Previous Article in Journal
Seismic Performance Assessment of Gravity Dams for Urban Flood Risk Mitigation Using the Scaled Boundary Finite Element Method (SBFEM)
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Hydrology
Volume 12
Issue 8
10.3390/hydrology12080210
hydrology-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Human-Induced Shifts in Yellow River Flooding: Population Threshold Effects in the Loess Plateau’s Primary Sediment Source Area (934 CE)

by
Tao Huang
1,*,† and
Yabin Li
2
1
College of Soil and Water Conservation Science and Engineering, Northwest A&F University, Yangling 712100, China
2
Shenyang Academy of Environmental Sciences, Shenyang 110167, China
*
Author to whom correspondence should be addressed.
†
Current address: Advanced Institute of Natural Sciences, Beijing Normal University, 18 Jinfeng Road, Zhuhai 519087, China.
Hydrology 2025, 12(8), 210; https://doi.org/10.3390/hydrology12080210
Submission received: 2 July 2025 / Revised: 5 August 2025 / Accepted: 7 August 2025 / Published: 11 August 2025
(This article belongs to the Section Water Resources and Risk Management)
Browse Figures
Review Reports Versions Notes

Abstract

Flooding frequency in the lower Yellow River (YR) exhibited an abrupt increase post-934 CE, causing catastrophic societal disruptions. However, the quantitative relationship between this abrupt increase and the intensification of human activity in the midstream Loess Plateau (LP)’s Primary Sediment Source Area (PSSA) remains uncertain. This study systematically evaluates the threshold effects of human activities on YR flooding through multi-proxy historical records, GIS-based spatial analysis, and nonparametric statistical tests. The results show that from 934 to 1102 CE, the population density in the PSSA surged from 1.3 to 19.8 persons/km2 (a 14.2-fold increase, p = 0.005). A 2400-year-scale comparison using 934 CE as the breakpoint revealed that the mean population density in this region increased from 5.2 to 51 persons/km2 (a 9.8-fold increase). This dramatic population surge drove a 1.4-fold increase in the cultivation rate (from 8.6% to 20.5%), leading to a 5.4-fold rise in sediment yield (1.6 × 108 → 1.02 × 109 t/yr, p = 0.035), a 10-fold acceleration in downstream sedimentation rate (0.3 → 3.3 cm/yr, p = 0.001), and ultimately a 5.5-fold escalation in flooding frequency (from 1.6 to 10.4 events per 20 years, p < 0.0001). The study identifies 19.8 persons/km2 as the ecological pressure threshold. It proposes converting population density to ecological pressure equivalents adjusted for soil–water conservation coverage (e.g., terracing/afforestation). When the equivalent ecological pressure exceeds 19.8 persons/km2, pre-defined sediment control measures (e.g., tillage restrictions/afforestation mandates) should be enforced in the PSSA.

1. Introduction

The Yellow River (YR) basin (Figure 1), the cradle of Chinese civilization [1], underwent deforestation-driven agricultural expansion [2], which—by increasing sediment loads—necessitated large-scale levee construction in the lower reaches during the Warring States period (475–221 BCE) [3]. However, these anthropogenic modifications triggered persistent flooding, solidifying the river’s reputation as ‘China’s Sorrow’ [4]. Historical records document more than 1500 levee breaches and 26 major channel avulsions across the YR, showing a time-dependent escalation that severely endangered downstream urban populations and livelihoods [5,6]. From the 4th century CE onward, the lower YR’s alluvial fan apex—Kaifeng—suffered seven catastrophic flood events [7]. The 1642 CE flood killed approximately 340,000 people in Kaifeng, leaving only 2000 documented survivors [8,9]. Therefore, deciphering the drivers of YR flooding and designing adaptive management strategies are essential to reduce escalating flood risks in its lower reaches. This flooding mitigation effort directly contributes to Sustainable Development Goal 6 (Clean Water and Sanitation), which calls for sustainable management of water resources and ecosystems [10].
Since ancient times, the YR’s sediment problem has been recognized, with the Loess Plateau (LP, 640,000 km2, Figure 1) identified as its primary source from the Qing Dynasty (1644–1911 CE) [11]. The region’s thick (>100 m), unconsolidated loess deposits are highly erosion-prone. Summer monsoon rains trigger flash floods, delivering massive sediment loads to the North China Plain, where perched river channels significantly increase levee failure risks [12,13]. However, since 1949, comprehensive soil and water conservation measures have been implemented across the LP [14], while a cascade reservoir system has been established along the mainstem (comprising 36 planned projects with a total storage capacity of 100.7 billion m3 and long-term effective storage of 50.5 billion m3) [15]. Sediment load has dramatically decreased from 1.628 billion tons (1919–1975) to 0.179 billion tons (2007–2014) [16], significantly below historical levels [6]. This reduction has led to substantial decreases in both the frequency of prolonged and high-magnitude floods and the occurrence of levee breaches, which have now virtually disappeared. Nevertheless, uncertainties persist regarding future water–sediment dynamics because of unpredictable climate conditions and the anticipated plateauing of soil conservation benefits, leaving potential flood risks. According to the “Hydrological Method” and “Soil Conservation Method”, some scholars predict that YR sediment loads may rebound to 300–500 million tons over the next 30–50 years [17], potentially even higher (950 million–1 billion tons) [18]. To identify the root causes of flood risks in the YR, the currently available century-long hydrological observations prove insufficient for developing long-term watershed management solutions. However, detailed historical records of flooding events spanning over two millennia provide invaluable data and case studies for eliminating flood risks.
Based on the above understanding, despite comprehensive analyses linking downstream flooding to (a) LP human activities [1,19,20] and (b) climatic variability [21,22,23], the mechanisms driving the post-934 CE surge in YR levee breaches remain unresolved. The flooding frequency post-934 CE was approximately 10-fold higher than in previous periods. Identifying the precise drivers behind this dramatic shift is critical for effective flood risk management. Resolving this requires accounting for erosional heterogeneity across the LP, with the Primary Sediment Source Area (PSSA, 226,000 km2, 35.3% of the region, Figure 1) yielding >90% of total sediments [16]. We hypothesize that abrupt anthropogenic perturbations in the PSSA dominantly drove the downstream levee breach frequency shift, though the mechanistic linkages require further validation. Tan’s Vegetation Degradation Hypothesis posits that pre-10th-century progressive vegetation loss in the LP systematically reduced the water retention capacity of downstream tributaries and lakes [19]. However, this hypothesis lacks a quantitative causal chain linking anthropogenic forcing (e.g., cultivation rate) to flooding frequency. Li’s Climatic Shift Hypothesis posits that the Medieval Climate Anomaly (MCA, 900–1300 CE) increased regional humidity, thereby inducing the observed hydrological regime shift [23]. However, its conclusions rely exclusively on tree-ring-based precipitation reconstructions from the eastern Tibetan Plateau (not the LP), fundamentally undermining their validity. While both hypotheses offer valuable insights, they treat the LP as a homogeneous unit, overlooking its erosion heterogeneity. This limitation underscores the necessity of evaluating the role of the PSSA. Specifically, current research faces fundamental constraints: (1) failure to distinguish whether human activities or climate dominate as driving factors, and (2) lack of clarification on the downstream flooding’s unique sensitivity to PSSA environmental changes. Moreover, existing studies have failed to establish a quantifiable ecological stress indicator system to elucidate abrupt changes in YR flooding frequency—a critical knowledge gap that diminishes the policy relevance for YR management. To bridge this gap, our study pioneers the ‘Ecological Stress Threshold’ framework, which quantitatively evaluates the critical regulatory effects of key drivers (e.g., population density and cropland area) on downstream flooding frequency. This innovative approach not only deciphers the dynamic mechanisms behind environmental regime shifts but also establishes a novel research paradigm with direct policy implications.
This study builds on two Warring States period (475–221 BCE) transitions in the YR system: (a) intensified anthropogenic pressures and (b) systematic levee construction (which standardized flood documentation). By synthesizing multi-source historical records—archival texts and land-use reconstructions—we construct a 2400-year anthropogenic intensity sequence for the LP, integrating core metrics including population density and cultivation rate. Using GIS-based spatial analysis and nonparametric statistical tests, we (i) identify critical transition points in human activity intensity, and (ii) unravel the nonlinear linkages between PSSA anthropogenic forcing and downstream flooding frequency. This study provides direct evidence of historical human–environment coupling mechanisms, informing modern YR basin management via erosion-sensitive zoning—particularly through enhanced vegetation restoration standards in the PSSA.

2. Methods

2.1. Study Region

The YR, extending 5464 km with a drainage area of 795,000 km2, lies within China’s mid-latitudes (30–40° N) (Figure 1). Originating on the Tibetan Plateau, the YR flows through the midstream LP (640,000 km2) and downstream North China Plain (250,000 km2) before discharging into the Bohai Sea (Figure 1). The basin’s long-term mean annual precipitation is 474 mm, showing strong intra-annual variability—70% occurs in July–September. The YR displays a classic water–sediment source decoupling—53% of runoff originates from the upper reaches, while 90% of sediment derives from the midstream PSSA (226,000 km2; Figure 1) [18]. This water–sediment imbalance creates a stark downstream disparity: Huayuankou Hydrological Station records an annual runoff of 4.8 × 1010 m3, just 5% of the Yangtze’s, yet carries 3× the Yangtze’s sediment load (1.63 × 109 t), producing an extreme concentration of 37.3 kg/m3. The PSSA includes major tributaries (Jinghe, Beiluohe, and Fenhe) that drain the LP, exhibiting three defining features: (a) hilly–gully topography, (b) sparse vegetation cover, and (c) semi-arid climate [16].

2.2. Data Sources and Processing

2.2.1. Population Density

To reconstruct spatiotemporal population density patterns across the LP, we analyze historical demographic records spanning nine benchmark years (2, 140, 282, 609, 752, 1102, 1207, 1820, and 2010 CE). To resolve mismatches between historical and modern administrative boundaries, we implemented GIS-based spatiotemporal standardization through four steps: (1) Geo-referencing and vectorizing LP administrative maps from Tan’s Historical Atlas of China using ArcMap 10.8 (CGCS2000 coordinate system) [24]; (2) Calculating population density (D = P/A) for each historical administrative unit, with population (P) from Lu & Teng [25] and jurisdictional area (A) from digitized boundaries; (3) Linking density values as attributes to vectorized boundaries to generate annual distribution maps; (4) Applying an area-weighting algorithm (Wij = Aij/Ai, where Aij = intersection area between historical unit i and modern prefecture j, Ai = total area of i) to integrate all data into China’s 2020 prefecture-level grid—yielding a standardized population density database.
With only nine years of population density spatial data available, we derived a 2400-year population density time series for the PSSA through: (1) applying area-weighting to convert prefecture-level population density data into the PSSA boundaries. (2) A linear regression model was established between LP population density (x) and PSSA population density (y): y = 0.9574x − 4.2662 (R2 = 0.9185, p < 0.001). We applied this regression equation to estimate PSSA population density for 340 BCE and 934 CE, using contemporaneous LP-wide demographic data as inputs. The PSSA population ratio is calculated as the PSSA population divided by the total Loess Plateau (LP) population.

2.2.2. Cultivation Rate

Cropland area data for the LP were sourced from the Historical Database of the Global Environment (HYDE), maintained by the Netherlands Environmental Assessment Agency (NEAA) (URL: http://themasites.pbl.nl/en/themasites/hyde/download/index.html; accessed on 1 October 2024). This dataset provides unparalleled temporal coverage (10,000 BCE to 2000 CE), albeit with a coarse spatial resolution (5-arcminute × 5-arcminute grids). To obtain accurate cropland area time series for the LP and PSSA, we implemented the following workflow: (1) Initial area calculation was conducted by aggregating HYDE grid cells within China’s borders, LP, and PSSA boundaries, with areal conversion (grid area × cell count); (2) A regression model was established between HYDE-derived cropland area for China (x) and historical cropland area reconstructed by Fang et al. [26] (y): y = 0.4947x (n = 15, R2 = 0.63, p < 0.01); (3) Data adjustment was performed by applying the regression to HYDE-derived LP and PSSA data, yielding calibrated estimates. The PSSA cultivation rate is calculated as cropland area divided by total PSSA area, whereas the PSSA cropland proportion quantifies its share of the LP’s total cropland area.
To synchronize with the nine benchmark years of population density analysis, we selected land-use distribution maps from the temporally closest available years (0, 100, 300, 600, 700, 1100, 1200, 1800, and 2000 CE) for spatial visualization. These distribution maps were generated by clipping HYDE global land-use data with the LP boundary. They depict only spatial pattern variations and have not undergone numerical calibration.

2.2.3. Sediment Yield, Sedimentation Rate, Flooding Frequency, and Precipitation

Sediment yield data for the YR were sourced from Ren [27], who estimated historical sediment fluxes by scaling modern measurements (1.6 × 109 t/yr) through paleoenvironmental differential analysis. A 2300-year downstream sedimentation rate sequence was established by integrating 27 datasets from borehole sediment thickness measurements, archeological dating, and radiocarbon (14C) chronologies [28].
Among all flood disasters in the YR basin, the levee breaches and overbank flooding in the lower reaches have attracted the most global attention, becoming a critical issue that successive dynasties prioritized. Historical records of flooding events in the lower YR can be traced back to as early as the Spring and Autumn and Warring States periods (770–221 BCE). Documentation of such events became increasingly frequent after the Western Han Dynasty (206 BCE-9 CE), with even more comprehensive records appearing in historical archives following the Northern Song Dynasty (960–1127 CE). These official documents typically contain structured information, including event dates, breach locations, disaster severity, and relief measures. Flooding event records for the lower YR were compiled from three historical sources (systematically collated from official archives and local chronicles): Annual Chronology of the Yellow River [5], Essential History of Yellow River Water Conservancy [29], and Chronicle of Yellow River Events [30]. Flooding events were categorized using traditional Chinese hydrological terminology: Jue (levee breach), Yi (bankfull overflow), Xi (channel avulsion), and Dashui (catastrophic flooding). In other words, flooding events encompass three equally significant mechanisms: (1) overtopping of dikes, (2) breaching of dikes, and (3) inundation following avulsion due to breaching. Following the statistical conventions of Essential History of Yellow River Water Conservancy (treating multiple flooding events within a single year as one record), this study identified 601 historical flooding events. To analyze long-term trends, this study conducted statistical analyses of flooding frequency using a 20-year moving time window.
Annual precipitation data were reconstructed from pollen assemblages in Gonghai Lake (38°54′ N, 112°14′ E, 1860 m asl), an alpine lake on the LP [31].

2.3. Statistical Methods

This study employed the Pettitt nonparametric test (R package trend, significance level α = 0.05) to detect abrupt change points in population density time series. Wilcoxon rank-sum tests (R wilcox.test, two-tailed, α = 0.05) assessed the significance of differences in five key variables—population density, sediment yield, sedimentation rate, flooding frequency, and precipitation—before versus after the turning year. The results are reported as exact p-values or median change ratios ((post − pre)/pre). To assess inter-group differences in YR flooding frequency, Cliff’s delta effect sizes were computed (R package effsize), with magnitude thresholds defined as: |Δ| ≥ 0.474 (large effect), 0.33 ≤ |Δ| < 0.474 (medium effect), and 0.147 ≤ |Δ| < 0.33 (small effect) [32]. All analyses were conducted using R 4.2.0.

3. Results

3.1. Millennial-Scale Abrupt Shifts in Population Density and Cultivation Rate

Historical records document a significant demographic shift in the PSSA following 934 CE (Figure 2a). Statistical analyses showed that although the Pettitt test (change-point detection) suggested 934 CE as a transitional year (p = 0.078, nonsignificant), the Wilcoxon rank-sum test (group comparison) demonstrated a significant 7.4-fold rise in median population density after this date (p = 0.005), robustly confirming 934 CE as a pivotal turning year. From 934 to 1102 CE, population density rose sharply from 1.3 to 19.8 persons/km2 (14.2× higher), demonstrating clear nonlinear growth dynamics. Over the 2400-year study span, the post-934 CE mean population density (51 persons/km2) exceeded pre-934 CE levels (5.2 persons/km2) by a factor of 9.8. Spatially, the share of the LP’s total population residing within the PSSA rose sharply from 17.2% to 33.8% (Figure 2a).
Over the past 2400 years, this region comprised merely 24% of the LP’s mean population but contributed 47% of its mean cropland area. This land-use disparity intensified post-1000 CE, as the cropland area proportion consistently surpassed 50% (Figure 2b). The cultivation rate rose sharply from 8.6% to 20.5% (1.4× higher), lagging slightly behind the abrupt population density increase—a historic turning point in land-use dynamics (Figure 2b). Figure 3 and Figure 4 demonstrate that this abrupt transition coincided with the northwestward expansion of human activities on the LP, shifting from basins/river valleys to erosion-sensitive hilly–gully areas—a spatial pattern paralleling concurrent population and cropland increases.

3.2. Chain Response of Sediment Yield and Sedimentation Rate

The abrupt surge in population density within the PSSA triggered marked transformations in the YR’s sediment production system and downstream depositional processes. As evidenced by these data (Figure 5), the YR’s sediment yield post-934 CE was significantly higher than in preceding periods (Wilcoxon rank-sum test, p = 0.035), with the mean value surging from 1.6 × 108 t/yr to 1.02 × 109 t/yr—a 5.4-fold increase (Cliff’s δ = 0.80). Concurrently, downstream sedimentation rates exhibited an even more pronounced synchronous shift (Wilcoxon rank-sum test, p = 0.001), escalating from 0.3 cm/yr pre-934 CE to 3.3 cm/yr thereafter—a 10-fold acceleration (Cliff’s δ = 1.00). Notably, the increase in sedimentation rate (10-fold) was significantly greater than the rise in sediment yield (5.4-fold), likely reflecting an amplification effect of downstream channels on sediment delivery. These findings demonstrate that anthropogenic forcing systematically regulated both sediment generation throughout the watershed and depositional dynamics downstream.

3.3. Threshold Effects of Flooding Frequency

Based on historical population data from the PSSA (increasing from 1.3 to 19.8 persons/km2 during 934–1102 CE), this study examined 934 CE as a hypothesized turning year. The Wilcoxon rank-sum test revealed a significant difference in YR flooding frequency before and after 934 CE (p < 0.0001), with an extremely large effect size (Cliff’s δ = 0.98). Specifically, the mean flooding frequency increased from 1.6 ± 0.9 events per 20 years (flood-free in 96% of intervals) pre-934 CE to 10.4 ± 4.2 events per 20 years (flood-free intervals reduced to 56%) thereafter (Figure 6). This hydrological shift showed strong spatiotemporal coupling with intensified human activities, featuring a 9.8-fold population density increase (5.1 → 51 persons/km2), a 1.4-fold reclamation rate rise (8.6% → 20.5%), 5.4-fold sediment yield growth (1.6 → 10.2 × 108 t/yr), and 10-fold downstream sedimentation acceleration (0.3 → 3.3 cm/yr). In the semi-arid LP, precipitation constitutes the dominant climatic control on soil erosion. Wilcoxon tests indicated significantly higher annual precipitation pre-934 CE (448 mm) versus post-934 CE (432 mm) (p < 0.001; Figure 6), yet this minor 16 mm difference starkly contrasted with abrupt anthropogenic changes in population density, cultivation rate, and sediment yield. These results demonstrate that hydrological transformations were dominantly human-induced, with precipitation variability exerting merely a secondary modulating effect. In addition, correlation analysis with a 20-year time interval revealed a weak negative trend (r = −0.295) between downstream flooding frequency and Loess Plateau rainfall over the past 2400 years, though this did not reach statistical significance (p = 0.15).

4. Discussions

4.1. How Population Growth and Cropland Expansion Triggered YR Flooding

Soil erosion refers to the detachment, transport, and deposition of soil and parent material driven by hydraulic, eolian, gravitational, and cryogenic forces [33]. As the primary sediment source for the YR basin, soil erosion on the LP is dominated by water erosion, with five key influencing factors: rainfall, topography, vegetation, soil properties, and soil conservation measures [12]. Among these controls, rainfall operates as an exclusively natural driver, whereas soil conservation measures—minimal before the modern era—constitute anthropogenic interventions. While topography, loess characteristics, and vegetation are inherently natural attributes, their baseline states have been significantly altered through anthropogenic modifications [34]. Studies demonstrate that vegetation cover is the dominant control on soil erosion rates across the LP [27]. Even in pre-anthropogenic eras, climatic oscillations (warm–cold/dry–wet cycles) regulated soil erosion intensity through vegetation community composition and canopy cover dynamics [35]. The historical shift from hunter–gatherer to pastoral–agricultural societies intrinsically required large-scale removal of native vegetation [36,37]. This destruction significantly amplifies the impact of natural factors such as rainfall and slope on soil erosion [35]. Controlled experiments conducted in the Ziwuling Forest Region of the Loess Plateau (LP) confirm that newly cultivated croplands generate significantly higher soil erosion modulus compared with both forested areas (Table 1) and grasslands (Table 2) under equivalent rainfall intensity and slope gradient conditions. This disparity primarily arises from the vegetation’s multi-layer protective mechanisms: the canopy layer intercepts rainfall, the litter layer regulates surface runoff, and the root system enhances soil structural stability and scouring resistance [38,39]. By modifying vegetation cover, anthropogenic activities have emerged as the primary regulator of soil erosion intensity throughout the LP.
In preindustrial societies, land-use/cover change (LUCC) was primarily driven by subsistence strategies (e.g., agro-pastoral systems) and demographic pressure [41]. Although both agricultural and pastoral practices alter natural vegetation, agriculture requires the complete removal of the original plant cover [19]. Ultimately, population size serves as the root determinant of anthropogenic ecosystem modifications [42]. Here, we demonstrate that the PSSA—despite covering merely 24% of the total study area—concentrates 47% of the region’s cropland. Dual-temporal analysis reveals: (1) an abrupt 14.2-fold surge in population density (1.3 → 19.8 persons/km2) during 934–1102 CE, and (2) a 9.8-fold increase in mean density (5.2 → 51 persons/km2) over the 2400-year period partitioned by the 934 CE threshold (Figure 2a). This demographic explosion drove cultivation rates above 20% (Figure 2b). These results indicate that post-934 CE, the PSSA experienced dual pressures—a dramatic increase in both population density and cultivation rate—providing crucial anthropogenic context for the subsequent changes in sediment load and flooding frequency in the YR system. To meet the food demands of the surging population, extensive deforestation and grassland conversion to cropland occurred, particularly in the PSSA dominated by hilly–gully topography (Figure 4). This land-use transformation increased sediment yield by 5.4-fold (Figure 5) and elevated downstream sedimentation rates by 10-fold (from 0.3 to 3.3 cm/yr). The massive increase in sediment load flowing through the flat North China Plain leads to rapid deposition, which accelerates riverbed aggradation and reduces channel flood capacity. Under these conditions, even small flood events may exceed the channel’s discharge capacity, overtopping or breaching levees and inundating surrounding farmland and villages. Ultimately, flooding frequency in the lower YR surged from 1.6 to 10.4 events per 20 years—a 5.5-fold increase (Figure 6). Critically, the short-term population explosion (14.2-fold increase) vastly outpaced the long-term mean trend (9.8-fold increase), confirming nonlinear acceleration of anthropogenic pressures post-934 CE. This cascading disaster chain (population growth → cropland expansion → vegetation loss → intensified erosion → channel sedimentation → flooding surge) definitively establishes post-934 CE population density as the paramount driver of both PSSA erosion and YR downstream flood regimes.
Integrated Wilcoxon rank-sum test results (population density: p = 0.005; sediment yield: p = 0.035; flooding frequency: p < 0.0001) demonstrate that when population density exceeds the critical threshold of 19.8 persons/km2, the impact of human activities on erosion–flooding processes becomes significantly stronger than the direct effects of precipitation changes—despite only a 4% decrease in mean annual rainfall, its variability increased markedly (p < 0.001). Furthermore, correlation analysis revealed no statistically significant relationship between downstream flooding frequency and LP rainfall (r = −0.295, p > 0.05), suggesting that rainfall has a relatively weak influence on flooding frequency. These findings not only provide quantitative evidence that the abrupt increase in YR flooding frequency was predominantly driven by human activities (versus climate variability) but also highlight a critical population density threshold with significant implications for modern basin management—with current PSSA density at 120% of this threshold, yet the Grain-for-Green Program has not fully incorporated this historical pattern. Our study makes two principal contributions compared with the vegetation gradual degradation hypothesis [19] and climate abrupt change hypothesis [23]: (1) By comparing climatic and anthropogenic influences, we have identified population surges as the primary driver of the substantial increase in downstream flooding frequency; (2) We have established a population density threshold of 19.8 persons/km2 in the LP’s PSSA as an ecological pressure indicator for the YR basin. This quantitative approach holds greater practical and scientific significance for YR management compared with qualitative studies in the existing literature.
The findings of this study are based on historical flooding records rather than modern quantitative discharge data. The former constitutes discrete data (with only two states: flooding or no flooding), while the latter represents continuous data. The rationale for using flooding frequency to characterize disasters in the lower YR lies in the fact that breaching and avulsion events are not solely the result of upstream floods, but rather a combined outcome of sediment deposition and whether upstream floods exceed the channel’s flood conveyance capacity. Flooding events do not occur only during major floods; when sediment deposition rates are excessively high, raising the riverbed and reducing the channel’s flood conveyance capacity, even minor floods can trigger flooding.
The identification of 934 CE as a turning point for YR flooding is based precisely on this type of discrete flooding event data. While the significant difference in flooding frequency before and after 934 CE may have been influenced by factors such as artificial breaches and warfare, it is crucial to emphasize that the periods before and after 934 CE both span a millennium-scale timeframe. The impact of warfare and artificial breaches is typically confined to decadal or multi-decadal scales and is unlikely to substantially affect millennial-scale differences. A key piece of evidence is that both before and after 934 CE—during the relatively stable periods of the Han, Tang, Song, Yuan, Ming, and Qing dynasties—the post-934 CE periods (Song (960–1271 CE), Yuan (1271–1368 CE), Ming (1368–1644 CE), Qing (1644–1911 CE)) exhibited significantly higher flooding frequencies than the pre-934 CE periods (Han (202 BCE-220 CE), Tang (618–907 CE)).

4.2. From Historical Lessons to Modern River Management Strategies

Historical studies demonstrate that the PSSA of the LP has exerted a progressively dominant influence on YR downstream flooding frequency since 934 CE. An in-depth analysis reveals that the 19.8 persons/km2 historical threshold represents the carrying capacity ceiling of natural vegetation systems under preindustrial agro-pastoral regimes. As a semi-arid climate zone reliant on rainfed agriculture, this region was forced to expand into ecologically vulnerable northwestern areas under population pressure, creating a feedback loop of ‘population growth → reclamation and expansion → intensified erosion’. However, modern governance practices reveal a critical paradigm shift: even with current population density (120 persons/km2) far exceeding historical thresholds, the establishment of artificial ecological replacement systems (e.g., the Grain-for-Green Program launched in 1999) has successfully balanced ecological integrity and socioeconomic development.
This transformation is anchored in two critical substitutions: (1) natural vegetation coverage (the historical determinant) has been replaced by artificial soil–water conservation engineering coverage (the modern driver), and (2) traditional rainfed agriculture (a historical erosion source) has been replaced with terraced/irrigated farming systems (modern erosion control measures). Studies confirm that terracing reduces sediment yield per unit area by 80% [43], whereas check-dam systems intercept >60% of slope-derived sediments [44]. Post-1999, vegetation restoration has contributed to 57% of total sediment reduction [45]. The synergistic effects of these measures have reduced the YR’s annual sediment load from its historical peak to 2 × 108 t (an 87.5% decrease) [16], with flooding events effectively eliminated (Figure 6).
The historical population density threshold of 19.8 persons/km2 revealed by this study holds critical contemporary significance, as it establishes a fundamental quantitative relationship between population pressure and ecological response. In modern governance contexts, the historical threshold should serve as an ‘ecological pressure conversion standard’: (1) convert actual population density to effective ecological pressure through soil and water conservation engineering (e.g., 60% terracing coverage can reduce 120 persons/km2 to an equivalent 48 persons/km2); (2) when a region’s converted value exceeds 19.8 persons/km2, it indicates insufficient engineering regulation and requires prioritized management; (3) establish closed-loop ‘threshold-triggered engineering response’ mechanisms in key areas such as the PSSA region. This approach retains the scientific validity of historical thresholds while addressing their application challenges under modern high-population-density conditions through quantifiable engineering regulatory capacity.

4.3. Limitations and Prospects

This study quantifies the drivers of historical YR flooding surges, but two methodological limitations remain. While the historical population density reconstruction covers nine temporal nodes, administrative-boundary area-weighting potentially masks PSSA’s inherent spatial variability. This method may inadequately capture population distribution differences between hills and valleys. The global land-use data (HYDE), though calibrated with Chinese records, lacks the resolution to precisely locate key erosion sources such as sloping croplands. Future research should focus on: (1) developing population-terrain models, (2) combining historical cropland data with remote sensing, and (3) reconstructing higher-resolution cropland spatial distribution data through historical records mining. Regarding the potential uncertainties in the YR flooding chronology—such as missing records during war periods—we acknowledge these limitations. However, flooding events inherently contribute to downstream sediment deposition. Therefore, future research should incorporate independent validation through geological dating records to enhance the reliability of the chronology. These improvements will enhance understanding of human–environment interactions and support YR basin governance. Moreover, this study only includes population density spatial data from nine benchmark years, which may introduce some uncertainty in results, such as the 19.8 persons/km2 ecological pressure threshold in the LP’s PSSA. In the future, historical records should be compiled to obtain population spatial density data from additional benchmark years to validate the accuracy of our findings.

5. Conclusions

This study confirms a 14.2-fold population density surge (1.3 → 19.8 persons/km2, p = 0.005) in the LP’s PSSA between 934 and 1102 CE (168 years). Comparative analysis spanning 2400 years (934 CE as threshold) shows a 9.8-fold rise in regional mean population density (5.2 → 51 persons/km2), driving cropland reclamation to 20.5%. This caused a 5.4× increase in sediment yield (1.6 × 108 → 1.02 × 109 t/yr, p = 0.035), leading to 10× faster downstream sedimentation (0.3 → 3.3 cm/yr, p = 0.001), which ultimately raised YR flooding frequency from 1.6 to 10.4 events per 20 years (p < 0.0001). This finding provides quantitative evidence of historical flooding regime shifts and establishes an ecological pressure conversion benchmark for modern YR management. Specifically, it proposes a closed-loop governance mechanism where: (1) soil–water conservation engineering transforms actual population density into effective ecological pressure; and (2) prioritized intervention activates in the PSSA when converted values exceed the 19.8 persons/km2 threshold.

Author Contributions

Conceptualization, T.H.; Methodology, Y.L.; Formal analysis, T.H.; Writing—original draft, T.H.; Writing—review & editing, T.H. and Y.L.; Supervision, T.H. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Chen, Y.Z.; Syvitski, J.P.M.; Gao, S.; Overeem, I.; Kettner, A.J. Socio-economic impacts on flooding: A 4000–year history of the Yellow River, China. AMBIO 2012, 41, 682–698. [Google Scholar] [CrossRef] [PubMed]
  2. Milliman, J.D.; Qin, Y.S.; Ren, M.; Saito, Y. Man’s influence on the erosion and transport of sediment by Asian rivers: The Yellow River (Huanghe) example. J. Geol. 1987, 95, 751–762. [Google Scholar] [CrossRef]
  3. Yan, T. Enlightenment from the decline of the lower Yellow River channel. Hydrogeol. Eng. Geol. 1988, 05, 27–31. [Google Scholar]
  4. Shu, L.; Finlayson, B. Flood management on the lower Yellow River: Hydrological and geomorphological perspectives. Sediment. Geol. 1993, 85, 285–296. [Google Scholar] [CrossRef]
  5. Shen, Y.; Zhao, S.; Zheng, D. Chronology of the Yellow River; Information Committee of the Military Commission: Nanjing, China, 1935. (In Chinese) [Google Scholar]
  6. Chen, Y.; Wang, K.; Lin, Y.; Shi, W.; Song, Y.; He, X. Balancing green and grain trade. Nat. Geosci. 2015, 8, 739–741. [Google Scholar] [CrossRef]
  7. Qiu, G. An investigation into piled–up cities under the Kaifeng City. China’s Cult. Relics 2004, 26, 70–77. (In Chinese) [Google Scholar]
  8. Ongley, D.E. The Yellow River: Managing the unmanageable. Water Int. 2000, 25, 227–231. [Google Scholar] [CrossRef]
  9. Xu, J. The water fluxes of the Yellow River to the sea in the past 50 years, in response to climate change and human activities. Environ. Manag. 2005, 35, 620–631. [Google Scholar] [CrossRef]
  10. Kaya, Y.; Sanli, F.B.; Abdikan, S. Refinement of ICESat-2 derived inland water surface levels with the TG20 local geoid model: In the case of Türkiye lakes. Phys. Chem. Earth 2005, 11, 139–155. [Google Scholar] [CrossRef]
  11. Zhang, H.Y. Discussion on Historical Strategies for River Management; Water Resources Press: Beijing, China, 1982. [Google Scholar]
  12. Zhao, G.J.; Mu, X.M.; Wen, Z.; Wang, F.; Gao, P. Soil Erosion, Conservation, and Eco-Environment Changes in the Loess Plateau of China. Land Degrad. Dev. 2013, 24, 499–510. [Google Scholar] [CrossRef]
  13. Wu, X.; Wang, H.; Bi, N.; Saito, Y.; Xu, J.; Zhang, Y.; Lu, T.; Cong, S.; Yang, Z. Climate and human battle for dominance over the Yellow River’s sediment discharge: From the Mid–Holocene to the Anthropocene. Mar. Geol. 2020, 42, 106188. [Google Scholar] [CrossRef]
  14. Jiao, J.Y.; Wang, Z.J.; Zhao, G.J.; Wang, W.Z.; Mu, X.M. Changes in sediment discharge in a sediment-rich re gion of the Yellow River from 1955 to 2010: Implications for further soil erosion control. J. Arid. Land 2014, 6, 540–549. [Google Scholar] [CrossRef]
  15. Zhang, C.H. The Yellow River: History, Present, Future; Science Press: Beijing, China, 2023. (In Chinese) [Google Scholar]
  16. Liu, X. Causes of the Sharp Reduction in Water and Sediment in the Yellow River in Recent Years; Science Press: Beijing, China, 2016. [Google Scholar]
  17. Yellow River Conservancy Commission, Ministry of Water Resources. Study on Water-Sediment Variations in the Yellow River; Yellow River Water Resources Press: Zhengzhou, China, 2014. (In Chinese)
  18. Yellow River Conservancy Commission, Ministry of Water Resources. Comprehensive Planning of the Yellow River Basin (2012–2030); Yellow River Water Resources Press: Zhengzhou, China, 2013. (In Chinese)
  19. Tan, Q. Why did the Yellow River keep a calm period for a long time after the Eastern Han Dynasty–to verify the view that the rational use of land in the middle reaches of the Yellow River is the decisive factor to eliminate water disasters in the lower reaches from the perspective of history. Acad. Mon. 1962, 2, 23–35. (In Chinese) [Google Scholar] [CrossRef]
  20. Shi, N. A Study of the Historical Geography of the Loess Plateau; The Yellow River Water Conservancy Press: Zhengzhou, China, 2001. (In Chinese) [Google Scholar]
  21. Xu, J.; Li, F. Response of lower Yellow River bank breachings to La Niña events since 924 CE. Catena 2019, 176, 159–169. [Google Scholar] [CrossRef]
  22. Wang, Y.; Su, Y. Influence of solar activity on breaching, overflowing and course-shifting events of the Lower Yellow River in the late Holocene. Holocene 2013, 23, 656–666. [Google Scholar] [CrossRef]
  23. Li, T.; Li, J.; Zhang, D.D. Yellow River flooding during the past two millennia from historical documents. Progress in Physical Geography: Earth Environ. 2020, 44, 661–678. [Google Scholar] [CrossRef]
  24. Tan, Q. The Historical Atlas of China (8 Volumes); SinoMaps Press: Beijing, China, 1996. [Google Scholar]
  25. Lu, Y.; Teng, Z. A Study of Historical Population by Province in China (2 Volumes); Shandong People’s Publishing House: Jinan, China, 2006. [Google Scholar]
  26. Fang, X.; Ye, Y.; Zhang, C.; Tang, C. Cropland cover change and its environmental impacts in the history of China. J. Palaeogeogr. 2019, 21, 160–174. [Google Scholar] [CrossRef]
  27. Ren, M. Sediment discharge of the Yellow River, China: Past, present and future—A synthesis. Acta Oceanol. Sin. 2015, 34, 1–8. [Google Scholar] [CrossRef]
  28. Xu, J. Sedimentation rates in the lower Yellow River over the past 2300 years as influenced by human activities and climate change. Hydrol. Process. 2003, 17, 3359–3371. [Google Scholar] [CrossRef]
  29. Editorial Committee of Summary of Yellow River Water Conservancy History. Summary of Yellow River Water Conservancy History; Yellow River Conservancy Commission: Zhengzhou, China, 2003. [Google Scholar]
  30. Yellow River Conservancy Commission. Chronicle of Major Events on the Yellow River; Yellow River Water Resources Press: Zhengzhou, China, 2001. [Google Scholar]
  31. Chen, F.; Xu, Q.; Chen, J.; Birks, H.J.B.; Liu, J.; Zhang, S.; Jin, L.; An, C.; Telford, R.J.; Cao, X.; et al. East Asian summer monsoon precipitation variability since the last deglaciation. Sci. Rep. 2015, 5, 11186. [Google Scholar] [CrossRef]
  32. Du, X.; Zhang, Y.; Jiang, H.; Dong, X. Influence of Canopy Environmental Characteristics on Regeneration of Nine Tree Species in Broadleaved Korean Pine Forests. Forests 2025, 16, 757. [Google Scholar] [CrossRef]
  33. Toy, T.J.; Foster, G.R.; Renard, K.G. Soil Erosion: Process, Prediction, Measurement, and Control; John Wiley & Sons, Inc.: New York, NY, USA, 2002. [Google Scholar]
  34. Ren, M.; Zhu, X.M. Anthropogenic influences on changes in the sediment load of the Yellow River, China, during the Holocene. Holocene 1994, 4, 314–320. [Google Scholar] [CrossRef]
  35. Tang, K.L.; Zhang, K.L.; Zheng, F.L.; Zha, X.; Wang, W.L.; Cai, Q.; Bai, H.Y. Analysis of Natural Erosion and Human–Accelerated Erosion in the Ziwuling Forest Region. Mem. Northwest Inst. Soil Water Conserv. Chin. Acad. Sci. Minist. Water Resour. 1993, 17, 17–28. (In Chinese) [Google Scholar]
  36. Chen, Y.Z.; Jing, K.; Cai, Q.G. Modern Erosion and Its Control on the Loess Plateau; Science Press: Beijing, China, 1998. [Google Scholar]
  37. Mu, X.M.; Zhao, G.J.; Gao, P.; Sun, W.Y. New Patterns of Water and Sediment Changes in the Loess Plateau; Science Press: Beijing, China, 2019. [Google Scholar]
  38. Zhu, X.M. The Influence of Vegetation Factors on Soil and Water Loss in the Loess Region. Acta Pedol. Sin. 1960, 8, 110–121. (In Chinese) [Google Scholar]
  39. Zhang, X.C.; Shao, M.A.; Huang, Z.B.; Lu, Z.F. Effects of Different Vegetation Types on Soil Erosion and Nitrogen Loss. Acta Ecol. Sin. 2002, 20, 1038–1044. [Google Scholar]
  40. Bai, H.Y.; Tang, K.L.; Zhang, K.L.; Chen, W.L.; Zha, X. Experimental Study on Human–Accelerated Erosion from Grassland Reclamation Using Artificial Rainfall Simulation. Mem. Northwest Inst. Soil Water Conserv. Chin. Acad. Sci. Minist. Water Resour. 1993, 17, 87–93. [Google Scholar]
  41. Li, H. Prehistoric Vegetation Conditions and Timber Resources in the Longxi Loess Plateau. Ph.D. Dissertation, Lanzhou University, Lanzhou, China, 2016. [Google Scholar]
  42. Yang, P.L. Studies on Historical Population Geography in the Middle Reaches of the Yellow River. In Research Collection on Environmental Evolution and Water–Sediment Dynamics in the Yellow River Basin; Wang, S.C., Ed.; China Ocean Press: Beijing, China, 1993; Volume 5. [Google Scholar]
  43. Liu, S.L.; Dong, Y.H.; Li, D.; Liu, Q.; Wang, J.; Zhang, X.L. Effects of different terrace protection measures in a sloping land consolidation project targeting soil erosion at the slope scale. Ecol. Eng. 2013, 53, 46–53. [Google Scholar] [CrossRef]
  44. Tian, P.; Zhao, G.J.; Mu, X.M.; Wang, F.; Gao, P.; Mi, Z.J. Check dam identification using multisource data and their effects on streamflow and sediment load in a Chinese Loess Plateau catchment. J. Appl. Remote Sens. 2013, 7, 63–72. [Google Scholar] [CrossRef]
  45. Wang, S.; Fu, B.J.; Piao, S.L.; Lv, Y.H.; Ciais, P.; Feng, X.M.; Wang, Y.F. Reduced sediment transport in the Yellow River due to anthropogenic changes. Nat. Geosci. 2015, 9, 38–41. [Google Scholar] [CrossRef]
Hydrology 12 00210 g001
Figure 1. Location map of the Yellow River (YR) basin.
Figure 1. Location map of the Yellow River (YR) basin.
Hydrology 12 00210 g001
Hydrology 12 00210 g002
Figure 2. Changes in human activity intensity in the PSSA of the LP. (a) Population density and its proportion to the LP total population. (b) Cultivation rate and its proportion to the LP total cultivated area. The dashed line in the figure denotes the turning year (934 CE) (the same applies below).
Figure 2. Changes in human activity intensity in the PSSA of the LP. (a) Population density and its proportion to the LP total population. (b) Cultivation rate and its proportion to the LP total cultivated area. The dashed line in the figure denotes the turning year (934 CE) (the same applies below).
Hydrology 12 00210 g002
Hydrology 12 00210 g003
Figure 3. Spatiotemporal variations in population density across the LP.
Figure 3. Spatiotemporal variations in population density across the LP.
Hydrology 12 00210 g003
Hydrology 12 00210 g004
Figure 4. Spatiotemporal changes in land-use types across the LP.
Figure 4. Spatiotemporal changes in land-use types across the LP.
Hydrology 12 00210 g004
Hydrology 12 00210 g005
Figure 5. Temporal changes in sediment yield from the YR and sedimentation rate in its lower reaches [27,28].
Figure 5. Temporal changes in sediment yield from the YR and sedimentation rate in its lower reaches [27,28].
Hydrology 12 00210 g005
Hydrology 12 00210 g006
Figure 6. Precipitation variability in the LP and flooding frequency in the lower YR.
Figure 6. Precipitation variability in the LP and flooding frequency in the lower YR.
Hydrology 12 00210 g006
Table 1. Comparative experiment on soil erosion under simulated rainfall: forestland vs. deforested land [35].
Table 1. Comparative experiment on soil erosion under simulated rainfall: forestland vs. deforested land [35].
Slope Gradient (°)Experimental TreatmentsRainfall Intensity (mm/min)Soil Erosion Modulus (t/km2/a)
20forestland1.260
1.910
2.376.67
deforested land1.26406.7
1.912244
2.374994.7
30forestland1.2638.67
1.919.33
2.3749.33
deforested land1.262000
1.914826.7
2.3711450
Table 2. Comparative experiment on soil erosion under simulated rainfall: grassland vs. reclaimed grassland [40].
Table 2. Comparative experiment on soil erosion under simulated rainfall: grassland vs. reclaimed grassland [40].
Slope (°)Experimental TreatmentsRainfall Intensity (mm/min)Soil Erosion Modulus (t/km2/a)
18–20grassland1.560
reclaimed grassland1.56469
23–25grassland1.560
reclaimed grassland1.561554
32–35grassland1.560
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Huang, T.; Li, Y. Human-Induced Shifts in Yellow River Flooding: Population Threshold Effects in the Loess Plateau’s Primary Sediment Source Area (934 CE). Hydrology 2025, 12, 210. https://doi.org/10.3390/hydrology12080210

AMA Style

Huang T, Li Y. Human-Induced Shifts in Yellow River Flooding: Population Threshold Effects in the Loess Plateau’s Primary Sediment Source Area (934 CE). Hydrology. 2025; 12(8):210. https://doi.org/10.3390/hydrology12080210

Chicago/Turabian Style

Huang, Tao, and Yabin Li. 2025. "Human-Induced Shifts in Yellow River Flooding: Population Threshold Effects in the Loess Plateau’s Primary Sediment Source Area (934 CE)" Hydrology 12, no. 8: 210. https://doi.org/10.3390/hydrology12080210

APA Style

Huang, T., & Li, Y. (2025). Human-Induced Shifts in Yellow River Flooding: Population Threshold Effects in the Loess Plateau’s Primary Sediment Source Area (934 CE). Hydrology, 12(8), 210. https://doi.org/10.3390/hydrology12080210

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Article metric data becomes available approximately 24 hours after publication online.
Back to TopTop