Correlation Analysis of Wetland Pattern Changes and Groundwater in Kaifeng Downstream of the Yellow River, China

Abstract

(1) Background: This study aims to provide a viable theoretical framework for wetland ecological restoration in the lower reaches of the Yellow River within the city of Kaifeng, China. (2) Methods: Using remote sensing and image interpretation to identify the long-term evolution characteristics of wetlands in the study area and analyzing the impact of runoff, riverway changes, and groundwater flow fields in the lower reaches of the Yellow River on wetland conditions along the Yellow River. (3) Results: With natural wetland as its major wetland type, the study area saw an increase in the total wetland area from 2000–2021. Among others, the total area of artificial wetlands increased by 43%, while that of flooding wetlands in natural wetlands decreased by 37%. Surface water discharge and water level saw a year-by-year drop. Moreover, the significant wandering and oscillations of riverways led to a direct impact on the area and stability of tidal flat wetlands. After 2010, affected by rainfall and exploitation, the groundwater level declined sharply. The degraded areas of artificial wetlands were mainly distributed at the northern embankment of the Yellow River, where the groundwater burial depth decreased significantly. In contrast, at the southern embankment, for the sake of the irrigation canal diverted from the Yellow River, new back river depressions had formed and helped build a more stable ecological environment. Yellow River water levels and discharge directly impacted the area of rivers and flooding wetlands. The decline in groundwater levels led to the degradation of ponds in artificial wetlands. (4) Conclusions: The reduction of groundwater exploitation and an adequate supply of diverted Yellow River water were conducive to the development of wetlands in the back river depressions on the outside of the Yellow River embankment.

1. Introduction

Wetlands, often referred to as the Earth’s kidneys, rank among the most important ecosystems globally. The lower reaches of the Yellow River, due to frequent wandering riverways and seasonal flood plains, feature a unique wetland ecosystem.

The downstream wetlands of the Yellow River boast rich ecological and water resources, playing an important role in maintaining the ecological balance of the Yellow River Basin [1,2]. However, the improper utilization of those wetlands, such as intensive human activities and development, has led to an overall shrinkage in wetland area, a drop of approximately 35.6% since the 1980s [3]. The ensuing considerable reduction of vegetation [4] leads to a decline in the quality of ecosystem services and biodiversity [5] and different degrees of pollution to wetland water [6].

Rainfall, groundwater, and Yellow River water are considered potential water sources to feed the riverside wetlands in the lower reaches of the Yellow River [7] with variant contributions spatially and temporally. For instance, flooding wetlands are fed by the Yellow River’s seasonal flooding and groundwater, while the non-flooding wetlands are fed by rainwater and groundwater [8,9]. The flooding wetlands along wandering riverways are susceptible to conditions of Yellow River water and sand, as well as the riverway boundaries [10]. Frequent riverway changes [11,12] trigger constant conversions between wetlands and waters [13,14].

Remote sensing satellite data, hailed for timeliness, wide coverage, and lower cost, serves as a primary data source for monitoring dynamic changes in wetlands [15]. Combining remote sensing and GIS data facilitates the analysis and promotion of research into the evolution of wetland morphology and spatial changes. Remote sensing methods support the quantitative analysis of regional landscape patterns [16,17] and the monitoring and assessment of landscape pattern changes. Those technologies aid in data analysis and can be utilized to formulate and implement measures to protect wetlands and slow down their degradation.

Research on the downstream Yellow River wetlands has so far mainly focused on the dynamic impacts of the Xiaolangdi Reservoir’s operation on single or multiple factors like runoff, sediment, riverway morphology, and wetland area. There are relatively few studies on the impact of surface water–groundwater systems on wetland development. The water and sediment of the Yellow River continue to change, and human activities are intense. This study explores the degree to which surface water and groundwater influence the wetland patterns along the Yellow River, with an analysis of the spatial and temporal variations in wetland locations and patterns from 2000 to 2021 (divided into five shorter periods). This study provides a viable theoretical framework for wetland ecological restoration in the lower reaches of the Yellow River. This study focuses on the wetlands in the Kaifeng downstream of the Yellow River, characterized by wandering riverways and dramatic changes in wetland landscape for the sake of overlong development and intense human activities, making it a representative region in the downstream wetlands of the Yellow River.

2. Study Area

Kaifeng, located on the south bank of the Yellow River, China (Figure 1a), lies in a temperate zone of monsoon-influenced climate with four distinct seasons. The temperature averages 14.52 °C, and the annual rainfall is 627.5 mm, the majority of which is concentrated in the summer months of July and August. As the only surface river in this region, this section of the Yellow River features a wide and shallow riverbed with significant oscillations in the mainstream, making it a typical wandering riverway prone to frequent flooding.

The area belongs to the upper-middle part of the Yellow River alluvial fan, characterized by a topography of Yellow River alluvial plains. The Yellow River embankment has a height difference of 4–10 m, with butterfly-shaped depressions between the embankments. Groundwater is mainly found in loose deposits of the Quaternary System, in which the aquifer bed’s thickness is 30–80 m. The lithology of the shallow aquifer bed consists mostly of sand layers of various grain sizes (Figure 1b).

3. Research Data and Methods

3.1. Research Data

The research data primarily includes groundwater level monitoring data and land use/land cover change (LUCC) data (Chinese Academy of Sciences, Beijing, China).

The former consists of 155 samples collected in June 2022 from shallow groundwater wells (Figure 1c). The monitoring was completed within a 15-day period, meeting the requirement of duration. The monitoring density was approximately 4 points per 100 km², fulfilling the requirement and drawing a relatively accurate picture of the groundwater flow characteristics.

The latter was derived from Landsat series satellite images with a 30-m resolution, sourced from the Geospatial Data Cloud (https://www.gscloud.cn/). Considering such factors as vegetation growth cycles, image acquisition times, and cloud cover, images from June to September of the years 2000, 2006, 2011, 2016, and 2021 were selected for analysis.

3.2. Method

3.2.1. Remote Sensing Interpretation

The preprocessing of remote sensing images mainly includes geometric correction, image cropping, image enhancement, and so on. The typical remote sensing interpretation signs are established, and the human–computer interactive interpretation method is used. Correct the classification results according to the historical data and the results of field investigation of simple random sampling (Figure 2) to ensure the accuracy of the final extraction. Wetland classification [18] is a national standard in China, which refers to the Ramsar classification system and is suitable for the actual situation of wetland types in China. According to the standard, the wetland in the lower reaches of the Yellow River is interpreted. There are only the following four wetland types in the study area, and the interpretation standards are shown in

Table 1. Remote sensing interpretation signs of main types of wetlands.

Wetland Types		Type Declarations	Remote Sensing Image Interpretation Signs
Natural wetlands	River	The water surface between the natural or artificially excavated river water level shoreline (excluding the reservoir water surface).
	Flooding wetlands	Tidal flats between naturally formed or artificially excavated river flood level shorelines.
Artificial wetlands	Ponds and ditches	Artificial excavation of water storage area.
	Reservoirs	An area of natural formation or artificial excavation in which the surface is permanently covered by water.

. A series of wetland distribution maps of the study area were drawn in ArcGIS 10.6.

3.2.2. Mann–Kendall Test

The Mann–Kendall test (MK test) is a non-parametric test method widely used and recommended by the World Meteorological Organization (WMO). It does not need the sample to follow a specific distribution and is less disturbed by outliers [19]. The M-K trend test is often used to evaluate the trend and significance of time series of meteorological and hydrological elements [20]. When the upward trend (UF) is positive, it shows an upward trend, and when UF is negative, it shows a downward trend; if the UF curve exceeds the critical line of the significant value, it indicates that the upward or downward trend is significant, and the range exceeded is defined as the mutation period; if the UF and the upper bound (UB) curves have an intersection point between the two critical lines, it indicates that this point is the start time of the mutation test. In this paper, the M-K trend test method is used to study the variation characteristics of rainfall and groundwater level and to reveal the trend change of long-time scale.

3.2.3. Absolute Principal Component Score-Multiple Linear Regression Model

The Absolute Principal Component Score-Multiple Linear Regression model (APCS-MLR model) is based on a principal component analysis (PCA) to obtain the absolute principal component factor score (APCS) and combined with the multiple linear regression model (MLR) to calculate the contribution rate of each factor to each sample [21]. (1) The principal component analysis was carried out after the standardization of various data, and the factor score was calculated. (2) A 0-concentration sample was introduced for all wetland elements, and principal component analysis was performed after standardization to calculate the factor score. (3) The APCS value of each wetland area is the factor score of each sample minus the factor score of the 0-content sample. (4) Multiple linear regression analysis was carried out with APCS as the independent variable and the measured value of wetland area as the dependent variable. The obtained regression coefficient can convert APCS into the contribution of each influencing factor to each sample. The formula [22] is as follows:

$${\mathrm{C}}_{\mathrm{i}}={\mathrm{b}}_{\mathrm{i}0}+\sum _{\mathrm{k}-1}^{\mathrm{k}}{\mathrm{b}}_{{\mathrm{k}}_{\mathrm{i}}}\mathrm{APC}{\mathrm{S}}_{\mathrm{k}}$$

(1)

where C_i is the measured value of various types of wetland area; b_i0 is the constant term of multiple linear regression; b_ki is the regression coefficient of multiple linear regression; APCS_k is the absolute main factor score of factor k; and b_ki × APCS_k is the contribution of factor k to C_i.

4. Results and Discussion

4.1. Spatial and Temporal Variations in the Wetland Area in the Lower Yellow River

Select wetland type information for the years 2000, 2006, 2011, 2016, and 2021 based on the remote sensing and image interpretation results (Figure 3).

The main wetland types in the study area include such natural wetlands as rivers and flooding wetlands and such artificial wetlands as ponds, ditches, and reservoirs. The rivers in the area are permanent rivers, but the riverway patterns change dramatically. Flooding wetlands were distributed on both sides of the Yellow River. Ponds were scattered mainly in low-lying depressions along the riverway and on the outside of the embankment.

From 2000 to 2021, the total wetland area showed an increasing trend, growing from 88.15 km² to 115.78 km² (Figure 4). The predominant wetlands in the area were natural wetlands, accounting for approximately 80% of the total wetland area. River wetlands gradually increased since 2000, peaking in 2006, followed by a gradual reduction in area from 2006 to 2016. After 2016, there was a rapid rebound, up 171% from 2000. A 37% reduction in seasonally flooded flat areas was observed, primarily driven by river channel instability [23] and altered hydrological dynamics, which induced frequent transitions between active river courses and floodplain ecosystems.

Artificial wetlands increased gradually from 16.34 km² in 2000 to 23.43 km² in 2021, with the ditch area growing at the same pace. Through field investigation and resident interviews, it was found that ponds on the outside of the north embankment were converted to farmland, farmland on the south bank of the riverway and development land on the outside of the southern embankment were converted into artificial wetlands thanks to the construction and renovation of water diversion facilities. The artificial wetland area increased by 63%. The reservoir area remained relatively stable.

The total area of wetlands in the lower reaches of the Yellow River experienced fluctuating changes from 2000 to 2021. The evolving wetland types have implications for the sustainable development of these crucial ecosystems in the lower reaches of the Yellow River. The relationships between the different land use types are shown in

Table 2. Land use transition matrix in the study area from 2000 to 2021 (km²).

2000	2021
	Non-Wetland	River	Flooding Wetlands	Reservoirs	Ponds and Ditches	Total
Non-wetland	657.15	19.59	10.06	5.34	0	692.14
river	0	19.51	4.69	0	0	24.20
Flooding wetlands	0	22.40	16.10	0	0	38.50
Ponds and ditches	2.30	0	0	13.11	0	15.41
Reservoirs	0.11	0	0	0	4.98	5.09
Total	659.56	61.50	30.85	18.45	4.98	775.34

. The width of the connecting line indicates the degree or proportion of the relationship between two types of data, and the arrow indicates the direction of the transfer. The wetland types in the area studied changed frequently. The rivers changed the most; 36% of the area was transformed from flooding wetlands, and 32% of the area was transformed from non-wetlands, followed by 33% of the area of flooding wetlands transformed from non-wetlands, and 15% of the area transformed from river water.

4.2. Wetland Hydrological Conditions

4.2.1. Variation in Discharge Characteristics of Yellow River Flow

Since the Xiaolangdi Reservoir has been put into use, the downstream discharge of the Yellow River has significantly decreased [24,25]. The frequency of streams at low flow levels surged significantly, with flows below 1000 m³/s comprising 67.12% of the entire year in 2022. This was except for a moderate flood period, which lasted for a certain duration and had a flow rate of about 4000 m3/s following reservoir regulation during the rainy season. According to data from the Garden Mouth Hydrologic Station, from 2008 to 2022, the Yellow River water level had fallen by approximately 1–2 m under the same discharge (Figure 5), indicating a corresponding depth of riverway erosion [26]. The persistent occurrence of low flows can lead to a progressive reduction in the main river channel’s width and depth, altering the river’s morphology and regime. This evolution will introduce new characteristics, which are likely to negatively affect the presence and health of downstream wetland ecosystems.

4.2.2. The Variation Law of Rainfall and Produced Quantity

Rainfall is a critical component of water resources in riverine wetlands, directly influencing the water balance of these ecosystems. Years with higher rainfall quantities result in abundant wetland water resources, which are essential for upholding the ecological equilibrium and biodiversity of the wetlands. Conversely, overexploitation of wetland water resources can cause a decrease in water levels, thereby adversely affecting the wetland ecosystem.

Rainfall within the study area exhibited a discernible pattern of fluctuation, trending slowly upwards (Figure 6a), with an incremental rise of 17.5 mm/a. This upward trajectory reached a peak in 2021. The 5-year moving average line fluctuated closely around the mean of the multi-year dataset, and the MK test should be employed to ascertain the significance of this increasing trend. The results from the MK test (Figure 6b) revealed that the UF statistic for precipitation prior to 1982 was negative. This suggested that the precipitation levels initially exhibited a decreasing trend followed by an increasing one. The effective intersection points of UF and UB were 5; however, throughout this period, the UF statistic did not cross the critical line, indicating that the change was not pronounced. The data demonstrated a relatively stable trend of fluctuation.

The extraction of groundwater in the region has generally followed a gradual decline (Figure 6c), with the 5-year moving average line oscillating around the long-term average. The MK test results (Figure 6d) reveal that groundwater extraction exhibited a fluctuating pattern prior to 2013, shifting to a consistent decline post-2013. The UF surpassed the critical threshold in 2010 and 2011, signifying a pronounced upward trend in groundwater extraction. However, the intersections of UF and the UB in 2012, 2016, and 2020 did not see UF crossing the critical line, suggesting a muted mutation in the trend.

4.2.3. Characteristics of Groundwater Flow Field and Its Influencing Factors

Based on the groundwater level survey data in May 2022, a groundwater flow field map during the low water level period was drawn. From Figure 7, it can be observed that the shallow groundwater runoff was controlled by recharge sources and exploitation. The continuous sedimentation of the Yellow River riverbed formed a natural groundwater watershed, and the groundwater runoff flowed from the Yellow River to both sides of the riverway. In the area around Xisanli Village, Jinglonggong Township Forest Farm, and Fengqiu County Forestry Station, a small groundwater cone of depression was formed due to groundwater exploitation, covering an area of approximately 50 km². The groundwater burial depth on both sides of the riverway was mostly between 4 and 7 m, while in farmland and forest areas, it was more than 10 m.

Two observation points with distinct groundwater burial depths were selected, situated in Jinglonggong Township and Chenqiao Town, to conduct a detailed analysis of the changes in groundwater depth. The patterns observed at both locations were consistent, with the trend lines indicating a rise in groundwater depth over time (Figure 8a,c). Prior to 2013, the 5-year moving average line was positioned below the multi-year average. The MK test was employed to derive the UF and UB curves. Between 1979 and 2009, the depth of groundwater exhibited a fluctuating pattern. However, from 2009 onwards, there was a noticeable upward trend in the depth of groundwater. Out of the 26 years when the UF value was negative, 1984 and 1985 saw values dropping below the critical threshold, corresponding to a significant decrease in groundwater depth. The intersection of the UF and UB curves occurred in 2011, within the range of ±1.96, marking the year when a sharp change in the shallow groundwater level began. Following 2016, the UF value consistently surpassed the critical value, indicating a substantial increase in groundwater depth (Figure 8b,d).

4.3. Drivers of Wetland Landscape Change

4.3.1. Correlation Analysis of the Yellow River on Wetland Formation Conditions

The riverway patterns changed dramatically (Figure 3). The swing of the main channel directly had an important impact on the existence, maintenance, and development of wetlands. In 2000, the riverway was relatively straight, while in 2006, the riverway’s meander coefficient increased significantly [27]. After 2011, the downstream riverway pattern remained relatively stable, yet the upstream tributaries crisscrossed, leading to complex riverway patterns. As the plane position of the river changed, the beaches on both sides also changed. It was difficult to form a real wetland in the main river channel.

The highest water level from 2002 to 2022 showed a trend of first decreasing and then increasing (Figure 9), with the maximum annual water level variation of about 3 m. The multi-year maximum water level decreased by 0.5 to 2.0 m, while the multi-year minimum water level decreased by 2.9 to 4.2 m.

The area of rivers was positively correlated with the water level, gradually expanding since 2000. From 2006 to 2016, the area of rivers gradually shrank with a decrease in the Yellow River water level. After 2016, it rapidly rebounded, consistent with the changes in the Yellow River water level (Figure 9). However, due to the increasing erosion of the riverway each year, the probability of flooding decreased; hence, potential wetlands, such as branch gullies and low-lying depressions, lost the feeding of Yellow River water, leading to a year-by-year reduction in flooding wetlands.

4.3.2. Correlation Analysis of the Groundwater Flow Field on Wetland Patterns

A water level variation map for the study area was drawn based on shallow groundwater levels in 1980 and 2022, and wetland degradation areas were delineated according to land use maps from 1980 and 2021. The groundwater level on both sides of the riverway, thanks to the lateral seepage feeding from the Yellow River, fell by less than 2 m, with a greater decrease observed farther from the riverway. On both sides of the northern embankment and the upper reaches of the southern embankment, the water level fell by more than 5 m (Figure 10). The study revealed that, compared to 2003, the groundwater level in the lower reaches of the Yellow River decreased. Among others, areas between Garden Mouth and Jiahetan witnessed a drop of an average of 4.87 m [28], consistent with the findings of this study.

The degraded areas of artificial wetlands were mainly distributed at the northern embankment of the Yellow River, where the groundwater burial depth decreased significantly. With an average evaporation of 1939 mm in the region, far outweighing the average annual rainfall of 645.2 mm, and due to the substantial decline in groundwater level, former ponds had been converted into farmland and development land. On the other hand, newly emerging artificial wetlands were located on both the inside and outside of the upper reaches of the southern embankment and on the inside of the downstream embankment. Despite the regional decline in water levels, the development of irrigation canals diverted from the Yellow River had bred a wealth of ponds on the outside of the southern embankment, while seeping from the Yellow River into new back river depressions on the inside of the embankment.

4.4. APCS-MLR Analysis

This study employed the APCS-MLR model to explore the influencing factors of wetland landscape changes in the study area, including five periods of wetland types, Yellow River water levels, rainfall, and groundwater burial depth. Based on the principle of characteristic value greater than 1, a total of three principal components were extracted, with a cumulative variance explanation rate of 80.29% (

Table 3. The principal component analysis results of groundwater parameters.

Indicator	Component 1	Component 2	Component 3
Pond and ditches	0.94	0.10	−0.17
River	0.22	0.83	0.17
Flooding wetlands	−0.01	0.89	−0.25
Yellow River water level	−0.14	0.73	0.10
Rainfall	−0.07	0.03	0.76
Groundwater burial depth (Jinglonggong)	0.87	−0.03	−0.07
Groundwater burial depth (Chenqiao)	0.90	−0.01	−0.25
Quantity of groundwater withdrawal	0.65	0.10	−0.17
Characteristic value	3.90	3.07	1.31
Contribution %	30.86	30.78	18.65
Cumulative contribution %	30.86	61.64	80.29

).

The contribution rate of the first principal component was 30.86%, among which ponds, groundwater burial depth (Jinglonggong), groundwater burial depth (Chenqiao), and quantity of groundwater withdrawal were significantly positively correlated with component 1. The shallow groundwater burial depth in the study area was significantly influenced by the quantity of groundwater withdrawal, and the degraded areas of ponds on the northern embankment were distributed in areas where the groundwater burial depth dropped greatly. Therefore, the first principal component can indicate that the degradation of ponds in artificial wetlands was caused by human activities. The contribution rate of the second principal component was 30.78%, among which rivers, flooding wetlands, and Yellow River water levels were significantly positively correlated with component 2. Therefore, the second principal component can indicate that Yellow River water levels influenced the area of rivers and flooding wetlands. The contribution rate of the third principal component was 18.65%, among which rainfall was significantly positively correlated with component 3. Therefore, the third principal component can indicate that meteorological conditions influenced the area of the wetlands.

Based on the analysis conducted, the relative contributions of each factor influencing wetland area were quantified (Figure 11). The linear regression coefficient (R2) between the observed wetland area data and the model’s predicted values exceeded 0.75, and the p-value associated with the regression equation was below 0.05. These statistics suggested that the model had high accuracy and reliability. These results indicated a strong linear regression fit. Human activities were found to be the primary influence on ponds and ditches, accounting for 66.8% of the area change. For riverine wetlands, the influence of human activities and Yellow River runoff was more balanced, contributing 43.0% and 30.0%, respectively. Flooding wetlands, however, were significantly affected by unspecified sources, with a contribution rate of 71.1%. The Yellow River is a typical sediment-laden river in China. Due to the water–sediment regulation of Xiaolangdi Reservoir, the sediment concentration of the Yellow River has been changing. River sediment concentration contributes to the growth of flooding wetlands. Therefore, it is speculated that the main reason for the unspecified source is the influence of the sediment concentration of the Yellow River.

5. Conclusions

By the aid of remote sensing and image interpretation and traditional hydrogeologic survey, we analyzed the evolutionary characteristics of the distribution pattern of the wetlands in the Kaifeng downstream of the Yellow River and drew the following conclusions:

(1)

With natural wetland as its major wetland type, the study area saw an increase in the total wetland area from 2000–2021. For natural wetlands, the total area of rivers increased while that of flooding wetlands decreased. For artificial wetlands, the total area increased.

(2)

The erosion of the riverway in the region made flooding increasingly less likely. The main riverway, even worse, was a typical wandering stream with significant oscillations, directly impacting the area of river water and flooding wetlands. The degraded areas of artificial wetlands were mainly distributed at the northern embankment of the Yellow River, where the groundwater burial depth decreased significantly. In contrast, at the southern embankment, for the sake of the irrigation canal diverted from the Yellow River, new artificial wetlands had formed, adding to the artificial wetland area.

(3)

Three influencing factors of wetland area along the Yellow River were identified in the study area, namely human factors, Yellow River runoff, and meteorological factors. According to the APCSMLR model, the influence of meteorological conditions on wetland area was small. The area of ditches and ponds was mainly affected by human factors, the area of river water surface was affected by human factors and the runoff of the Yellow River, and flooding wetlands was greatly affected by unknown sources.

(4)

The reduction of groundwater exploitation and an adequate supply of diverted Yellow River water were conducive to the development of wetlands in the back river depressions outside the Yellow River embankment.