Spatiotemporal heterogeneity in meteorological and hydrological drought patterns and propagations influenced by climatic variability, LULC change, and human regulations

This study aims to quantify meteorological–hydrological drought propagations and examine the potential impacts by climatic variability, LULC change (LULC), and human regulations. An integrated observation-modeling framework quantifies drought propagation intervals and assesses mechanisms influencing hydrological droughts. Meteorological droughts are characterized using the Standardized Precipitation Evapotranspiration Index (SPEI), and hydrological droughts are assessed through the Standardized Streamflow Index (SSI) across diverse climatic zones. Cross-correlation analysis between SPEI and SSI time series identifies the lag time associated with the highest correlation as the drought propagation interval. Mechanisms are investigated via a coupled empirical-process modeling framework incorporating the Soil and Water Assessment Tool (SWAT). Discrepancies between simulated and observed SSI time series help quantify the extent of human regulation impacts on hydrological drought characteristics and propagation. The Yellow River Basin (YRB), divided into six subzones based on climate characteristics, is selected as the case study. Key findings include: (1) Meteorological droughts were extremely severe across most YRB during the 1990s, while the 2000s showed some mitigation primarily due to precipitation increases. (2) Hydrological droughts and propagation times from meteorology to hydrology demonstrated substantial spatiotemporal variability. In general, summer propagation times were shorter than other seasons. (3) Propagation times were shorter in arid regions with cropland or built-up land cover versus grassland and woodland, while the reverse held for humid regions. (4) Human regulations prolonged propagation times, likely due to reservoir regulations designed to overcome water deficits. While the YRB is the focus of this paper, the methodologies and findings are applicable to other regions worldwide to enhance drought forecasting and water resource management. In various hydrological and climatic contexts worldwide.

assessment [38][39][40][41][42][43] , knowledge gaps remain exploring the mechanistic drivers of climatic variation, LULC change, and human regulation on drought evolution and propagation.As an intensely managed basin with inherent aridity and hydro-climate diversity, the YRB provides an ideal case study for investigating how meteorological drought signals propagate into hydrological drought events.By revealing the processes governing drought development and propagation in this complex human-natural system, improved hydrological drought monitoring and early warning systems can be developed by Water Resources Management Center and Drought Prevention and Disaster Reduction Department.
The primary objectives of this study are three-fold: (1) Investigate multi-scale temporal and spatial evolution dynamics of meteorological and hydrological droughts; (2) Quantify propagation times from meteorological to hydrological drought; and (3) Reveal potential influence mechanisms of climatic factors, land use/land cover (LULC) changes, and human regulations on hydrological drought and propagations.

Methodologies Meteorological drought index
Selecting or developing appropriate drought indices is crucial for investigating complex drought evolution dynamics, propagations, and governing mechanisms.Numerous drought indices have been proposed, including meteorological indices such as the Standardized Precipitation Index 44 (SPI), Precipitation Anomaly Percentage 45 (PAP), Standardized Precipitation Evapotranspiration Index 46 (SPEI), and Reconnaissance Drought Index 47 (RDI).The SPEI and RDI are advantageous for capturing meteorological drought as they incorporate precipitation and temperature as key driving factors 48,49 .Compared to the RDI, the SPEI is more sensitive to environmental fluctuations with clearer evaluation criteria, assessing deviations in precipitation minus evapotranspiration from normal conditions 50 .
Therefore, this study utilizes the SPEI as the meteorological drought index.SPEI fits a distribution function to precipitation-evaporation differences, enabling sensitivity to environmental changes and effective correlations with hydrological drought.Details on SPEI calculation and classification are available in Shang et al. 50and Table 1.
The water balance equation, based on the principle of conservation of mass, provides reasonably reliable estimates of actual evapotranspiration (AET) at the catchment scale by calculating AET as the residual of precipitation and runoff terms.However, it is limited by the availability and accuracy of the other water budget components, and is sensitive to errors in precipitation and streamflow measurements.To overcome these limitations in quantifying the atmospheric moisture deficit, potential evapotranspiration (PET) is utilized to derive the Standardized Precipitation Evapotranspiration Index (SPEI) for drought analysis.
There are three primary methods for estimating PET-the Penman-Monteith 51 , Priestley-Taylor 52 , and Hargreaves equations 53 .The Penman-Monteith approach calculates PET based on aerodynamic and radiative factors, requiring extensive meteorological data inputs of wind speed, humidity, radiation, and temperature.In contrast, the Priestly-Taylor method utilizes radiation and temperature to determine PET responses to climate variability.The Hargreaves equation only requires temperature and extraterrestrial radiation, providing reliable PET estimates in arid regions but performing poorly in humid environments or with significant wind.Critically, Hargreaves does not account for moisture limitation effects on PET.Among the available options, the Penman-Monteith equation has the most rigorous theoretical basis and fully considers the multiple factors influencing evapotranspiration like atmospheric flow and environmental interactions, providing robust PET estimations across regions and climate conditions.Therefore, this study adopts the Physically-based Penman-Monteith method to calculate PET for deriving SPEI drought indices.

Hydrological drought index
Appropriate meteorological and hydrological drought indices are foundational for achieving the study objectives.For characterizing hydrological drought, several standardized indices have been developed including the Surface Water Supply Index 54 (SWSI), Palmer Hydrological Drought Index 55 (PHDI), Standardized Runoff Index 56 (SRI), and Standardized Streamflow Index 57 (SSI).Among these, the SSI is advantageous due to simpler calculation and lower data requirements 7 .The SSI effectively depicts water storage dynamics in rivers and lakes, enabling assessment of climate and human impacts.The streamflow simulated by the swat model in this study is only affected by climate and land use changes, while the observed streamflow is influenced by climate change and human activities (including land use changes and human regulation) Therefore, simulated streamflow-based SSI is utilized to characterize hydrological droughts under climate and LULC change scenarios, while observed streamflow-based SSI represents hydrological droughts under coupled climate change and human activity scenarios.Details on SSI computation are available in Svensson et al. 57 , with drought classification matching that of the SPEI (Table 1).

SWAT model
Appropriate model selection and performance assessment provides confidence in subsequent investigations of complex drought dynamics.The Soil and Water Assessment Tool (SWAT) is widely applied for simulating complex hydrological processes under climate variation and land use/land cover (LULC) change scenarios [58][59][60][61][62] .
Hence, the SWAT model is selected to simulate the hydrological process in this study.
Because the SWAT hydrological model requires 2-3 years of flow data to preheat the model in the early stages of operation, eliminating the default value of model parameters as zero.Therefore, the research period in this study is from 1968 to 2010.To conduct research in different time periods, the change point analysis with Mann-Kendell trend test on the naturalized streamflow time series from 1968 to 2010 at five hydrological stations.The results showed that there was one change point in the naturalized streamflow time series that occurred in 1990.Hence, based on the change point in 1990, the study period (1968-2010) is divided into baseline period (1968-1990) and impact period (1991-2010).The impact period is further divided into two periods of 1991-2000  (1990s) and 2001-2010 (2000s) according to intergenerational differences.
To accurately represent variable underlying surfaces, this study implements the time-varying parameter approach in SWAT proposed by Li et al. 17,63 calibrating the model for three periods (1968-1990, 1990s, 2000s)  using corresponding LULC maps.This enables aggregated characterization of distinct LULC impacts on drought propagation.
As reported in Li et al. 63 , NSE and R 2 exceeded 0.6 and 0.7 for all periods in both calibration and validation, and PBIAS remained within 25% (Table 2).Therefore, SWAT reasonably simulated hydrological processes under climate and LULC changes across the six subzones of the YRB, enabling robust analysis of spatiotemporal drought propagations and mechanisms.The Tangnaiai, Lanzhou, Toudaoguai, Huaxian and Huanyuankou in Table 2 are the hydrological stations in the YRB, and their locations are marked in Fig. 3.

Drought propagation time
Meteorological drought indices (e.g.SPEI) are based on precipitation and evapotranspiration, representing deficits in the atmospheric water balance.Hydrological drought indices (e.g.SSI) quantify deficits in surface waters, groundwater, and streamflow that constitute the water sources for human use.As meteorological water deficits propagate through terrestrial systems, it takes time for reduced precipitation and moisture supply to manifest as deficiencies in rivers, lakes, reservoirs and groundwater.Hence, comparing correlation analysis between meteorological and hydrological drought indices provides insight into propagation time from precipitation-evapotranspiration imbalances to resultant surface water, groundwater, and streamflow deficits.This study cross-correlates 1-month Standardized Streamflow Index (SSI-1) time series with 1-to 24-month accumulated Standardized Precipitation Evapotranspiration Index (SPEI) scales using Pearson correlation coefficients.The SPEI accumulation scale exhibiting maximum correlation with SSI-1 is denoted SPEI-n, where n serves as an indicator of drought propagation time (Fig. 1).Longer SPEI accumulation scales strongly correlated with SSI-1 signify extended propagation times, and vice versa.The SPEI-n series for month i is then extracted to represent the meteorological drought conditions n months prior, enabling quantification of propagation lags from meteorological to hydrological drought events.
To clearly clarify the design ideas in of this study, the research flow chart is provided in Fig. 2.

Case study of the Yellow River Basin
The Yellow River Basin (YRB), located between 95°-119°E and 32°-41°N, is the second largest river basin in China and sixth largest globally, with a length of 5464 km and drainage area of 752,443 km 2 (Fig. 3, generated using ArcMap 10.8 software).Originating on the Tibetan Plateau, the basin meanders eastward through several provinces before discharging into the Bohai Sea.
The YRB has a large area with varied topography, resulting in significant spatial precipitation differences ranging from 1021 mm in the rainy southeast to 123 mm in drier northwest areas 67,68 Mean annual temperatures vary from − 4 to 14 °C between upper and lower reaches.The YRB renowned for its complex climatic patterns, diverse land use types, and intensive human activities.Hence, the YRB was selected for this study due to its unique environment that allows for the examination of interactions between meteorological and hydrological droughts and the impact of human activities.Water resource allocation and management within the basin are critical for ecological security and socio-economic development in China and the broader East Asian region.Therefore, research on the YRB not only has local significance but also offers insights for global water resource management and drought early warning and defense.To enable analysis of spatial drought dynamics, the basin is delineated into six subzones based on climate characteristics (Fig. 3), with details provided in Li et al. 63 .The hydrological stations such as Tangnaihai, Lanzhou, Toudaoguai, Huaxia and Huayuankou are pivotal for monitoring water flow and quality, providing essential data for flood control, water resource management, and environmental protection.They also serve as key points for dividing the YRB into manageable subzoness for detailed drought analysis.

Datasets
Daily meteorological data spanning 1966-2010 was acquired from 121 stations across the Yellow River Basin (YRB) (Fig. 3), including precipitation, humidity, wind speed, solar radiation, and maximum/minimum temperatures, obtained from the National Climate Center of China.Monthly naturalized and measured streamflow data for the same period was collected from five hydrological stations (Fig. 3) under the administration of the Yellow River Conservancy Commission.Naturalized streamflow excludes human regulation influences and was used for calibrating SWAT model hydrological parameters.Elevation data was extracted from the 30 m resolution NASA Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM).Soil and land use/land cover (LULC) maps for 1980, 2000, and 2010 were obtained from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences.

Meteorological drought evolution characteristics
Standardized Precipitation Evapotranspiration Index (SPEI) timescales reflect abnormal precipitation and evapotranspiration states over different antecedent accumulation periods.For instance, SPEI-1 and SPEI-3 indicate monthly and seasonal moisture anomalies, respectively.Multi-scale SPEI series (1-24 month accumulations) were thus calculated across YRB subzones from 1968-2010 to examine meteorological drought distribution spatially and temporally.Results are presented in Fig. 4.
Analysis of Fig. 4 reveals increasingly severe meteorological drought across most subzones prior to 2000, especially during the 1990s.However, partial drought relief occurred in the 2000s for many subzones.As an example, subzone D exhibited a shift from extreme 1990s drought to 2000s mitigation.Comparing average annual precipitation and evapotranspiration for the 1968-1990 baseline, 1990s, and 2000s provides insight into drivers.While evapotranspiration showed minimal change (468, 442, 459 mm), precipitation declined sharply from 571 mm (baseline) to 500 mm (1990s), before recovering slightly to 527 mm (2000s).This suggests the 1990s drought primarily resulted from dramatically reduced precipitation, while 2000s easing stemmed from mi-nor precipitation recovery.Figure 4 elucidates spatiotemporal meteorological drought dynamics across the climatically diverse Yellow River Basin, highlighting intense 1990s drought followed by 2000s moisture deficit mitigation for most subzones.Under the climate-LULC change scenario (Fig. 5a), the severity of hydrological drought in the cold Plateau region (e.g., subzone A) progressively increased over time.However, in the transition region between the midtemperate continental and the plateau cold climates (subzone B) and the arid region (subzone C), the hydrological drought conditions improved after 1990.In contrast, since 2000, severe hydrological droughts were observed in the midstream (subzone D and subzone E), while the downstream region (subzone F) experienced some relief after 2005.
When accounting for the impacts of human regulations, such as reservoir storage, irrigation, and industrial and domestic water usage (Fig. 5b), it was observed that, apart from the significant alleviation of hydrological drought in subzone E after 1990, the severity of hydrological drought in the other subzones progressively worsened over time.Particularly since 1990, the occurrence of hydrological drought became more frequent.After 2000, the arid region (subzone C) and the humid region (subzone F) experienced particularly severe hydrological droughts.to 24 months and SSI-1 under climate-LULC change scenarios across the six subzones within the YRB during the baseline period, the 1990s, and the 2000s.Analyzing the heat maps, it becomes evident that during the baseline period, the correlation between meteorological and hydrological droughts was notably higher in the upstream region compared to the middle and downstream regions.However, in the 1990s and 2000s, this correlation exhibited minimal spatial variability across the subzones.Temporally, the correlation between meteorological and hydrological droughts in the upstream region of the YRB initially declined and then increased.In contrast, the middle and downstream regions displayed a consistent increasing trend.This trend could be attributed to the rapid urban development observed in the middle and downstream regions of the YRB since the 1990s.Regarding propagation time, noticeable spatial variations were observed within the YRB.Generally, the upstream region exhibited longer propagation times compared to the middle and lower reaches.Temporally, there were negligible differences in propagation time between the baseline period and the 1990s.However, in the 2000s, the propagation time experienced a significant prolongation.

Propagation time over time
The propagation time from meteorological to hydrological drought varied substantially over time and space.In the humid subzone F, propagation times fluctuated most over time, providing an example to preliminarily illustrate temporal variations and analyze key influences.During the baseline period, subzone F propagation times were relatively short, concentrated in 1-5 months.Roughly, summer propagation took slightly longer at 4-5 months, likely due to higher summer rainfall frequencies facilitating runoff generation.Compared to the baseline, April, June, August and October propagation times lengthened to 15-21 months in the 1990s.In the 2000s, March to August propagation increased further to 19-23 months.

Propagation time in space
Spatially, substantial differences emerged in propagation times, especially summer, between the arid subzone C and humid subzone F in the 2000s.The subzones C and F provide examples to preliminarily illustrate spatial variations in meteorological-to-hydrological drought propagation and discuss potential key influences.Compared to the subzone F, subzone C propagation times were relatively short, concentrated in just 1-6 months.

Modeling experiments
Based on the preliminary analysis of the propagation mechanism, it is evident that cropland, woodland, grassland and built-up areas have significant influences on the propagation time from meteorological to hydrological drought across the six subzones with differing climate characteristics.In this section, four modeling experiments (EPMC, EPMW, EPMG and EPMB) are conducted using the SWAT model for the latest decade of the impact period (2000s) to examine how these four LULC types affect drought propagation time in the various climate subzones of the YRB.The setup of the four experiments is identical except for differences in the LULC maps representing four extreme LULC scenarios, where all LULC types in the YRB are transformed completely into cropland (EPMC), woodland (EPMW), grassland (EPMG), and built-up areas (EPMB).All other inputs such as DEM and soil data remain constant across the four experiments.Based on the simulated streamflow from these experiments, the SSI at a one month time scale (SSI-1) is calculated for the 2000s across the six YRB sub-zones.The correlations of SPEI-n (n = 1-24 months) with SSI-1 are analyzed for the four modeling experiments as shown in Fig. 7.By comparing the strongest SPEI-n and SSI-1 correlations across the four modeling experiments and six subzones, we can investi-gate how climate and LULC characteristics influence the propagation time from meteorological to hydrological drought in the YRB, thereby further elucidating the physical processes governing the propagation mechanism.
Figure 7 shows high correlations between meteorological and hydrological droughts across the four modeling experiments in the YRB, except for the arid subzone C. Regarding propagation times, noticeable spatial differences emerged among the four experiments, indicating significant impacts of both climate and LULC characteristics.Generally, summer propagation times were shorter across subzones and experiments, suggesting meteorological drought readily propagates to hydrological drought in summer irrespective of climate zone or LULC type.This is likely due to higher summer temperatures increasing evaporation rates and accelerating meteorological-to-hydrological drought propagation.
More broadly, the spatial propagation time variations highlight how local climate and LULC influence meteorological drought propagation into hydrological drought in complex basins like the YRB.The modeling experiments reveal propagation times lengthen under wetter climate regimes and with increased cropland or built-up LULC types.These insights can inform monitoring and management strategies to improve meteorological-tohydrological drought prediction in different climate and land use contexts.

Influence of human regulations on propagation time
Figure 8a, b, and c display heat maps of the correlations between SSI-1 and SPEI accumulation scales of 1-24 months under climate-human activity scenarios across six sub-basins in the Yellow River Basin during the baseline period, 1990s, and 2000s, respectively.The results show pronounced spatial differences in the correlations between meteorological and hydrological droughts when considering human regulations such as reservoir storage, irrigation, and industrial and domestic water use.Compared to Fig. 6, the correlations decreased substantially in most sub-basins, especially in the middle and lower reaches.The propagation time exhibited an overall prolonging trend but was more dispersed within the year.This suggests that human regulations weaken the link between meteorological and hydrological drought and add complexity to the hydrological processes in the YRB.www.nature.com/scientificreports/ to determine how human regulations affect the propagation time from meteorological to hydrological drought.
In contrast to Figs. 6a-c, 8a-c showed notably decreased correlation, likely due to the water storage provided by two major reservoirs upstream-Longyangxia and Liujiaxia.

The complex interplay of climate change, land use change, and human regulation on drought propagation mechanism
Temporal changes primarily relate to climate and land use/land cover (LULC) shifts in recent decades.Precipitation and temperature changes in spring and summer, along with LULC transitions quantified using a transition matrix, were analyzed for the subzone F over the past 30 years.The results are shown in Tables 3 and 4, respectively.This analysis reveals how climate variations and LULC changes over recent decades have influenced propagation time lengthening in the humid subzone.Table 3 shows little change in average spring and summer precipitation and temperature over time in the subzone F, indicating climate change may not be the primary driver prolonging propagation times in these seasons.The LULC transition matrix in Table 4 reveals that from 1980 to 2010, substantial woodland and grassland area trans-formed into cropland, at 17.81% and 31.32%respectively.Meanwhile, 11.71% of cropland converted to built-up area.These transitions likely accelerated surface runoff generation in the hydrological cycle, as croplands and built-up areas have lower surface resistance and faster flow velocities compared to woodlands and grasslands when it rains.Consequently, the LULC changes appear to have prolonged meteorological-to-hydrological drought propagation times in the humid subzone, especially during rainy spring and summer seasons.
This discrepancy likely relates to contrasting climate and land surface characteristics.The arid subzone C in the Loess Plateau receives under 300 mm annual precipitation, while the humid subzone F exceeds 600 mm.Additionally, the subzone C is dominated by grasslands, whereas croplands prevail in the subzone F (Fig. 9a and  b).Croplands clearly generate surface runoff more readily following frequent rainstorms compared to grasslands in humid regions.Consequently, different climate regimes and land use likely drive longer meteorological-tohydrological drought propagation times in the humid cropland versus the arid grassland.
More interestingly, substantial propagation time differences emerged across climate zones despite similar underlying land surfaces (LULC types), especially between arid and humid regions.For instance, propagation times in the arid subzone C were shorter in the cropland (EPMC) and built-up (EPMB) experiments versus the woodland (EPMW) and grassland (EPMG) experiments.Conversely, the humid subzone F showed longer propagation times under cropland and built-up versus woodland and grassland experiments.
These differences likely stem from how underlying surfaces affect surface water and groundwater dynamics differently in arid versus humid climates.The EPMC, EPMB, EPMW and EPMG experiments reflect cropland, built-up, woodland and grassland land covers, respectively.Clearly, croplands and built-up areas generate more surface runoff when it rains compared to woodlands and grasslands.This runoff can temporarily delay hydrological drought onset in rainy humid regions like the subzone F, effectively prolonging meteorological-to-hydrological drought propagation.This aligns with earlier results showing lengthened propagation times when wood-lands/ grasslands converted to croplands in the subzone F.
In contrast, woodlands and grasslands better retain soil water and recharge groundwater versus croplands and built-up areas.This alleviates hydrological drought in drier regions with less rainfall.Consequently, woodlands and grasslands likely cause slower precipitation deficit propagation into hydrological drought in arid regions like the subzone C compared to croplands and built-up areas.Reservoir operations can smooth hydrographs, resulting in higher low flows and lower peak flows, thus disturbing the natural rainfall-runoff process.Moreover, propagation time, especially in the 2000s, was prolonged under human regulation, primarily attributable to the effective regulation of reservoirs often designed to mitigate water deficits, thereby prolonging the meteorological to hydrological drought propagation time to some degree.
Overall, this research underscores the intricate relationship between climate change, land use change, and human regulation in influencing drought dynamics.It reveals that while climate change sets the broader stage for drought occurrence, land use practices and human regulations, such as reservoir management, play pivotal roles in modulating drought propagation times across different climatic zones.In humid regions, urbanization and agricultural expansion can increase surface runoff, delaying the onset of hydrological droughts and extending drought cycles.Conversely, in arid zones, natural vegetation like forests and grasslands are crucial in maintaining soil moisture and supporting groundwater recharge, thus mitigating hydrological droughts and decelerating the transition from meteorological to hydrological drought.The operation of reservoirs introduces another layer of complexity, as it can disrupt natural hydrological patterns, sometimes prolonging droughts by altering rainfallrunoff processes.This multifaceted impact is most evident in the way land use changes and human regulatory measures, such as the strategic operation of reservoirs, have been found to significantly influence the temporal and spatial characteristics of drought propagation, highlighting the need for integrated management approaches that consider the diverse effects of these factors.

Uncertainty estimation the SWAT model
While comparing simulated and observed streamflow enables evaluating SWAT model performance, uncertainties can arise from limitations in capturing complex hydrological processes 69 .For example, SWAT relies on calibrated parameters to represent basin-specific water budgets, introducing subjectivity and constraints from observational data availability in parameter selection and validation.Additionally, conceptual model structural uncertainty stems from the simplification of intricate hydrological process representations 70 .Input data uncertainty can also emerge from errors in measurement, resolution, or spatial interpolation 9,71 .
To reduce such uncertainties in future work, several approaches could improve scientific understanding of the intricate YRB water cycle dynamics.Implementation of flexible Bayesian frameworks that allow model parameters and process representations to vary spatially and evolve temporally, based on assimilation of integrated meteorological, ecological, agricultural, and hydrological datasets, may overcome structural deficiencies.Tight coupling of subsurface and groundwater compartments could enhance representation of base-flow and water table dynamics governing hydrological memory and drought persistence.Additionally, dynamic model parameter estimation methods and fusion with complementary models can help constrain uncertainties.

Application and limitation of the research results
The findings in this study can advance understanding of how climate types, land surface characteristics, and human regulations like water resource development influence drought evolution and propagation across complex basins such as the YRB.Such insights have policy relevance for land use planning and water management strategies aimed at facilitating effective drought prediction and mitigation.The results highlight the utility of considering local climate, land cover, and human factors when developing integrated approaches to monitor and manage droughts in complex basins like the YRB.
However, it is important to acknowledge key limitations of this study.Namely, the current analysis lack of considering potential nonlinearities in drought propagation processes.In reality, the propagation from meteorological to hydrological drought involves nonlinear propagation characteristics in addition to linear connections.Hence, a promising avenue for future work is to elucidate nonlinear propagation mechanisms linking different drought types under changing climate and human pressures.Analyses quantifying nonlinear drought impacts and feedbacks could provide deeper scientific understanding to advance predictive capabilities and inform adaptive water management strategies.Coupling empirical analyses with process-based models may aid in disentangling climate change and human controls on coupled meteorological-hydrological drought dynamics across complex basins like the YRB.

Conclusion
Investigating linkages between meteorological and hydrological drought events and diagnosing drivers governing their propagation lay groundwork for developing integrated forecasting frameworks leveraging meteorological warnings for predicting emergent hydrological extremes.This study uses an observation-modeling framework integrating the Standardized Precipitation Evapotranspiration Index (SPEI) and Standardized Streamflow Index (SSI) under observed and simulated scenarios to quantify meteorological-to-hydrological drought connections and isolate propagation mechanisms.With the intensely-managed YRB as a testbed, key conclusions regarding spatiotemporal drought dynamics are as follows:

Figure 1 .
Figure 1.Schematic diagram of propagation time from meteorological drought to hydrological drought.

Figure 2 .Figure 3 .
Figure 2. The research flow chart of the study.

Figure 4 .
Figure 4. Spatiotemporal distributions of monthly SPEI across YRB subzones at varying accumulation scales, with darker red shading denoting more intense meteorological drought conditions.

Figure 5 .
Figure 5. Temporal and spatial evolutions of monthly SSI at different accumulation time-scales under climate-LULC change scenario (a), and under climate-human activities scenario (b) in the YRB.Darker red color implies more severe hydrological droughts.
https://doi.org/10.1038/s41598-024-56526-zwww.nature.com/scientificreports/Subzone C exhibited relatively greater variability in correlation and propagation time compared to other subzones when considering the influence of human regulations.Therefore, subzone C was analyzed as a case study

Figure 6 .Figure 7 .Figure 8 .
Figure 6.Correlations of SPEI accumulation scales of 1-24 months with SSI-1 under climate-LULC change scenarios over six subzones in the YRB during the baseline period (a), the1990s (b), and the 2000s (c).The black dot denotes the strongest correlation, and the corresponding meteorological cumulative scale is the propagation time.

Figure 9 .
Figure 9. Area percentages of LULC types in subzone C (a) and subzone F (b).

( 1 )
Meteorological drought severity rose from 1968 to 2000 before stabilizing post-2000, apart from sub-humid zones where agricultural expansion has maintained atmospheric moisture losses.Hydrological drought trends showed higher climate dependency, with alleviation upstream but continued intensification downstream.(2)Summer propagation times from meteorological to hydrological drought events are shorter than other seasons across climate zones, attributed to heightened evaporation rates accelerating atmospheric moisture deficit transfer.(3) Longer propagation time arise in humid croplands and built-up regions relative to humid grasslands/ woodlands because the former's surface runoff hinders onset of rainy season hydrological droughts.Arid croplands and built-up areas contrastingly have shorter meteorological-hydrological drought propagation time than arid grasslands/woodlands due to the latter's soil water retention slightly buffering hydrological drought.(4) Human regulations via extensive water storage infrastructure prolong basin-wide propagation times, especially post-2000, as controlled reservoir releases overcome natural water limitations and delay meteorological drought realization as hydrological droughts.https://doi.org/10.1038/s41598-024-56526-z

Table 1 .
The classification of SPEI and SSI.

Table 2 .
Results of calibration and validation in three periods measured by the three metrics.

Table 3 .
Average precipitation and temperature in spring and summer in different decades.

Table 4 .
Transfer percentages of LULC area during of period of 1980-2010.