Multiscale Spatiotemporal Dynamics of Drought within the Yellow River Basin (YRB): An Examination of Regional Variability and Trends

Drought, a pervasive and recurrent natural hazard, profoundly impacts diverse facets of human endeavors globally [1,2]. The Intergovernmental Panel on Climate Change, in its Sixth Assessment Report, emphasizes that climate change is intensifying drought conditions worldwide through elevated temperatures and shifting precipitation patterns [3]. Particularly, in China, drought emerges primarily due to inadequate precipitation, amongst other contributing elements. Drought’s hallmark characteristics include its high frequency, prolonged duration, and extensive impacts. It represents a cumulative process where a persistent deficit in regional surface water, primarily influenced by precipitation and evapotranspiration, takes a central role amidst various meteorological factors [4,5]. The frequent and enduring nature of drought not only precipitates significant economic losses but also triggers a cascade of adverse effects such as acute water shortages, accelerated land desertification, and an increased incidence of sandstorms and dust storms [6,7,8]. Consequently, timely and precise drought monitoring is imperative for the judicious management of regional water resources, bolstering socio-economic progress, and fostering the sustainable development of the ecological environment.

Drought manifests in four principal forms: socioeconomic, agricultural, hydrological, and meteorological. Typically, the onset of other drought types is heralded by a meteorological drought, which usually precedes them [9,10,11]. Typically, a diminution in precipitation precipitates the onset of a meteorological drought. This sequence of events invariably leads to a reduction in soil moisture levels, thereby impeding the effective irrigation of crops and consequently instigating agricultural drought due to diminished crop yields. When it progresses to a certain stage, such conditions can culminate in a decrease in river flows or reservoir storage levels, thereby triggering a hydrological drought. These cascading effects extend to the socio-economic realm, where the fundamental water requirements of society and the economy are unmet, ultimately resulting in a socio-economic drought characterized by substantial losses in human life, property, and overall societal and economic stability. Consequently, an in-depth examination of meteorological drought is instrumental in enhancing our knowledge and understanding of various other drought typologies.

The meteorological drought index serves as an indispensable instrument for the monitoring, early warning, and quantitative assessment of drought severity. Prominent indices such as the Standardized Precipitation Index (SPI) [12,13,14], Standardized Precipitation Evapotranspiration Index (SPEI) [15,16,17,18,19,20,21,22], and the Palmer Drought Severity Index (PDSI) are extensively utilized in current drought studies [22,23,24,25,26,27,28]. Due to their distinct foundational backgrounds and drought-inducing factors, these indices vary in regional suitability. Consequently, the accurate analysis of regional drought characteristics hinges on the judicious selection of appropriate drought indices. The PDSI, introduced by Palmer in 1965 [29], represents a sophisticated meteorological drought index. It integrates a range of critical factors, including historical precipitation patterns, water supply–demand equilibrium, and both actual and potential evapotranspiration rates. Despite its comprehensive approach, the PDSI falls short in capturing the multi-dimensional characteristics of drought phenomena. In an enhancement to the original model, the self-calibrating Palmer Drought Severity Index (scPDSI) has been developed [23,30,31,32,33,34,35]. This advanced index dynamically adjusts PDSI parameters in response to specific climatic data from individual stations, thus offering a more tailored and effective tool for assessing local climate conditions [36]. Furthermore, the SPEI further broadens the scope by incorporating variables such as temperature and precipitation, addressing the multi-scale dimensions of drought, and providing a more holistic understanding than the SPI [37,38]. Acknowledging the intricate genesis and diverse factors influencing drought, our study extensively employs both scPDSI and SPEI indices to meticulously investigate the spatial and temporal dynamics of drought within the designated study region.

Although interpolation analysis on the station-based SPEI values was conducted earlier in our study, it is important to note that such interpolated outcomes are susceptible to variations induced by topographical features and spatial station distribution. To mitigate these effects and enhance the robustness of our analysis, we have conducted a comparative assessment, juxtaposing SPEI values calculated at station level with those derived from 0.5-degree-resolution gridded SPEI datasets over a temporal scale of 3 months. Figure 10 presents the time series analysis of SPEI values computed from station data juxtaposed with those obtained from gridded datasets, spanning from 1961 to 2017 in the YRB. The depicted correlation coefficient of 0.93 signifies a robust congruence between the two datasets, substantiating the efficacy of gridded SPEI values in capturing the hydrological dynamics of the region. This high fidelity suggests that gridded SPEI data are well suited for future evaluations of drought conditions within the basin.

Following the temporal series analysis of SPEI values computed from station-based measurements versus gridded datasets, we sought to elucidate the spatial applicability of gridded data within our study domain. This involved leveraging station data to extract corresponding values from the gridded dataset at precise locations. Correlation coefficients were then calculated for each station in relation to its respective gridded data point. Employing the kriging interpolation technique, this process culminates in Figure 11, which illustrates the spatial correlation of gridded SPEI data across the study area. The analysis depicted in Figure 11 reveals a discernible spatial gradient in correlation levels, extending from the northwest towards the southeast, with an increasing trend in correlation coefficients that range between 0.61 and 0.88. Within the delineated sub-regions of the study area, the correlation levels are ranked as follows: BH (0.865) > SH (0.858) > LS (0.845) > HL (0.817) > IF (0.749) > LH (0.736) > AL (0.729) > LL (0.713). This demonstrates a commendable degree of correlation between the Standardized Precipitation Evapotranspiration Index (SPEI) values derived from station data and those obtained from grid-based datasets. Consequently, for subsequent drought assessments at the grid scale, grid-based data emerge as a viable analytical resource. Nevertheless, it is pertinent to note that the resolution of grid data currently prevalent for drought assessments is comparatively coarse, necessitating enhancements for precise evaluations in smaller catchments or scenarios demanding higher resolution data.