that regions south of 29° N were generally more humid than those to the north, reflecting an overall north–south gradient in moisture conditions. The spatial patterns of aridity varied markedly across seasons. During summer (Figure 2c) and autumn (Figure 2d), the spatial distributions of AI were similar, with relatively arid conditions in the central part of the basin, whereas the northern and southern regions remained more humid. In contrast, winter (Figure 2e) and spring (Figure 2b) exhibited the opposite patterns, characterized by higher humidity in the central basin and drier conditions in the northern and southern extremities.

Table 2. AI Statistical Values of the Basin-wide.

	Spring	Summer	Autumn	Winter	Annual
Max	0.98	1.75	1.53	1.15	1.11
Min	0.26	1.01	0.52	0.29	0.77
Std	0.02	0.02	0.03	0.03	0.01
Mean	0.60	1.36	0.93	0.62	0.95
Category	Dry subhumid	Humid	Wet subhumid	Dry subhumid	Wet subhumid

presents the statistical metrics of annual and seasonal AI in the LRB. The annual mean AI was 0.95, ranging from 0.77 to 1.11, classifying the basin as wet sub-humid according to the UNEP aridity classification. However, significant seasonal variations were observed. In summer, AI ranged from 1.01 to 1.75 (mean: 1.36), indicating a humid climate. Autumn had a mean AI of 0.93, falling within the wet sub-humid category. In contrast, winter and spring exhibited drier conditions, with mean AI values of 0.62 and 0.60, respectively, corresponding to a dry sub-humid climate. These results highlight a distinct seasonal pattern: the basin is humid in summer and autumn but becomes drier in winter and spring.

3.2. Climatological Changes

This section analyzed the long-term trends of AI across the entire basin. Considering the significant influence of precipitation on AI, we divided the study area into 26 regions at 0.5° latitude intervals, analyzed the long-term trends of AI in these 26 regions, and examined the relationships between AI and its trend rates with latitude.

3.2.1. Basin-Wide AI Trends

Table 3. Basin-wide AI Trend Rate.

	Spring	Summer	Autumn	Winter	Annual
Tendency rate	0.015	−0.007	0.002	0.005	0.003
Z-value	2.39 *	−1.41	0.56	0.30	0.75

Note: * indicates statistical significance at p < 0.05.

presents the trend detection results for annual and seasonal AI in the LRB. AI exhibited distinct variation trends across seasons, with an overall tendency toward increased wetness in the basin. At the annual scale, AI showed a non-significant increasing trend. However, seasonal analysis revealed divergent patterns; spring demonstrated a significant upward trend (p < 0.05), suggesting a potential climate shift from dry to wet sub-humid. This transition may have important implications for spring agriculture, livestock, and the ecosystems in the basin. Autumn and winter showed positive trend slopes (though insignificant), indicating weak moistening tendencies. In contrast, summer exhibited a non-significant decreasing trend, with the absolute trend magnitude being greater than that of autumn/winter but weaker than that of spring. This seasonal divergence in aridity highlights the complex climatic dynamics of the basin.

3.2.2. Regional AI Trends

Figure 3 shows the M–K test results for AI across 26 latitudinal zones in the LRB. Annual AI exhibited a significant increasing trend (α = 0.05) in high-latitude regions (north of 32° N), indicating that these areas are becoming more humid, while no significant trends are observed in mid- and low-latitude regions. For seasonal variations, spring AI showed widespread increasing trends throughout the basin, with particularly significant increases north of 29° N. Summer AI displayed increasing trends north of 33° N but non-significant decreasing trends in other regions. Autumn AI demonstrated mostly non-significant increasing trends, except in the southern basin (21° N–24° N). Winter AI showed significant increasing trends in the four latitudinal zones (31° N–33° N). Overall, AI in all four seasons in the LRB showed upward trends at the northernmost part of the basin (33° N–34° N), with spring aridity particularly exhibiting a significant upward trend in high-latitude areas (29° N–34° N). These results reflect the varying responses of basin aridity to global climate change across different regions and seasons. The positive Z-values increased with latitude, with all values being greater than 0 in the northernmost region (33° N–34° N), further confirming the wetting trend in the northern basin.

3.3. Climate Jump

3.3.1. Basin-Wide Climate Jump

Table 4. Abrupt change detection results of AI.

	Annual	Spring	Summer	Autumn	Winter
abrupt point	1961	1988	1972	1972	1946
p-value	p > 0.05	p < 0.05	p > 0.05	p > 0.05	p > 0.05

presents the abrupt change detection results for AI in the basin from 1940 to 2022. Annual AI in the LRB exhibited an abrupt change in 1961. Abrupt changes also occurred in 1946 (winter), 1972 (summer and autumn), and 1988 (spring). However, only the abrupt change in spring AI was significant (p < 0.05). Therefore, we compared the variations in AI before and after this abrupt change.

Table 5. Comparison before and after abrupt of spring AI.

	Mean	Median	Variance	Std. Dev.	Min	Max	Max/Min
Pre-1988	0.56	0.54	0.02	0.13	0.27	0.82	3.04
Post-1988	0.56	0.64	0.03	0.16	0.29	0.98	3.38

compares the six statistical indicators of spring AI before and after an abrupt change. Following this change, both the mean and median of spring AI showed significant increases, aligning with the conclusion in Section 3.2 that spring AI increased, indicating a trend toward wetter conditions in the basin during spring. Additionally, the post-change period exhibited larger standard deviations and extreme value ratios than those of the pre-change period, reflecting greater variability in aridity. Specifically, maximum AI increased from 0.82 (pre-change) to 0.98 (post-change), with this upward shift being more pronounced than that of the minimum values, thereby driving the observed increases in both the standard deviation and extreme value ratio. The significant abrupt change in spring AI (p < 0.05) suggests a climatic shift toward wetter springs in recent years, which may benefit spring crop growth in the basin. However, the concurrent rise in the standard deviation and extreme value ratio implies that seasonal drought variability could become more pronounced, potentially exacerbating springtime water management challenges.

3.3.2. Regional Climate Jump

The Pettitt test was applied to detect abrupt changes in AI across the 26 latitudinal zones of the basin from 1940 to 2022. Although all zones exhibited abrupt changes in AI, the timing of these shifts varied significantly. Notably, only spring AI changes in 1988 (29.5° N–33° N) and 1974 (33° N–34° N) passed the significance test (p < 0.05). To quantify the magnitude of the change, we calculated the variability ratio—$\mathrm{Ratio}=\left({\overline{AI}}_{\mathrm{post}}-\overline{A{I}_{\mathrm{pre}}}\right)/{\overline{AI}}_{\mathrm{pre}}\times 100$—for each latitudinal zone before and after the abrupt changes. The results are presented in

Table 6. AI Ratio and abrupt change point by regions.

Latitude (°N)	Annual		Spring		Summer		Autumn		Winter
	Ratio	Change Point	Ratio	Change Point	Ratio	Change Point	Ratio	Change Point	Ratio	Change Point
21.25	4.53	1963	16.3	1970	6.86	1951	−9.92	1993	16.43	1958
21.75	4.89	1964	17.5	1970	−4.86	1972	−10.28	1993	16.68	1958
22.25	3.93	1964	18.35	1970	−6.43	1972	−12.02	1998	14.3	1958
22.75	2.92	1963	14.97	1988	−6.93	1972	−13.66	2002	5.81	1999
23.25	−5.61	2009	17.26	1988	−7.1	1972	−13.82	2002	−10.49	1969
23.75	−5.68	2009	20.47	1996	−7.44	1974	16.13	1967	−9.07	1969
24.25	4.3	1965	24.13	1996	−7.26	1987	19.75	1967	34.15	1946
24.75	−5.99	2009	23.2	1996	−6.86	1987	21.91	1970	−8.32	1972
25.25	3.52	1965	19.79	1996	−7.07	1975	22.11	1970	42.08	1946
25.75	−7.62	2009	18.81	1974	−7.29	1975	20.47	1970	43.4	1946
26.25	−8.68	2009	16.22	1974	−7.86	1975	18.18	1970	−15.02	1994
26.75	−9.61	2009	14.03	1974	−8.45	1975	15.35	1976	−17.58	1994
27.25	−10.4	2009	13.67	1974	−8.58	1975	15.65	1976	−18.49	1994
27.75	−11.67	2009	15.93	1974	−8.36	1975	15.52	1976	−17.38	1994
28.25	−11.74	2009	17.95	1974	−7.83	1975	14.75	1976	−16.34	1994
28.75	−10.37	2009	20.24	1974	−11.47	2005	15.15	1976	−20.94	2007
29.25	4.78	1984	20.75	1988	−12.3	2006	13.79	1976	−18.35	2007
29.75	−7.19	2006	20.98 *	1988	−12.29	2006	12.35	1976	34.27	1947
30.25	4.98	1985	22.09 *	1988	−12.39	2006	10.62	1985	8.7	1989
30.75	5.45	1985	23.84 *	1988	−12.3	2006	10.2	1985	14.39	1989
31.25	5.36	1985	26.22 *	1988	−10.93	2006	10.04	1978	19.34	1989
31.75	5.15	1985	25.79 *	1988	−8.9	2006	12.13	1963	20.52	1989
32.25	5.22	1974	25.82 *	1988	−3.9	1971	12.58	1963	19.11	1989
32.75	5.08	1980	26.04 *	1988	−2.82	1971	13.09	1963	19.61	1995
33.25	7.08	1998	26.27 *	1974	5.16	2003	13.55	1963	17.95	1995
33.75	9.06	1998	28.26 *	1974	7.76	2003	13.88	1963	19.64	1995
Legend
AI change		Slight→Substantial (Decrease)			Slight→Substantial (increase)

Note: * indicates significance at p < 0.05. Color coding: red cells indicate significant decrease in aridity, blue cells indicate significant increase in aridity. Color intensity reflects the magnitude of change—darker colors represent larger changes.

, which demonstrates distinct spatial patterns in the intensity of AI shifts across the basin.

As shown in Table 6, the variability in AI differed across latitudinal zones. In spring, the post-abrupt change AI values in all 26 latitudinal zones were consistently higher than those before the change, with AI variability increasing with latitude. Except for 33.25° N and 33.75° N, the mean summer AI values after the abrupt changes were lower than those before the shift. Autumn AI exhibited a distinct north–south divergence: post-change values in the southern basin were lower than pre-change levels, whereas the central and northern regions showed the opposite pattern. Winter displayed more complex characteristics, with the largest increases in AI before and after the abrupt change occurring at 25.25° N and 25.75° N (42.08 and 43.40%, respectively), whereas the most significant decreases were observed between 26.25° N and 29.25° N. These findings demonstrate that the abrupt change characteristics of AI vary across seasons and regions.

3.4. Relationship Between AI and Latitude

3.4.1. Characteristics of Annual AI Variation with Latitude

Figure 4 illustrates the complex relationship between AI and latitude in the LRB. South of 25° N and within the 27.5° N–30.5° N region, AI exhibited a significant negative correlation with latitude (p < 0.001). Conversely, in the 25.5° N–27.5° N and 30.5° N–33.5° N regions, AI positively correlated with latitude (p < 0.001). The latitudinal variation in AI differed across subregions, with the southern basin (south of 25.5° N) displaying the most rapid change; AI decreased by approximately 0.12 per degree of increasing latitude. The northern region (north of 30.5° N) followed, where AI increased by approximately 0.09 per degree of latitude.

3.4.2. Characteristics of Seasonal AI Variation with Latitude

Figure 5 presents the relationship between AI and latitude across the four seasons in the LRB. Seasonal AI–latitude patterns can be categorized into two distinct types: spring and winter (Figure 5a,d) exhibited similar characteristics, forming one group, whereas summer and autumn (Figure 5b,c) displayed another coherent pattern, representing the second group.

South of 24.75° N in the basin, AI showed consistent negative correlations with latitude across all four seasons. Except for winter, which failed to pass the significance test (p > 0.05), all other seasons showed significant negative correlations (p < 0.05), indicating that AI decreased with increasing latitude in this region year-round. In the area north of 30.5° N, AI exhibited significant positive correlations with latitude in all four seasons (p < 0.05), suggesting that AI increased with latitude in this region. Notably, the relationship between AI and latitude became more complex in the central part of the basin (24.75° N–30.5° N). During spring and winter, the AI–latitude relationship displayed opposite patterns north and south of approximately 27.25° N, with significant positive correlations (p < 0.001) south of 27.25° N and significant negative correlations (p < 0.001) north of this latitude. However, such a transitional pattern was absent in summer, and autumn–summer showed consistent negative correlations (p < 0.05), whereas autumn demonstrated uniform positive correlations (p < 0.05) throughout the central basin.

In summary, although a correlation exists between AI and latitude in the LRB, this relationship demonstrates significant seasonal and zonal variations.

3.5. Relationship Between Long-Term AI Changes and Latitude

3.5.1. Relationship Between Annual AI Tendency Rate and Latitude

Figure 6 illustrates the relationship between the AI trend rate and latitude across the 26 latitudinal zones in the LRB. The AI trend rate exhibited significant spatial differentiation and was approximately demarcated by 27.5° N. North of 27.5° N, the AI trend rate and latitude were significantly negatively correlated (p < 0.001), indicating a wetting trend that intensified with increasing latitude. Conversely, south of 27.5° N, a significant negative correlation was observed (p < 0.01), reflecting a drying tendency. Overall, 18 latitudinal zones showed positive trend rates (>0), suggesting a general wetting trend across the basin. Notably, the trend rates north of 27.5° N were substantially greater than those in the southern regions, demonstrating more pronounced wetting in the northern areas of the basin.

3.5.2. Relationship Between Seasonal AI Tendency Rate and Latitude

Figure 7 illustrates the relationship between the mean AI tendency rate and latitude across the four seasons. Figure 7a depicts the correlation between the spring AI tendency rate and latitude, revealing uniformly positive values that indicate an overall wetting trend throughout the basin during spring. This moistening trend intensified progressively with increasing latitude, showing a significant positive correlation between the tendency rate and latitude (p < 0.01). Figure 7b shows the relationship between the summer AI tendency rate and latitude, demonstrating a transition from negative to positive values from south to north. The tendency rate exhibited a significant positive correlation with latitude (p < 0.01). The relationship between the autumn AI tendency rate and latitude (Figure 7c) followed a pattern similar to that of summer. Figure 7d presents the correlation between winter AI tendency rate and latitude, showing a distinct divergence at approximately 27° N. In the northern basin, the AI tendency rate exhibited a significant positive correlation with latitude (p < 0.01), whereas a significant negative correlation (p < 0.01) was observed with latitude in the southern region. These findings indicate that in the northern basin, areas closer to the Tibetan Plateau at higher latitudes demonstrate greater AI tendency rates, reflecting an accelerated moistening trend. These results collectively demonstrate that spatial variations in AI across the LRB are significantly influenced by latitudinal gradients.

3.6. Sensitivity Analysis

Considering that the Aridity Index (AI) is simultaneously influenced by both precipitation and potential evapotranspiration (PET), to gain a deeper understanding of the underlying mechanisms of the aforementioned spatiotemporal variation characteristics of AI, this study employed a multiple linear regression method to analyze the sensitivity of AI to precipitation and PET in the Lancang River Basin, thereby identifying the dominant factors driving aridity changes in different regions. The results are shown in

Table 7. Sensitivity analysis of AI.

	Pre	Pet	R2	p-Value
Spring	2.03	0.06	0.95	<0.01
Summer	1.19	−0.84	0.94	<0.01
Autumn	2.65	0.06	0.91	<0.01
Winter	2.06	−0.57	0.80	<0.01
Annual	0.87	−0.24	0.99	<0.01

.

The results revealed that precipitation was the dominant factor controlling the AI. The sensitivity coefficients of precipitation were consistently higher than those of potential evapotranspiration, confirming that precipitation was the primary driver of AI variability. Although the absolute values of the precipitation and evapotranspiration sensitivity coefficients were comparable during summer, the sensitivity coefficient of precipitation remained higher than that of potential evapotranspiration. This indicated that while both precipitation and evapotranspiration influenced summer aridity, precipitation remained the dominant controlling factor.

4. Discussion

This study first examined the spatial distribution characteristics of AI in the LRB, followed by an analysis of its long-term trends and abrupt changes. Furthermore, the relationship between AI (including its tendency rate) and latitude was investigated. The results demonstrate that both AI and its tendency rate exhibit significant latitudinal dependence, indicating that the spatiotemporal patterns of aridity in the LRB are strongly influenced by latitude.

4.1. Long-Term Trends of AI

Under the context of global warming, Southwest China has generally trended toward becoming more humid [30,59,60,61,62]. Some studies have indicated that precipitation in the entire Lancang River Basin has shown a weak increasing trend over the past 55 years, with a climate tendency rate of 0.28 mm/10a [37]. These conclusions are consistent with our finding that the basin’s AI shows a non-significant upward trend on an annual scale. The underlying mechanism is that under the influence of global climate warming, water vapor transport from the Indian Ocean and Western Pacific to this region has intensified, leading to increased precipitation and consequently resulting in elevated aridity index in the Lancang River Basin. The sensitivity analysis in Table 7 reveals that the sensitivity coefficient of spring precipitation (2.03) is substantially higher than that of potential evapotranspiration (0.06), indicating that precipitation is the dominant factor controlling spring AI. Studies have demonstrated that spring precipitation in the Lancang River basin exhibits a significant increasing trend with a climate tendency rate of 8.20 mm/10a [37]. Additionally, peach blossom floods occur annually during February-April (spring season) near 28° N in the basin, characterized by significantly increased precipitation [63]. These factors collectively contribute to the significant increase in spring AI in the Lancang River basin.

Conversely, the basin’s summer AI shows a non-significant declining trend. Studies have reported that summer precipitation in the Lancang River Basin exhibits a decreasing trend with a climate tendency rate of −7.33 mm/10a [37]. Furthermore, research has revealed that summer precipitation in the Lancang River Basin underwent an interdecadal abrupt change from abundant to scarce around 2002, with particularly notable decreases in June precipitation leading to a shortened wet season [38]. Moreover, potential evapotranspiration in the Lancang River Basin shows a significant increasing trend [32]. These factors collectively contribute to the decline in summer aridity index.

4.2. Influence of Latitude on AI Spatial Distribution

The current study revealed a strong correlation between AI in the LRB and latitude, influencing both its spatial distribution and temporal trends. On annual and four seasonal scales, AI in areas south of 24.75° N in the basin demonstrated significant negative correlations with latitude. Studies have indicated that the southern of the basin experiences both significant warming (with an annual temperature trend rate of 0.264 °C/10a) and a decreasing precipitation trend (annual precipitation decreased at a rate of −9.91 mm/10a) [64]. The significant temperature increase leads to enhanced potential evapotranspiration, while the continuous decline in precipitation contributes to a significant decrease in the AI in regions south of 24.75° N within the basin, with both factors collectively driving this trend.

In areas north of 30.5° N in the basin, the AI for all four seasons, as well as annual AI, showed significant positive correlations with latitude. Studies have indicated that both annual and spring precipitation and temperature have significantly increased in the upstream region above the Changdu Station within the watershed, and this result may have led to a significant increase in AI in this region [64]. Furthermore, Studies have indicated that the southern Tibetan Plateau is characterized by a winter-dry warm climate zone (CW), where precipitation exhibits a negative correlation with elevation [65]. The southern Lancang River Basin belongs to this climatic region; consequently, precipitation decreases with increasing latitude, resulting in a corresponding decline in the AI. In contrast, in the northern part of the basin, topographic uplift occurs as the latitude increases, resulting in increased precipitation and elevated AI. Therefore, the occurrence of two different relationships between AI and latitude in the southern and northern parts of the basin is reasonable.

In the central part of the basin (24.75° N–30.5° N), spring, winter, and annual AI exhibited two distinct relationships with latitude: a positive correlation south of 27.25° N and a negative correlation north of 27.25° N. This pattern is primarily attributed to the high precipitation zone associated with the “spring flood” phenomenon near 27.25° N [63]. The maximum precipitation near 27.25° N resulted in higher AI values at 27.25° N compared with those in adjacent areas. As AI peaks around 27.25° N, it exceeds values in neighboring latitudinal zones, leading to positive and negative correlations with latitudes south and north of 27.25° N, respectively, in the central basin. This phenomenon is related to valley topography in the central basin, where interactions between valley orientation and airflow direction create complex spatial precipitation patterns. In regions north of 27.25° N, the towering mountain systems impede the southwest monsoon, preventing moisture from penetrating deep into the inland areas, which results in a declining AI. Conversely, in regions south of 27.25° N, the relatively lower elevation and river valley corridors facilitate moisture transport, leading to an increasing AI south of 27.25° N [66]. However, the summer and autumn patterns in the central basin differed from those in winter and spring. Summer AI showed an overall negative correlation with latitude, indicating decreasing aridity with increasing latitude, whereas autumn AI exhibited a positive correlation, indicating increasing aridity with latitude. These seasonal differences reflect distinct climatic mechanisms controlling summer and autumn precipitation compared with winter and spring. Summer represents the rainy season when the entire basin is generally influenced by monsoons, with an abundant and relatively uniform precipitation distribution. However, as latitude increases, the precipitation increment may be smaller than the evaporation increment, leading to decreased AI [38]. Autumn, as a transitional period from the rainy to the dry season, may involve specific water vapor transport pathways and precipitation mechanisms that create spatial patterns that differ from those of other seasons [66].

4.3. Influence of Latitude on AI Tendency Rate

This study revealed that AI trend rates in the LRB are influenced by latitude. Using approximately 27.5° N as the boundary, the northern part of the basin exhibited positive AI trend rates that were positively correlated with latitude, whereas the southern part exhibited negative AI trend rates that were negatively correlated with latitude. Previous studies have demonstrated that latitude not only influences the magnitude of precipitation variation, but also determines the direction of precipitation increase and decrease [3]. Research has revealed that latitude may be a critical factor controlling the spatial distribution of annual precipitation variability across China [61], and the findings of this study indirectly corroborate this hypothesis. Simultaneously, studies have shown that precipitation trend rates are predominantly positive in high-altitude regions, whereas they are negative in low-altitude areas [61], which is consistent with the AI change characteristics observed in the present study. This reflects the influence of precipitation patterns on AI in the LRB and validates our sensitivity analysis finding that precipitation is the dominant factor affecting AI. However, it must be acknowledged that our sensitivity analysis only considered two factors: precipitation and potential evapotranspiration.

The AI trend rate in the LRB varies with latitude, with the AI trend rate gradually increasing as latitude increases, indicating that high-latitude regions become humid at a faster rate. This essentially reflects the fact that as latitude increases, elevation increases, and high-elevation areas become humid at a faster rate. This conclusion is consistent with the EDWE phenomenon identified in previous studies, namely that the increasing trend of precipitation intensifies with rising elevation [5]. Furthermore, studies have indicated that elevation is the primary factor influencing precipitation in mountainous regions, converting atmospheric water vapor into precipitation through topographic lifting mechanisms [4]. As elevation in the LRB increases with latitude, the orographic lifting effect is more pronounced in high-latitude regions, the orographic lifting effect is more pronounced in high-latitude regions, further promoting increased precipitation and strengthening wetting trends. Meanwhile, due to the enhanced North Atlantic Oscillation (NAO) activity, the intensified NAO has promoted the transport of increased water vapor to high-latitude regions by westerly circulation, resulting in increased precipitation in the northern Tibetan Plateau [67]. These factors collectively contributed to the more pronounced wetting trends in the high-latitude regions of the LRB.

4.4. Characteristics of Climate Jump in AI and Regional Trends

This study revealed that spring AI in the LRB experienced an abrupt change in 1974 within the 33° N–34° N region. Studies have indicated that the climate transition from arid to humid conditions on the Tibetan Plateau first occurred in 1974 [68], consistent with our findings. The magnitude of the abrupt change in spring AI was consistently higher in the northern part than that in the southern part, with the variability increasing progressively with increasing latitude. This indicates that the LRB has become increasingly humid during spring in recent decades, with the northern region showing a more pronounced humidification trend. This phenomenon may be attributed to increased spring precipitation due to climate warming [34]. Simultaneously, spring warming in the northern region may have caused earlier snowmelt, thereby increasing the surface water availability and promoting the humidification process [60]. Additionally, the LRB exhibits a longitudinal distribution pattern, where higher latitudes correspond to higher elevations and closer proximity to the Tibetan Plateau. Under the context of global warming, the Tibetan Plateau demonstrates heightened sensitivity and responsiveness to global climate change [4], resulting in more pronounced abrupt changes in aridity in basin areas adjacent to the Tibetan Plateau.

4.5. Consideration of Elevational Factors in Latitudinal Analysis

The Lancang River Basin exhibits a north–south longitudinal distribution, with elevations generally increasing with latitude throughout the basin. This natural topographic-latitudinal correspondence enables the 0.5° latitudinal zoning method to effectively capture elevational gradient effects implicitly. Each latitudinal belt roughly corresponds to different elevation ranges; therefore, latitudinal analysis actually integrates the influence of elevational factors on the spatial distribution of the aridity index. Compared with traditional elevation-based analyses using station observations, the latitudinal analysis method demonstrates distinct advantages when processing gridded reanalysis data. Gridded data undergo spatial interpolation and smoothing processing, and point-by-point elevational analysis may introduce uncertainties, whereas averaging within latitudinal belts can effectively reduce data noise and provide more robust regional climate signals.

In the Lancang River Basin, due to the high consistency between topographic structure and latitudinal gradient, latitudinal analysis can simultaneously capture both latitudinal climate variations and elevational effects, providing a scientifically sound framework for basin-scale aridity change analysis. We acknowledge that the current study has limitations in explicit elevational analysis. Future research could consider combining higher-resolution topographic and climate data to conduct specialized elevation-based zoning analyses, particularly for topographically complex sub-basins, which would provide deeper insights into elevation-climate interactions.

5. Conclusions

We utilized high-resolution reanalysis data and employed the AI established by the UNEP to define climate types. We further analyzed the spatiotemporal distribution characteristics of AI in the LRB and its relationship with latitude through linear slope analysis, the M–K trend test, and Pettitt change point detection. The main conclusions drawn from this study are as follows:

(1)

The significant increase in spring AI in the LRB, coupled with non-significant changes in summer, autumn, and winter, resulted in a non-significant increasing trend in annual AI, indicating that the basin is becoming more humid. The dry-wet variations in the basin exhibit distinct seasonal differences.

(2)

The AI-latitude relationship in the Lancang River Basin demonstrates pronounced seasonal and spatial heterogeneity. South of 24.75° N, the relationship between AI and latitude consistently showed a negative correlation, whereas a consistently positive relationship was observed north of 30.5° N. In the central basin region, spring–winter and summer–autumn seasons displayed contrasting relationships. During spring and winter, approximately 27.25° N served as a boundary where AI showed opposite north–south relationship patterns: a positive correlation south of 27.25° N and a negative correlation north of 27.25° N. During summer and autumn, AI in the central basin did not exhibit this transitional pattern, with summer and autumn showing consistently negative and positive correlations, respectively.

(3)

The AI trend changes in the Lancang River Basin exhibit significant spatial differentiation characteristics. Using 27.5° N as the boundary line, the northern region shows a humidification trend that intensifies with increasing latitude. In seasonal variations, spring shows humidification across the entire basin, while summer and autumn exhibit a gradient change from aridification to humidification from south to north, with winter displaying the most significant north–south differentiation. Overall, the basin demonstrates a general humidification trend, with faster humidification rates in high-latitude areas closer to the Tibetan Plateau, indicating that latitude is an important geographic factor affecting dry-wet changes in the Lancang River Basin.

This study revealed the relationships between AI spatial distribution, trends, trend magnitudes, and latitude in the LRB, reflecting the complex response of the basin’s climate system to global change. Therefore, when formulating basin water resource management strategies and climate change adaptation measures, it is important to fully consider the differential characteristics across different latitudinal zones and seasons to enhance the effectiveness of water resource management and ecological protection. However, this study focused on the variations in aridity and its relationship with latitude, without considering other meteorological factors that influence aridity, such as temperature and wind speed. Future research should comprehensively examine the effects of these meteorological variables on aridity, quantify the contribution of each meteorological factor to aridity index changes, and provide deeper insights into the driving mechanisms underlying aridity index variations.