Nothing Special   »   [go: up one dir, main page]
Next Article in Journal
Monitoring the Impact of Floods on Water Quality Using Optical Remote Sensing Imagery: The Case of Lake Karla (Greece)
Previous Article in Journal
Effects of Low-Molecular-Weight Organic Acids on the Transport of Polystyrene Nanoplastics in Saturated Goethite-Coated Sand Columns
Previous Article in Special Issue
Identifying the Layout of Retrofitted Rainwater Harvesting Systems with Passive Release for the Dual Purposes of Water Supply and Stormwater Management in Northern Taiwan
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 16
Issue 23
10.3390/w16233501
water-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 thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Multi-Scale Spatial Relationship Between Runoff and Landscape Pattern in the Poyang Lake Basin of China

by
Panfeng Dou
1,*,†,
Yunfeng Tian
2,†,
Jinfeng Zhang
1 and
Yi Fan
3
1
School of Civil Engineering, Putian University, Putian 351100, China
2
Key Laboratory of the Ministry of Education for Coastal and Wetland Ecosystems, College of the Environment and Ecology, Xiamen University, Xiamen 361102, China
3
College of Environmental and Biological Engineering, Putian University, Putian 351100, China
*
Author to whom correspondence should be addressed.
†
These authors contributed equally to this work.
Water 2024, 16(23), 3501; https://doi.org/10.3390/w16233501
Submission received: 30 October 2024 / Revised: 29 November 2024 / Accepted: 2 December 2024 / Published: 5 December 2024
(This article belongs to the Special Issue Watershed Hydrology and Management under Changing Climate)
Browse Figures
Figure 1
<p>Location of the Poyang Lake Basin.</p> "> Figure 2
<p>Elevation and sub-basin division in the Poyang Lake Basin. Note: The numbers in the right part of this figure represent the sub-basin numbers.</p> "> Figure 3
<p>A comparative analysis of the flow dynamics observed and simulated in the Poyang Lake Basin. Note: 201101 represents January 2011.</p> "> Figure 4
<p>Spatial–temporal pattern of runoff in the Poyang Lake Basin from 2011 to 2020.</p> "> Figure 5
<p>Spatial distribution of landscape types and proportions in the Poyang Lake Basin.</p> "> Figure 6
<p>Spatial distribution of geographically weighted regression coefficients between runoff and landscape type in the Poyang Lake Basin.</p> "> Figure 7
<p>Spatial distribution of key landscape metrics at the landscape and class scales in the Poyang Lake Basin.</p> "> Figure 8
<p>Spatial distribution of geographically weighted regression coefficients between runoff and landscape metrics at landscape and class scales in the Poyang Lake Basin. Note: R<sup>2</sup> = N/A indicates that the GWR model was not successfully constructed.</p> "> Figure 9
<p>Ecological restoration pathways in sub-basins of the Poyang Lake Basin.</p> ">
Versions Notes

Abstract

:
Runoff research serves as the foundation for watershed management, and the relationship between runoff and landscape pattern represents a crucial basis for decision-making in the context of watershed ecological protection and restoration. However, there is a paucity of research investigating the multi-scale spatial relationship between runoff and landscape patterns. This study employs the Poyang Lake Basin (PLB) as a case study for illustrative purposes. The construction of the soil and water assessment tool (SWAT) model is the initial step in the process of carrying out runoff simulation, which in turn allows for the analysis of the spatial–temporal characteristics of runoff. Subsequently, Pearson’s correlation analysis, global linear regression and geographically weighted regression (GWR) models are employed to examine the impact of landscape composition on runoff. Finally, the spatial relationship between runoff and landscape pattern is investigated at the landscape and class scales. The results of the study demonstrate the following: (1) runoff in the PLB exhibited considerable spatial–temporal heterogeneity from 2011 to 2020. (2) Forest was the most prevalent landscape type within the PLB. Landscape composition’s impact on runoff exhibited non-linear characteristics, with forest, cropland, barren, and grassland influencing runoff in decreasing order. (3) A spatial relationship between runoff and landscape pattern was observed. At the landscape scale, patch diversity significantly influenced runoff, and reducing patch diversity primarily increased runoff. At the class scale, forest and cropland patch areas had the greatest impact on runoff, potentially enhanced by improving patch edge density. (4) Nine sub-basins needing ecological restoration were identified, with restoration pathways developed based on spatial relationships between runoff and landscape patterns. This study elucidates the impact of landscape composition and pattern on runoff, thereby providing a basis for informed decision-making and technical support for the ecological restoration and management of the watershed.

1. Introduction

Runoff represents the primary conduit through which humans access water resources and is therefore of paramount importance for the sustenance of life and health, socio-economic development, and the protection of ecosystems [1,2]. The accurate simulation of runoff is fundamental to the development of a comprehensive understanding of the ecohydrological processes and mechanisms that occur within a watershed [3,4]. This understanding enables the formulation of more accurate watershed management strategies, the improvement of the efficiency of water resources utilization, and the subsequent assurance of the safety of water resources within the watershed. The most commonly used aggregate hydrological models include the Stanford, Sacramento, Xinanjiang and SWMM models, among others. These models consider the watershed as a whole for simulation, and their structure is relatively simple. However, they are unable to take into account the spatial distribution inhomogeneity of climate, topography, vegetation and other elements, which may result in inaccurate simulation results [5,6]. Distributed hydrological models that are commonly employed include the soil and water assessment tool (SWAT) model, the VIC model, the MIKE SHE model, and the HBV model, among others. These types of model consider the spatial heterogeneity of the geographic elements within the watershed and divides it into multiple computational units, each of which is characterized by its own unique subsurface conditions and rainfall conditions. This allows for more accurate simulation of the hydrological processes within the watershed. Furthermore, the physical mechanism is clear, which enables a deeper understanding of the hydrological phenomena [7,8]. The SWAT model, which is representative of distributed hydrological models, has been widely applied to runoff simulation due to the maturity of its models and the high accuracy of its simulations [9].
Landscape pattern is the spatial arrangement of different sizes, shapes and types of patches resulting from the coupling of natural factors and human activities at different scales [10,11]. Landscape pattern is an important factor influencing runoff, which is mainly reflected in the fact that the area, shape, distribution and diversity of patches affect the water cycle process, including evapotranspiration, interception and infiltration, which in turn affect the formation and transformation of runoff [12,13]. Along with rapid socio-economic development, large-scale urban and rural construction, conversion of cropland to forests and grasslands, and other human activities have changed the landscape composition and pattern, interfered with the eco-hydrological processes of watersheds, and significantly affected the formation, transformation, and spatial–temporal distribution of runoff [14,15,16], resulting in the frequent occurrence of droughts, floods, soil erosion, and other catastrophic events. In this context, there is an urgent need for ecohydrological watershed studies to analyze the effects of landscape composition and pattern on runoff. Existing studies mainly focus on the spatial and temporal variations of runoff and the analysis of its influencing factors [17,18], which provide important references for watershed ecological management, but the study of the multi-scale spatial relationship between runoff and landscape pattern has rarely been reported.
The abundant water resources of the Poyang Lake Basin (PLB) provide a guarantee for the sustainable development of the socio-economic–natural ecosystem. However, in recent years, it has been facing serious challenges of drought and ecological degradation [19,20,21,22]. The evidence suggests that climate change has a more pronounced influence on runoff than land use in the PLB [23]. However, given the relatively low cost of modifying land use, investigating the relationship between landscape pattern and runoff is of great practical significance. This can aid decision-makers in optimizing the landscape pattern of the watershed to efficiently enhance runoff. Based on the above analysis, (1) the SWAT model was constructed for the purpose of simulating eco-hydrological processes in the PLB, and the spatial–temporal characteristics of runoff were analyzed from 2011 to 2020; (2) the impact of landscape composition on runoff was analyzed using linear and spatial regression models; and (3) landscape metrics were calculated using Fragstats 4.2 software, and the spatial relationship between runoff and landscape pattern was explored at both landscape and type scales. The objective of this study is to elucidate the impacts of landscape composition and pattern on runoff, thereby furnishing a scientific foundation and technical assistance for the ecological protection and restoration of the watershed.

2. Materials and Methods

2.1. Study Area

The PLB (113°35′~118°29′ E, 24°29′~30°05′ N) is situated on the southern bank of the middle and lower reaches of the Yangtze River (Figure 1). The region encompasses five major rivers: the Gan River, the Fu River, the Xin River, the Rao River, and the Xiu River. The total area of the basin is approximately 162,200 square kilometers. Poyang Lake, as the largest freshwater lake in China and an important storage component of the Yangtze River system, occupies an area of 3150 square kilometers at the normal water level (ranging from 14 to 15 m), expands beyond 4125 square kilometers during high water levels (reaching 20 m), and shrinks to merely 500 square kilometers at low water levels (dropping to 12 m). The study area is situated within a subtropical humid monsoon climate, characterized by abundant heat, precipitation, a long frost-free period and four distinct seasons. The average annual temperature is 17.1 °C, the average annual sunshine hours are 1970, the average annual precipitation is 1426.4 mm and the relative humidity is maintained at approximately 80%. The PLB serves as a crucial ecological barrier along the Yangtze River. However, in recent years, the persistent drought and low rainfall during the dry season in the PLB has had a significant impact on the social and economic development of the basin. This includes issues such as urban and rural water supply, agricultural irrigation, shipping and transportation, and has also resulted in serious ecological problems, including the degradation of wetlands and difficulties in maintaining aquatic organism habitats.

2.2. Data Needs and Sources

SWAT is a data-driven distributed hydrologic model. The data required for its construction primarily encompass land use/land cover (LULC), digital elevation model (DEM), meteorology, soil, and hydrology (Table 1). The data utilized for landscape pattern analysis are predominantly LULC.

2.3. Methodology

2.3.1. Construction of the SWAT Model

The initial step involved delineating the sub-basin and hydrological response unit (HRU). Based on the DEM data (Figure 2) and the D8 algorithm, the catchment area threshold (297,838 ha) recommended by the SWAT model was employed to extract the water system and divide the sub-basin boundaries. These were then corrected and adjusted according to the actual river network pairs present in the study area. Ultimately, the PLB was divided into 31 sub-basins (Figure 2) and 1152 HRUs.
Subsequently, the model calibration and validation were conducted. The SWAT–CUP 2019 software’s sequential uncertainty fitting (SUFI-2) algorithm was employed to conduct a sensitivity analysis on the model parameters. In a sequential manner, beginning with the upstream area and subsequently moving to the downstream area, the measured runoff data from the Lijiadu and Waizhou hydrologic stations were utilized to assess the runoff of the PLB. Upon attaining the optimal simulation, the measured runoff data from the Hukou hydrologic station were incorporated to complete the runoff simulation of the PLB. In accordance with the extant literature, 20 runoff parameters were initially identified for analysis [24,25,26]. Through a global sensitivity analysis, the 10 most representative parameters were then selected. After 7400 iterations, the sensitivity analysis of SWAT model parameters for the PLB was completed (Table 2). The applicability of the SWAT model was evaluated using the fitting coefficient R2 and the Nash–Sutcliffe efficiency (NSE) coefficient. These values, which range from 0 to 1, indicate the quality of the model simulation effect, with a value closer to 1 indicating a more optimal result. It is generally accepted that the model simulation results are credible when the coefficient of determination (R2) exceeds 0.6 and the NSE coefficient exceeds 0.5 [27,28].

2.3.2. Analysis of Landscape Pattern

The establishment of landscape metrics is a key step in the analysis of landscape pattern, including the principles of comprehensiveness, typicality and independence [29,30], and the selected metrics should be able to accurately measure and reflect the landscape pattern with clear meaning. The variance inflation factor (VIF) is the ratio of the variance of the regression coefficient estimates to the variance of the assumption that there is no linear correlation between the independent variables, which is an indicator of the severity of multicollinearity. It is generally believed that, if the value of the VIF is greater than 10, there is a serious problem of multicollinearity between the independent variables; the closer the VIF is to 1, the lower the degree of multicollinearity [31].
The landscape pattern was initially analyzed in terms of four dimensions: area, shape, distribution and diversity. Thereafter, metrics with an evident correlation in the calculation formula, metrics obtained by arithmetical averaging and metrics that did not reflect landscape heterogeneity were excluded from the landscape metrics. The Fragstats 4.2 software was used to retain and calculate 25 landscape metrics (Table 3). Subsequently, covariance diagnostics were performed on the aforementioned 25 landscape metrics, and the landscape metrics with the smallest VIF in each of the four dimensions were selected.
In light of the findings of the covariance diagnosis, the edge density (ED) was identified as the most appropriate measure for characterizing the patch area, the area-weighted related circumscribing circle (CIRCLE_AM) was selected for characterizing the patch shape, the interspersion and juxtaposition index (IJI) was deemed the most suitable for characterizing the patch distribution, and the patch richness density (PRD) was chosen for characterizing the patch diversity in this study. The landscape metrics were calculated using the following formula:
E D = E A × 10000
where E is the total length of the edge in the landscape or class, and A is the total area of the landscape, multiplied by 10,000 to convert to hectares. ED ≥ 0, a larger ED can result in a smaller patch area.
C I R C L E = 1 − a i a i c
where ai is the area of patch i, and aic is the area of the smallest circumscribing circle around patch i. 0 < CIRCLE < 1, CIRCLE approaches 0 for circular patches and approaches 1 for elongated linear patches one cell wide.
I J I = − ∑ j = 1 m p i j p i ln ⁡ p i j p i ln ⁡ m − 1 × 100
where pij is the edge length of the adjacency between patch type i and patch type j, pi is the total edge length of patch type i, and m is the total number of patch types adjoining patch type i, multiplied by 100 to convert to a percentage. At the landscape scale, IJI calculates the total dispersion and juxtaposition of individual patch types. 0 < IJI ≤ 100, IJI approaches 0 as the distribution of adjacencies between individual patch types becomes increasingly unequal, and IJI = 100 when all patch types are equally adjacent to all other patch types.
P R D = m A × 10000 × 100
where m is the number of patch types present in the landscape, excluding the landscape boundary if present, A is the total landscape area, multiplied by 10,000 and 100 to convert to 100 ha.

2.3.3. Analysis of Correlation

In this study, three analytical techniques were employed to examine the correlation between runoff and landscape composition and pattern: Pearson correlation analysis, global linear regression, and geographically weighted regression (GWR) models. Among the aforementioned methods, GWR is a local spatial regression analysis method. It allows the model parameters to vary spatially by introducing spatial coordinates as weights, thereby enabling the capture and accounting of local features and spatial non-stationarity in spatial data, thus providing a more accurate analysis of local spatial relationships [32]. The calculation formula is as follows:
y i = β 0 ( u i , v i ) + ∑ j = 1 p β j ( u i , v i ) x j + ε i
where i = 1, 2, 3… n; yi is the runoff of the ith sub-basin; β0 is the regression constant term for the ith sub-basin, which varies with the geographical location (ui,vi); β0 (ui,vi) is the regression coefficient of the jth independent variable for the ith sub-basin, which also varies with the geographical location (ui,vi); xj is the ith sub-basin, the value of the jth independent variable; and εi is the error term for the ith sub-basin.

3. Results

3.1. Spatial–Temporal Patterns of Runoff

3.1.1. Calibration and Validation of the SWAT Model

The SWAT model is an appropriate tool for simulating runoff in the PLB. The observed flow values of three hydrological stations (Hukou, Lijiadu and Waizhou) exhibit a similar trend to the simulated values of the SWAT model (Figure 3). The coefficient of determination (R2) and NSE values for the calibration period are greater than 0.7, indicating a superior model simulation effect. Among the three hydrological stations, Lijiadu exhibits the highest fit, with an R2 and NSE value of 0.83. The fit effect of the validation period is more pronounced than that of the calibration period, with the R2 and NSE values for the three hydrological stations remaining above 0.7.

3.1.2. Spatial–Temporal Characteristics of Runoff

The runoff exhibited fluctuating changes in the PLB from 2011 to 2020 (Figure 4, Table 4), with a maximum value of 1086.28 mm in 2016, a minimum value of 570.21 mm in 2011, and a multi-year average of 861.17 mm. The years 2012, 2014, 2015, 2016, 2019 and 2020 exhibited runoff values that exceeded the multi-year average, indicating that these were years of abundant precipitation. In contrast, the years 2011, 2013, 2017 and 2018 were distinguished by relatively low runoff, which aligns with the characteristics of dry water years. The coefficients of variation in runoff for 2018, 2019 and 2011 are relatively high, and the differences in runoff are greater than those observed in other years. With regard to the spatial distribution of runoff, a discernible pattern of heterogeneity was observed, with the extremely high levels (>1000 mm) concentrated in the downstream area and the extremely low levels (<700 mm) observed in the upstream area. With regard to the sub-basins, the areas with the highest values of the multi-year average of runoff are sub-basins No. 8, 6 and 4, which have a mean runoff value of 1121.14 mm, 1084.57 mm and 1061.94 mm, respectively. In contrast, sub-basin No. 23, which has a mean runoff value of 659.02 mm, represents the area with the lowest value.
In 2020, the runoff of the PLB was 923.77 ± 169.13 mm (mean ± SD), exhibiting an overall spatial pattern of “high in the north and low in the south” (Figure 3). The area with the highest runoff values (exceeding 1000 mm) is distributed across 11 sub-basins (35.48%), including No. 15 (1195.90 mm), No. 14 (1180.22 mm) and No. 8 (1176.94 mm) in the northern part of the study area. Conversely, the area with the lowest runoff values (below 700 mm) is distributed across the No. 28 (669.38 mm) and No. 25 (679.77 mm) sub-basins in the southern part of the study area.

3.2. Landscape Composition Analysis and Its Impact on Runoff

The forest landscape type is the most prevalent in the PLB (Figure 5), with a forest cover of 61.74% in 2020. The proportion of cropland (26.24%) is the second highest, followed by other landscape types, which collectively account for less than 5%. The smallest proportion is that of barren, which is less than 0.01%. In terms of spatial distribution, cropland was concentrated in the northern part of the study area, while forest was mainly distributed in the southern part. In terms of sub-basins, 67.74% were found to have forests as the dominant landscape type, while 22.58% were identified as having cropland as the dominant landscape type.
The impact of landscape composition on runoff is not linear; rather, it displays a nonlinear character in the PLB (Figure 6). The R2 of the global linear regression model for the runoff to proportion of forest, cropland, grassland, and barren was only 0.04 at the maximum. In contrast, the fit of the GWR model was significantly superior to that of the global linear regression model, with R2 values of 0.78 and 0.75 for cropland and forest, respectively (Figure 7). Furthermore, the R² values for grassland and barren were greater than 0.5, indicating the presence of a spatial relationship between runoff and landscape type. The GWR coefficients indicate that the influence of landscape type on runoff is ranked in descending order as forest (mean of GWR coefficient = 0.18), cropland (mean of GWR coefficient = 0.12), bare ground (mean of GWR coefficient = 0.11), and grassland (mean of GWR coefficient = 0.01).

3.3. Multi-Scale Spatial Relationship Between Runoff and Landscape Pattern

3.3.1. Landscape Pattern Analysis

At the landscape scale, the high-value area of ED (ED > 8) was primarily distributed across eight sub-basins (Figure 7), including No. 21, 19, and 23, etc. Notably, there was no sub-basin of CIRCLE_AM larger than 0.8. The high-value area of IJI (IJI > 80) included two sub-basins, No. 28 and 2, and the distribution of patches in these sub-basins was homogeneous. The high-value area of PRD (PRD > 0.0085) included two sub-basins, No. 23 and 28, and these sub-watersheds exhibited relatively high patch diversity.
At the class scale, the mean value of ED for grassland was relatively low, while the patch area was considerable. In contrast, the mean value of CIRCLE_AM for cropland and forest was high, and the patch shape was approximately linear. In comparison, the patch shape of grassland was more circular. The IJI for cropland was notably high, and the distribution of patch neighbors was the most uniform.

3.3.2. Examination of Spatial Relationship

A spatial relationship exists between runoff and landscape pattern in the PLB. The Pearson significance of runoff and landscape metrics ED, CIRCLE_AM, IJI, and PRD were all greater than 0.05 (Table 5), which did not pass the significance test. In contrast, the GWR model demonstrated superior fit and an R2 value greater than 0.5 at both the landscape and class scales (Figure 8), indicating the presence of a spatial relationship between runoff and landscape pattern. At the landscape scale, the GWR model for runoff and PRD exhibited the optimal fit (R2 = 0.71), while at the class scale, the GWR model for runoff and IJI demonstrated the optimal fit (R2 = 0.72). Furthermore, neither the Pearson correlation analysis nor the GWR model of runoff with the CIRCLE_AM was successfully constructed due to the limited extent of bare land and its distribution in the sub-basin.

3.3.3. Spatial Relationship at Landscape Scale

At the landscape scale, the mean values of the GWR coefficients of runoff with ED, CIRCLE_AM, IJI, and PRD were 0.15, 0.15, −0.23, and −0.60, respectively (Figure 8). These results indicate that, at the landscape scale, patch diversity had the greatest effect on runoff, while patch distribution, area, and shape had relatively little effect on runoff. The mean value of the GWR coefficient of runoff and PRD is negative, indicating that PRD exerts a negative influence on runoff as a whole. Furthermore, the higher the patch diversity, the lower the runoff. The mean value of the GWR coefficient of runoff and IJI is negative, indicating that IJI has a negative influence on runoff. Consequently, the greater the uniformity of the neighbor distribution of patches, the lower the runoff. Conversely, the mean value of the GWR coefficient of runoff and ED is positive, indicating that ED has a positive influence on runoff. Therefore, the higher the patch edge density, the higher the runoff. The mean value of the GWR coefficient between runoff and CIRCLE_AM is positive, indicating that CIRCLE_AM has a positive influence on runoff. Furthermore, the closer the patch shape is to linear, the higher the runoff.

3.3.4. Spatial Relationship at Class Scale

At the class scale for forest, the mean values of the GWR coefficients of runoff with ED, CIRCLE_AM, and IJI were 0.38, 0.10, and −0.23, respectively (Figure 8). These results indicate that the patch area of the forest had the greatest effect on runoff, while the patch distribution and shape had a lesser effect. The results indicated that ED had an overall positive effect on runoff, whereby the greater the edge density of the forest patches, the higher the runoff. Similarly, CIRCLE_AM had a positive effect on runoff, whereby the closer to linear the patch shape of the forest, the higher the runoff. In contrast, IJI had a negative effect on runoff, whereby the more uneven the neighborhood distribution of the forest patches, the higher the runoff.
At the class scale for cropland, the mean values of the GWR coefficients of runoff with ED, CIRCLE_AM, and IJI were 0.35, 0.33, and −0.16, respectively. These results indicate that the patch area and shape of cropland exert a greater influence on runoff, with an overall positive effect. Conversely, the patch distribution of cropland exerts a lesser influence on runoff, with an overall negative effect.
At the class scale for grassland, the mean values of GWR coefficients for runoff with ED, CIRCLE_AM, and IJI were 0.09, −0.12, and −0.001, respectively. This indicates that the patch area, shape, and distribution of grassland had a minor impact on runoff, with the patch distribution of grassland exerting almost no influence on runoff.
At the class scale for barren, the mean values of GWR coefficients of runoff with ED and IJI were 0.20 and 0.18, respectively. This indicates that the patch area and distribution of barren had a relatively minor effect on runoff, with an overall positive influence.
A cross-sectional comparison at the class scale demonstrated that the patch area had the most significant impact on forest runoff, while exerting the least influence on grassland runoff. The analysis revealed that patch shape had the greatest impact on cropland runoff and the least impact on forest runoff. Ultimately, the distribution of patches had the greatest impact on forest runoff and the least impact on grassland runoff.

3.4. Ecological Restoration Pathways in Sub-Basins

Nine sub-basins, including No. 18, 23, and 25, were identified as requiring ecological restoration due to their low (700–800 mm) and extremely low (<700 mm) runoff levels (Figure 1). Subsequently, restoration pathways were formulated based on the positive or negative impacts of landscape metrics on sub-basin runoff, specifically whether the GWR coefficient was greater than 0 or less than 0 (Figure 9). Specifically, eight sub-basins, namely No. 18, 23, 25, along with five others, require an increase in patch edge density, whereas sub-basin No. 31 requires a decrease. The patch shape of sub-basins No. 18, 23, and 26 should tend towards linearity, whereas those of six sub-basins, including No. 25, 27, and 28, should tend towards circularity. Sub-basins No. 18 and 25 require an increase in patch neighbor distribution, whereas seven sub-basins, including No. 23, 26, and 27, require a decrease. Patch diversity reduction is necessary in all nine sub-basins.

4. Discussion

4.1. Sensitivity Analysis and Parameter Optimization of the SWAT Model

The SWAT model incorporates a multitude of parameters, with the default parameters primarily derived from empirical equations pertaining to the hydrological environment of the United States. The initial simulation results differ significantly from the actual observations of the PLB, indicating the necessity for a sensitivity analysis of the parameters. This analysis is essential for identifying the key parameters affecting the hydrological processes in the PLB and adjusting the range of sensitivity parameters to achieve closer alignment between the simulation results and the actual observations [33]. This approach ensures that the model simulation meets the required accuracy standards. The SWAT–CUP software offers five distinct approaches to parameter rate determination. Among these, the SUFI-2 algorithm, employed in the present study, offers the dual advantages of producing the most accurate simulations and the smallest uncertainty with the fewest simulations, while also demonstrating the highest simulation efficiency. This algorithm is currently the most utilized by researchers [34,35,36].
A sensitivity analysis revealed that the parameters SOL_AWC, SOL_CBN, and HRU_SLP exert a significant influence on runoff in the PLB. The SOL_AWC parameter reflects the amount of water that the soil can store, which is a crucial factor for plant absorption and transpiration. In the PLB, elevated SOL_AWC values indicate that the soil is capable of retaining greater quantities of water, which subsequently reduces the volume of surface runoff [37,38,39]. Given the diverse soil types and land use practices present within the Poyang Lake watershed, alterations in SOL_AWC have significant ramifications for the watershed’s water retention capacity and runoff generation. Although SOL_CBN is primarily associated with the dynamics of soil organic matter and nutrient cycling, it indirectly influences the hydrological characteristics of the soil. The content of soil organic matter affects the porosity and water infiltration rate of the soil, which, in turn, affects the formation and movement of runoff [40,41]. In the PLB, alterations in land use and land cover may result in modifications to the SOL_CBN, which subsequently influences the hydrological processes and runoff within the basin. HRU_SLP represents a crucial topographic parameter that affects hydrological processes within the basin. In the Poyang Lake watershed, the presence of a diverse range of topographical features gives rise to changes in HRU_SLP, which in turn influence the velocity and direction of water flow. This, in turn, affects the generation of runoff and the timing of convergence [42,43,44].

4.2. Impact of Landscape Composition on Runoff

In this study, we analyzed the effect of landscape composition on runoff. The findings indicated that forests exert the most significant influence on runoff, which is consistent with the conclusions of other scholars [45,46,47]. Forests, croplands, and grasslands are all integral components of terrestrial ecosystems, and they all play a pivotal role in the generation and regulation of runoff. However, in comparison to farmland and grassland, forests exert a more pronounced influence on runoff. This is likely to be attributable to the fact that forests possess a superior capacity for water retention and storage, which serves to regulate the reduction of flood flow, the minimization of flood peaks, the slowing down of the flooding process, and the augmentation of base flow. During floods, forests are capable of absorbing and storing considerable quantities of precipitation, thereby reducing flood flows and flood peaks. Furthermore, during periods of low precipitation, the water stored in the soil can be released as seepage, recharging rivers and reservoirs and increasing the runoff of rivers during dry periods [48,49,50].

4.3. Impact of Landscape Pattern on Runoff

The findings of this study suggest that, when considered at the landscape scale, patch diversity exerts the most significant influence on runoff. This may be attributed to the fact that patch diversity alters microtopography, which in turn affects water flow paths and velocities, thereby reducing runoff. Furthermore, interactions between different patch types, such as source–sink relationships for resource redistribution and complementary ecological niches, can influence the hydrological regulation function of the overall landscape [51,52]. Watershed runoff is affected by a combination of natural and human-induced factors, with spatial differences in the distribution of these factors. The global linear relationship between runoff and landscape pattern does not fully reflect the complex, non-linear relationship between these variables. In this study, the GWR model was employed to reflect the spatial differentiation of the contribution of landscape units to the regression equations. This was achieved by attaching the parameters of correlation and heterogeneity of the landscape unit itself to the change in the correlation, thereby enhancing the plausibility of the results of the spatial relationship analysis [53,54].
It is not advisable to reduce patch diversity, as it serves as a critical factor in maintaining ecosystem health and stability. Patch diversity provides a range of ecosystem services, including supporting species diversity and enhancing ecosystem resistance and resilience. However, if the necessity arises to address increased runoff by considering reductions in patch diversity, several potential measures can be explored. Firstly, the configuration of different types of patches can be optimized through the implementation of rational land use planning. In ecologically sensitive areas, prioritizing the protection of existing forest and grassland patches is of utmost importance to prevent overdevelopment and destruction. Furthermore, implementing measures for the gradual restoration of damaged ecological patches is essential. Secondly, establishing connections and integration between patches is also significant. For dispersed farmland patches, implementing land improvement and farmland infrastructure measures can facilitate centralization and connectivity, thereby enhancing land use efficiency. In the case of forest and grassland patches, establishing ecological corridors can reinforce ecological connections between patches, facilitating species migration and gene flow.
At the class scale, the patch area of forest and cropland demonstrated the most significant impact on runoff. In the patch edge area, a reduction in vegetation cover results in a decrease in water infiltration, which consequently increases the rate and volume of surface water flow [55]. Consequently, high edge density may lead to an increase in runoff. Furthermore, an increase in edge density results in a larger contact area between patches, thereby enhancing the potential for material, energy, and species exchange. This process facilitates the material cycle and energy flow within the ecosystem [56].
In practice, measures aimed at increasing the density of forest patch edges to enhance runoff have primarily focused on optimizing water mobility and pooling capacity through the strategic layout of forest patches and the application of ecological engineering techniques. Firstly, the configuration of forest patches is designed in a logical manner, particularly in areas designated for water source conservation and in the upstream sections of watersheds. This is intended to create a more open and fluid hydrological network, which is accomplished by increasing the length and connectivity of forest patch edges. This can be realized through the strategic planning of ecological corridors, buffer zones, or water-source conservation forests, which function as runoff channels and regulators, thereby enhancing water mobility and pooling efficiency. In the context of agricultural land, the configuration and arrangement of agricultural land patches are adjusted with the goal of increasing edge length and connectivity. This is primarily achieved through the implementation of farmland improvement and planning initiatives. Configuring farmland patches into shapes with a greater proportion of concave and convex edges, such as wavy or jagged patterns, can facilitate the extension of water flow paths and, consequently, augment the volume of runoff. Simultaneously, ensuring the unobstructed flow of ecological corridors between farmland patches is of utmost importance, as it facilitates water movement and pooling.

4.4. Limitations

In this study, we primarily utilized the sub-basin as the unit of landscape analysis, with a particular focus on the influence and binding of topographic factors on the landscape [57]. The division of HRUs in the SWAT model integrates topography, soil, and land use elements, thereby more accurately representing the spatial heterogeneity of landscape units. Consequently, the next research direction is to analyze the influence of landscape pattern on runoff by using HRUs as a landscape unit.
In this study, the impact of landscape patterns on runoff was isolated for analysis. However, the impact of dams, which are prominent examples of human engineering, is also a pivotal factor in runoff dynamics. Dams regulate the volume of runoff across different seasons by impounding and releasing water, thereby influencing the overall runoff pattern. Furthermore, obtaining data on the volume of water impounded and released by dams is challenging. Consequently, the next phase of research will focus on elucidating the impact of dams on the balance between water supply and demand within the basin, which poses a significant challenge.
Furthermore, SWAT+ is an enhanced water resources modeling tool that builds upon the original SWAT model through optimized code management and improved representation of watershed processes. This enhancement enables the model to more accurately simulate hydrologic processes within a watershed. Therefore, it is recommended that SWAT+ be employed in the subsequent stages of water resources research.

5. Conclusions

In this study, the SWAT model, GWR model, and landscape pattern analysis were used to elucidate the spatial–temporal pattern of runoff in the PLB, reveal the impacts of landscape composition and pattern on runoff, and propose corresponding ecological restoration pathways. The principal findings are as follows:
(1)
The runoff is distinguished by notable spatial–temporal heterogeneity, with the runoff exhibiting fluctuating changes from 2011 to 2020.
(2)
The PLB is predominantly characterized by forest landscapes, with a forest cover of 61.74% in 2020. The impact of landscape composition on runoff exhibits a non-linear characteristic, with the order of impact being forest > cropland > barren land > grassland. Forest has the most significant impact on runoff.
(3)
There is a spatial relationship between runoff and landscape patterns. At the landscape scale, patch diversity has the most significant impact on runoff. Consequently, runoff can be optimized primarily by reducing patch richness. At the class scale, the patch area of forests and croplands has the greatest impact on runoff, which can be enhanced primarily by increasing the density of patch edges and facilitating the circulation and flow of materials within the landscape.
(4)
Nine sub-basins requiring ecological restoration were identified, and restoration pathways were developed based on the spatial relationships between runoff and landscape patterns.

Author Contributions

Conceptualization, P.D. and Y.T.; methodology, P.D. and Y.T.; software, P.D. and Y.T.; validation, P.D. and Y.T.; formal analysis, P.D. and Y.T.; investigation, J.Z.; resources, Y.F.; data curation, Y.F.; writing—original draft preparation, P.D. and Y.T.; writing—review and editing, P.D. and Y.T.; visualization, P.D., Y.T. and J.Z.; supervision, P.D.; project administration, P.D.; funding acquisition, P.D. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Social Science Foundation of Fujian Province, grant number FJ2024C086 and the Startup Fund for Advanced Talents of Putian University, grant number 2023012.

Data Availability Statement

Links to the data are available in the Section 2 of this article.

Acknowledgments

The authors are very grateful to the editors and reviewers for their generous help in the publication of this paper.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Vörösmarty, C.J.; Hoekstra, A.Y.; Bunn, S.E.; Conway, D.; Gupta, J. Fresh water goes global. Science 2015, 349, 478–479. [Google Scholar] [CrossRef] [PubMed]
  2. Qin, Y.; Mueller, N.D.; Siebert, S.; Jackson, R.B.; AghaKouchak, A.; Zimmerman, J.B.; Tong, D.; Hong, C.P.; Davis, S.J. Flexibility and intensity of global water use. Nat. Sustain. 2019, 2, 515–523. [Google Scholar] [CrossRef]
  3. Wang, S.; Zhang, K.; Chao, L.J.; Li, D.H.; Tian, X.; Bao, H.J.; Chen, G.D.; Xia, Y. Exploring the utility of radar and satellite-sensed precipitation and their dynamic bias correction for integrated prediction of flood and landslide hazards. J. Hydrol. 2021, 603, 126964. [Google Scholar] [CrossRef]
  4. Yu, C.; Hu, D.Y.; Shao, H.Y.; Dai, X.A.; Liu, G.; Wu, S. Runoff simulation driven by multi-source satellite data based on hydrological mechanism algorithm and deep learning network. J. Hydrol. Reg. Stud. 2024, 52, 101720. [Google Scholar] [CrossRef]
  5. Devia, G.K.; Ganasri, B.P.; Dwarakish, G.S. A review on hydrological models. Aquat. Proc. 2015, 4, 1001–1007. [Google Scholar] [CrossRef]
  6. Seiller, G.; Anctil, F.; Perrin, C. Multimodel evaluation of twenty lumped hydrological models under contrasted climate conditions. Hydrol. Earth Syst. Sci. 2012, 16, 1171–1189. [Google Scholar] [CrossRef]
  7. Rodriguez, F.; Andrieu, H.; Morena, F. A distributed hydrological model for urbanized areas–Model development and application to case studies. J. Hydrol. 2008, 351, 268–287. [Google Scholar] [CrossRef]
  8. Xie, K.; Liu, P.; Zhang, J.Y.; Wang, G.Q.; Zhang, X.J.; Zhou, L.T. Identification of spatially distributed parameters of hydrological models using the dimension-adaptive key grid calibration strategy. J. Hydrol. 2021, 598, 125772. [Google Scholar] [CrossRef]
  9. Tan, M.L.; Gassman, P.W.; Yang, X.Y.; Haywood, J. A review of SWAT applications, performance and future needs for simulation of hydro-climatic extremes. Adv. Water Resour. 2020, 143, 103662. [Google Scholar] [CrossRef]
  10. Dong, J.Q.; Jiang, H.; Gu, T.W.; Liu, Y.X.; Peng, J. Sustainable landscape pattern: A landscape approach to serving spatial planning. Landsc. Ecol. 2022, 37, 31–42. [Google Scholar] [CrossRef]
  11. Xiong, Y.; Mo, S.H.; Wu, H.P.; Qu, X.Y.; Liu, Y.Y.; Zhou, L. Influence of human activities and climate change on wetland landscape pattern—A review. Sci. Total Environ. 2023, 879, 163112. [Google Scholar] [CrossRef] [PubMed]
  12. Lyu, L.T.; Bi, S.Q.; Yang, Y.; Zheng, D.F.; Li, Q. Effects of vegetation distribution and landscape pattern on water conservation in the Dongjiang River basin. Ecol. Indic. 2023, 155, 111017. [Google Scholar] [CrossRef]
  13. Liu, C.X.; Chen, Y.N.; Huang, W.J.; Fang, G.H.; Li, Z.; Zhu, C.G.; Liu, Y.C. Climate warming positively affects hydrological connectivity of typical inland river in arid Central Asia. NPJ Clim. Atmos. Sci. 2024, 7, 250. [Google Scholar] [CrossRef]
  14. Lyu, Y.S.; Chen, H.; Cheng, Z.; He, Y.T.; Zheng, X. Identifying the impacts of land use landscape pattern and climate changes on streamflow from past to future. J. Environ. Manag. 2023, 345, 118910. [Google Scholar] [CrossRef] [PubMed]
  15. Zhu, Z.T.; Wang, H.L.; Yang, Z.H.; Huai, W.X.; Huang, D.; Chen, X.H. Landscape Pattern Changes of Aquatic Vegetation Communities and Their Response to Hydrological Processes in Poyang Lake, China. Water 2024, 16, 1482. [Google Scholar] [CrossRef]
  16. Hou, Y.P.; Wei, X.H.; Zhang, M.F.; Creed, I.F.; McNulty, S.G.; Ferraz, S.F.B. A global synthesis of hydrological sensitivities to deforestation and forestation. Forest Ecol. Manag. 2023, 529, 120718. [Google Scholar] [CrossRef]
  17. Guo, W.X.; Hu, J.W.; Wang, H.X. Analysis of runoff variation characteristics and influencing factors in the Wujiang River Basin in the past 30 years. Int. J. Environ. Res. Public Health 2021, 19, 372. [Google Scholar] [CrossRef] [PubMed]
  18. Qin, J.; Liu, Y.X.; Chang, Y.P.; Liu, S.W.; Pu, H.Z.; Zhou, J.X. Regional runoff variation and its response to climate change and human activities in Northwest China. Environ. Earth Sci. 2016, 75, 1366. [Google Scholar] [CrossRef]
  19. Wang, Y.Y.; Molinos, J.G.; Shi, L.L.; Zhang, M.; Wu, Z.S.; Zhang, H.; Xu, J. Drivers and changes of the Poyang Lake wetland ecosystem. Wetlands 2019, 39, 35–44. [Google Scholar] [CrossRef]
  20. Liu, H.; Zheng, L.; Jiang, L.; Liao, M.W. Forty-year water body changes in Poyang Lake and the ecological impacts based on Landsat and HJ-1 A/B observations. J. Hydrol. 2020, 589, 125161. [Google Scholar] [CrossRef]
  21. Tian, M.M.; Mao, J.Q.; Wang, K.; Xu, D.D. Spatio-temporal heterogeneity of ecological water level in Poyang Lake, China. Ecol. Inform. 2024, 82, 102694. [Google Scholar] [CrossRef]
  22. Liu, M.Y.; Zhang, P.P.; Cai, Y.P.; Chu, J.W.; Li, Y.L.; Wang, X.; Li, C.H.; Liu, Q. Spatial-temporal heterogeneity analysis of blue and green water resources for Poyang Lake basin, China. J. Hydrol. 2023, 617, 128983. [Google Scholar] [CrossRef]
  23. Zou, R.; Wang, X.J.; Yin, Y.X.; Ma, X.Y.; Yang, X.Q.; Huang, P.N.; Ullah, I. Comparing spatio-temporal propagation patterns of hydrological and meteorological droughts: Insights from SWAT modelling in the poyang lake basin. Catena 2024, 243, 108183. [Google Scholar] [CrossRef]
  24. Verma, S.K.; Prasad, A.D.; Verma, M.K. An assessment of ongoing developments in water resources management incorporating SWAT model: Overview and perspectives. Nat. Environ. Pollut. Technol. 2022, 21, 1963–1970. [Google Scholar] [CrossRef]
  25. Feng, Q.Y.; Chen, L.D.; Yang, L.; Yen, H.; Wang, R.Y.; Wu, F.; Feng, Y.; Raj, C.; Engel, B.A.; Omani, N.; et al. A distributed model parameter optimization toolbox performing multisite calibration in the lump and distributed mode for the SWAT model. Environ. Modell. Softw. 2023, 168, 105785. [Google Scholar] [CrossRef]
  26. Xiang, X.; Ao, T.Q.; Xiao, Q.T.; Li, X.D.; Zhou, L.; Chen, Y.; Bi, Y.; Guo, J.Y. Parameter sensitivity analysis of SWAT modeling in the Upper Heihe River Basin using four typical approaches. Appl. Sci. 2022, 12, 9862. [Google Scholar] [CrossRef]
  27. Bennour, A.; Jia, L.; Menenti, M.; Zheng, C.L.; Zeng, Y.L.; Asenso Barnieh, B.; Jiang, M. Calibration and validation of SWAT model by using hydrological remote sensing observables in the Lake Chad Basin. Remote Sens. 2022, 14, 1511. [Google Scholar] [CrossRef]
  28. Zhao, L.X.; Li, H.Y.; Li, C.H.; Zhao, Y.L.; Du, X.Q.; Ye, X.Y.; Li, F.P. Enhanced SWAT calibration through intelligent range-based parameter optimization. J. Environ. Manag. 2024, 367, 121933. [Google Scholar] [CrossRef]
  29. Zhou, Z.X.; Li, J. The correlation analysis on the landscape pattern index and hydrological processes in the Yanhe watershed, China. J. Hydrol. 2015, 524, 417–426. [Google Scholar] [CrossRef]
  30. Jia, Y.Q.; Tang, L.N.; Xu, M.; Yang, X.Y. Landscape pattern indices for evaluating urban spatial morphology–A case study of Chinese cities. Ecol. Indic. 2019, 99, 27–37. [Google Scholar] [CrossRef]
  31. Cheng, J.H.; Sun, J.; Yao, K.S.; Xu, M.; Cao, Y. A variable selection method based on mutual information and variance inflation factor. Spectrochim. Acta A 2022, 268, 120652. [Google Scholar] [CrossRef] [PubMed]
  32. Sisman, S.; Aydinoglu, A.C. A modelling approach with geographically weighted regression methods for determining geographic variation and influencing factors in housing price: A case in Istanbul. Land Use Policy 2022, 119, 106183. [Google Scholar] [CrossRef]
  33. Shi, P.; Chen, C.; Srinivasan, R.; Zhang, X.S.; Cai, T.; Fang, X.Q.; Qu, S.M.; Chen, X.; Li, Q.F. Evaluating the SWAT model for hydrological modeling in the Xixian watershed and a comparison with the XAJ model. Water Resour. Manag. 2011, 25, 2595–2612. [Google Scholar] [CrossRef]
  34. Sao, D.; Kato, T.; Tu, L.H.; Thouk, P.; Fitriyah, A.; Oeurng, C. Evaluation of different objective functions used in the sufi-2 calibration process of swat-cup on water balance analysis: A case study of the Pursat river basin, Cambodia. Water 2020, 12, 2901. [Google Scholar] [CrossRef]
  35. Wu, H.J.; Chen, B. Evaluating uncertainty estimates in distributed hydrological modeling for the Wenjing River watershed in China by GLUE, SUFI-2, and ParaSol methods. Ecol. Eng. 2015, 76, 110–121. [Google Scholar] [CrossRef]
  36. Narsimlu, B.; Gosain, A.K.; Chahar, B.R.; Singh, S.K.; Srivastava, P.K. SWAT model calibration and uncertainty analysis for streamflow prediction in the Kunwari River Basin, India, using sequential uncertainty fitting. Environ. Process. 2015, 2, 79–95. [Google Scholar] [CrossRef]
  37. Athira, P.; Sudheer, K.P. Calibration of distributed hydrological models considering the heterogeneity of the parameters across the basin: A case study of SWAT model. Environ. Earth Sci. 2021, 80, 131. [Google Scholar] [CrossRef]
  38. Cai, Y.F.; Zhang, F.; Shi, J.C.; Johnson, V.C.; Ahmed, Z.; Wang, J.G.; Wang, W.W. Enhancing SWAT model with modified method to improve Eco-hydrological simulation in arid region. J. Clean. Prod. 2023, 403, 136891. [Google Scholar] [CrossRef]
  39. Panikkar, U.R.; Srivastav, R.; Srivastava, A. Multiscale variability of hydrological responses in urbanizing watershed. Remote Sens. 2023, 15, 796. [Google Scholar] [CrossRef]
  40. Basset, C.; Abou Najm, M.; Ghezzehei, T.; Hao, X.X.; Daccache, A. How does soil structure affect water infiltration? A meta-data systematic review. Soil Tillage Res. 2023, 226, 105577. [Google Scholar] [CrossRef]
  41. Vereecken, H.; Huisman, J.A.; Hendricks Franssen, H.J.; Brüggemann, N.; Bogena, H.R.; Kollet, S.; Javaux, M.; Kruk, J.; Vanderborght, J. Soil hydrology: Recent methodological advances, challenges, and perspectives. Water Resour. Res. 2015, 51, 2616–2633. [Google Scholar] [CrossRef]
  42. Jiang, L.P.; Zhu, J.H.; Chen, W.; Hu, Y.M.; Yao, J.; Yu, S.; Jia, G.L.; He, X.Y.; Wang, A.Z. Identification of suitable hydrologic response unit thresholds for soil and water assessment tool streamflow modelling. Chin. Geogr. Sci. 2021, 31, 696–710. [Google Scholar] [CrossRef]
  43. Li, Y.; DeLiberty, T. Evaluating hourly SWAT streamflow simulations for urbanized and forest watersheds across northwestern Delaware, US. Stoch. Environ. Res. Risk A 2021, 35, 1145–1159. [Google Scholar] [CrossRef]
  44. Pang, S.J.; Wang, X.Y.; Melching, C.S.; Feger, K.H. Development and testing of a modified SWAT model based on slope condition and precipitation intensity. J. Hydrol. 2020, 588, 125098. [Google Scholar] [CrossRef]
  45. Shrestha, M.K.; Recknagel, F.; Frizenschaf, J.; Meyer, W. Assessing SWAT models based on single and multi-site calibration for the simulation of flow and nutrient loads in the semi-arid Onkaparinga catchment in South Australia. Agr. Water Manag. 2016, 175, 61–71. [Google Scholar] [CrossRef]
  46. Chiang, L.C.; Shih, P.C.; Lu, C.M.; Jhong, B.C. Strategies analysis for improving SWAT model accuracy and representativeness of calibrated parameters in sediment simulation for various land use and climate conditions. J. Hydrol. 2023, 626, 130124. [Google Scholar] [CrossRef]
  47. Zhang, X.; Srinivasan, R.; Bosch, D. Calibration and uncertainty analysis of the SWAT model using Genetic Algorithms and Bayesian Model Averaging. J. Hydrol. 2009, 374, 307–317. [Google Scholar] [CrossRef]
  48. Liu, D.J.; Esquivel-Muelbert, A.; Acil, N.; Astigarraga, J.; Cienciala, E.; Fridman, J.; Kunstler, G.; Matthews, T.J.; Ruiz-Benito, P.; Sadler, J.P.; et al. Mapping multi-dimensional variability in water stress strategies across temperate forests. Nat. Commun. 2024, 15, 8909. [Google Scholar] [CrossRef]
  49. Farooqi, T.J.A.; Portela, R.; Xu, Z.; Pan, S.L.; Irfan, M.; Ali, A. Advancing forest hydrological research: Exploring global research trends and future directions through scientometric analysis. J. For. Res. 2024, 35, 128. [Google Scholar] [CrossRef]
  50. Li, K.; Wang, G.X.; Gao, J.H.; Guo, L.M.; Li, J.L.; Guan, M.H. The rainfall threshold of forest cover for regulating extreme floods in mountainous catchments. Catena 2024, 236, 107707. [Google Scholar] [CrossRef]
  51. Sun, W.Y.; Mu, X.M.; Gao, P.; Zhao, G.J.; Li, J.Y.; Zhang, Y.Q.; Chiew, F. Landscape patches influencing hillslope erosion processes and flow hydrodynamics. Geoderma 2019, 353, 391–400. [Google Scholar] [CrossRef]
  52. Wu, J.S.; Sha, W.; Zhang, P.H.; Wang, Z.Y. The spatial non-stationary effect of urban landscape pattern on urban waterlogging: A case study of Shenzhen City. Sci. Rep. 2020, 10, 7369. [Google Scholar] [CrossRef] [PubMed]
  53. Wang, N.X.; Yan, H.Z.; Long, K.L.; Wang, Y.T.; Li, S.X.; Lei, P. Impact of greenspaces and water bodies on hydrological processes in an urbanizing area: A case study of the Liuxi River Basin in the Pearl River Delta, China. Ecol. Indic. 2023, 156, 111083. [Google Scholar] [CrossRef]
  54. Shi, P.; Li, P.; Li, Z.B.; Sun, J.M.; Wang, D.J.; Min, Z.Q. Effects of grass vegetation coverage and position on runoff and sediment yields on the slope of Loess Plateau, China. Agric. Water Manag. 2022, 259, 107231. [Google Scholar] [CrossRef]
  55. Vanneste, T.; Depauw, L.; De Lombaerde, E.; Meeussen, C.; Govaert, S.; De Pauw, K.; Sanczuk, P.; Bollmann, K.; Brunet, J.; Calders, K.; et al. Trade-offs in biodiversity and ecosystem services between edges and interiors in European forests. Nat. Ecol. Evol. 2024, 8, 880–887. [Google Scholar] [CrossRef]
  56. Willmer, J.N.G.; Puettker, T.; Prevedello, J.A. Global impacts of edge effects on species richness. Biol. Conserv. 2022, 272, 109654. [Google Scholar] [CrossRef]
  57. Li, J.; Zhou, Z.X. Coupled analysis on landscape pattern and hydrological processes in Yanhe watershed of China. Sci. Total Environ. 2015, 505, 927–938. [Google Scholar] [CrossRef]
Water 16 03501 g001
Figure 1. Location of the Poyang Lake Basin.
Figure 1. Location of the Poyang Lake Basin.
Water 16 03501 g001
Water 16 03501 g002
Figure 2. Elevation and sub-basin division in the Poyang Lake Basin. Note: The numbers in the right part of this figure represent the sub-basin numbers.
Figure 2. Elevation and sub-basin division in the Poyang Lake Basin. Note: The numbers in the right part of this figure represent the sub-basin numbers.
Water 16 03501 g002
Water 16 03501 g003
Figure 3. A comparative analysis of the flow dynamics observed and simulated in the Poyang Lake Basin. Note: 201101 represents January 2011.
Figure 3. A comparative analysis of the flow dynamics observed and simulated in the Poyang Lake Basin. Note: 201101 represents January 2011.
Water 16 03501 g003
Water 16 03501 g004
Figure 4. Spatial–temporal pattern of runoff in the Poyang Lake Basin from 2011 to 2020.
Figure 4. Spatial–temporal pattern of runoff in the Poyang Lake Basin from 2011 to 2020.
Water 16 03501 g004
Water 16 03501 g005
Figure 5. Spatial distribution of landscape types and proportions in the Poyang Lake Basin.
Figure 5. Spatial distribution of landscape types and proportions in the Poyang Lake Basin.
Water 16 03501 g005
Water 16 03501 g006
Figure 6. Spatial distribution of geographically weighted regression coefficients between runoff and landscape type in the Poyang Lake Basin.
Figure 6. Spatial distribution of geographically weighted regression coefficients between runoff and landscape type in the Poyang Lake Basin.
Water 16 03501 g006
Water 16 03501 g007
Figure 7. Spatial distribution of key landscape metrics at the landscape and class scales in the Poyang Lake Basin.
Figure 7. Spatial distribution of key landscape metrics at the landscape and class scales in the Poyang Lake Basin.
Water 16 03501 g007
Water 16 03501 g008
Figure 8. Spatial distribution of geographically weighted regression coefficients between runoff and landscape metrics at landscape and class scales in the Poyang Lake Basin. Note: R2 = N/A indicates that the GWR model was not successfully constructed.
Figure 8. Spatial distribution of geographically weighted regression coefficients between runoff and landscape metrics at landscape and class scales in the Poyang Lake Basin. Note: R2 = N/A indicates that the GWR model was not successfully constructed.
Water 16 03501 g008
Water 16 03501 g009
Figure 9. Ecological restoration pathways in sub-basins of the Poyang Lake Basin.
Figure 9. Ecological restoration pathways in sub-basins of the Poyang Lake Basin.
Water 16 03501 g009
Table 1. Data needs and sources.
Table 1. Data needs and sources.
NameData DescriptionData Sources
Land use/land cover (LULC)The dataset comprises a raster with a spatial resolution of 30 m, which classifies LULC types including cropland, forest, grassland, barren, water, and impervious.1985–2022 China’s land cover products (http://irsip.whu.edu.cn/resources/CLCD.php, accessed on 16 May 2023)
Digital elevation model (DEM)The data are provided at a spatial resolution of 30 m, with units in meters.Geospatial data cloud (https://www.gscloud.cn, accessed on 16 May 2023)
MeteorologyThe dataset comprises daily precipitation, temperature, relative humidity, hours of sunshine, average wind speed, and other meteorological variables from six stations: Yongxin, Duchang, Zhangshu, Dexing, Nanfeng, and Xingguo. The data span the period from 2009 to 2020 and are presented in text format.National Meteorological Science Data Center (http://data.cma.cn, accessed on 6 May 2023)
Soil types1:1 million items of data on soil types in China.Institute of Soil Science, Chinese Academy of Sciences (http://www.issas.cas.cn, accessed on 10 May 2023)
Soil propertiesThe dataset includes information on the number of soil horizons, the maximum root depth in the soil profile, the depth from the soil surface to the soil subsoil, the organic matter content, and the soil particle composition.Soil Science Database (http://vdb3.soil.csdb.cn, accessed on 10 May 2023)
HydrographyThe dataset comprises monthly runoff data from three hydrologic stations (Hukou, Lijiadu, and Waizhou) for the period 2009–2020, in text format.China Hydrographic Yearbook: Hydrographic Data of the Yangtze River Basin
Table 2. Parameter sensitivity analysis of the SWAT model.
Table 2. Parameter sensitivity analysis of the SWAT model.
NamePhysical SignificanceRangeOptimal ValueSensitivity Ranking
SOL_AWCSoil water available capacity[−1, 1]−0.1151
SOL_CBNSoil carbon content[0.05, 10]7.1852
HRU_SLPHydrologic response unit slope[0, 1]0.4733
ESCOEvaporation soil cover coefficient[0, 1]1.2944
CN2SCS curve number[35, 98]20.4125
SOL_KSaturated hydraulic conductivity[0, 2000]−372.5316
CANMXMaximum canopy storage[0, 100]−27.0097
SLSUBBSNSubbasin length for overland flow[10, 150]58.9168
CH_N2Manning’s roughness coefficient for the main channel[0, 1]−0.0199
SOL_BDSoil bulk density[0.9, 2.5]1.84710
Table 3. Landscape metrics and applicable scales.
Table 3. Landscape metrics and applicable scales.
DimensionLandscape MetricsAbbreviationApplicable Scales
AreaLargest Patch IndexLPILandscape/Class
Edge DensityEDLandscape/Class
AreaAREA_AMLandscape/Class
Radius of GyrationGYRATE_AMLandscape/Class
ShapePerimeter-Area RatioPARA_AMLandscape/Class
Shape IndexSHAPE_AMLandscape/Class
Fractal Dimension IndexFRAC_AMLandscape/Class
Related Circumscribing CircleCIRCLE_AMLandscape/Class
Contiguity IndexCONTIG_AMLandscape/Class
AggregationEuclidean Nearest-Neighbor DistanceENN_AMLandscape/Class
Patch DensityPDLandscape/Class
Landscape Division IndexDIVISIONLandscape/Class
Splitting IndexSPLITLandscape/Class
Effective Mesh SizeMESHLandscape/Class
Interspersion and Juxtaposition IndexIJILandscape/Class
Aggregation IndexAILandscape/Class
Landscape Shape IndexLSILandscape/Class
Patch Cohesion IndexCOHESIONLandscape/Class
DiversityPatch Richness DensityPRDLandscape
Shannon’s Diversity IndexSHDILandscape
Simpson’s Diversity IndexSIDILandscape
Modified Simpson’s Diversity IndexMSIDILandscape
Shannon’s Evenness IndexSHEILandscape
Simpson’s Evenness IndexSIEILandscape
Modified Simpson’s Evenness IndexMSIEILandscape
Table 4. Statistics for runoff in the Poyang Lake Basin from 2011 to 2020 (Unit: mm).
Table 4. Statistics for runoff in the Poyang Lake Basin from 2011 to 2020 (Unit: mm).
YearMeanStandard DeviationCoefficient of Variation
2011570.21139.140.24
2012912.84152.880.17
2013796.9122.240.15
20141002.3121.870.12
2015954.41155.450.16
20161086.28237.470.22
2017759.47121.560.16
2018633.88186.580.29
2019971.69246.950.25
2020923.77169.130.18
Table 5. A Pearson correlation analysis was conducted to examine the relationship between runoff and landscape metrics in the Poyang Lake Basin. Note: N/A because PRD is only applicable at the landscape scale.
Table 5. A Pearson correlation analysis was conducted to examine the relationship between runoff and landscape metrics in the Poyang Lake Basin. Note: N/A because PRD is only applicable at the landscape scale.
ScalePearson CorrelationEDCIRCLE_AMIJIPRD
Landscape scaleCoefficient0.120.14−0.19−0.35
Sig. (2-tailed)0.530.440.320.06
Class scale_ForestCoefficient0.150.05−0.32N/A
Sig. (2-tailed)0.430.790.08N/A
Class scale_CroplandCoefficient0.280.31−0.05N/A
Sig. (2-tailed)0.130.090.79N/A
Class scale_GrasslandCoefficient−0.26−0.06−0.04N/A
Sig. (2-tailed)0.160.750.83N/A
Class scale_BarrenCoefficient0.15N/A0.14N/A
Sig. (2-tailed)0.41N/A0.46N/A
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

Dou, P.; Tian, Y.; Zhang, J.; Fan, Y. Multi-Scale Spatial Relationship Between Runoff and Landscape Pattern in the Poyang Lake Basin of China. Water 2024, 16, 3501. https://doi.org/10.3390/w16233501

AMA Style

Dou P, Tian Y, Zhang J, Fan Y. Multi-Scale Spatial Relationship Between Runoff and Landscape Pattern in the Poyang Lake Basin of China. Water. 2024; 16(23):3501. https://doi.org/10.3390/w16233501

Chicago/Turabian Style

Dou, Panfeng, Yunfeng Tian, Jinfeng Zhang, and Yi Fan. 2024. "Multi-Scale Spatial Relationship Between Runoff and Landscape Pattern in the Poyang Lake Basin of China" Water 16, no. 23: 3501. https://doi.org/10.3390/w16233501

APA Style

Dou, P., Tian, Y., Zhang, J., & Fan, Y. (2024). Multi-Scale Spatial Relationship Between Runoff and Landscape Pattern in the Poyang Lake Basin of China. Water, 16(23), 3501. https://doi.org/10.3390/w16233501

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

Article Metrics

Back to TopTop