Analysis and Dynamic Evaluation of Eco-Environmental Quality in the Yellow River Delta from 2000 to 2020

Abstract

With the rapid development of urbanization and population growth, the ecological environment in the Yellow River Delta has undergone significant changes. In this study, Landsat satellite data and Google Earth Engine (GEE) were utilized to dynamically evaluate the changes in eco-environmental quality in the Yellow River Delta region using the remote sensing ecological index (RSEI). Additionally, the CASA model was used to estimate net primary productivity (NPP) and explore the relationship between vegetation NPP, land-use and land-cover change (LUCC), and eco-environmental quality to reveal the complexity and related factors of eco-environmental quality changes in this region. The results show that: (1) Over the past 20 years, the eco-environmental quality in the Yellow River Delta region has changed in a “V” shape. The eco-environmental quality near the Yellow River Basin is relatively better, forming a diagonal “Y” shape, while the areas with poorer eco-environmental quality are mainly distributed in the coastal edge region of the Yellow River Delta. (2) The response of vegetation NPP to eco-environmental quality in the Yellow River Delta region is unstable. (3) Urban construction land in the Yellow River Delta region is strongly correlated with RSEI, and the absolute value of the dynamic degree of land use is as high as 8.78%, with significant land transfer changes. The correlation between arable land and RSEI is weak, while coastal mudflats are negatively correlated with RSEI, with the minimum absolute value of the dynamic degree of land use being −1.01%, and significant land transfer changes. There is no correlation between forest land and RSEI. Our research results can provide data support for the eco-environmental protection and sustainable development of the Yellow River Delta region and help local governments to take corresponding measures.

1. Introduction

Ecological environment is the foundation of human survival and development, as well as the fundamental strategy for sustainable development. Since the founding of the People’s Republic of China, rapid urbanization and development have promoted the high-speed economic growth of the country, but have also brought a series of ecological environment problems, which severely constrain the sustainable development of regions [1]. Therefore, to study the patterns of ecological environment changes, dynamic and continuous monitoring of the spatiotemporal variations in ecological environment quality has become a focus of academic research and relevant government departments, which is of great significance for regional sustainable development.

In recent years, with the development of technology, various remote sensing monitoring indices such as NDVI (normalized vegetation index), FVC (fractional vegetation cover), wetlands, etc., have been widely used for monitoring and evaluating urban, grassland, forest, wetland, and other ecosystems. Researchers such as Xu and Hu [2,3] have analyzed and evaluated the ecological environment changes in urban and its characteristic ecological areas based on remote sensing ecological indices. Zhao, Shang, and Ma [4,5,6] have evaluated the ecological environment quality and its ecological security changes in grassland and natural landscapes through FVC and NDVI. Zhang and Hu [7,8] have explored the temporal and spatial changes in natural reserves and the response to environmental climate factors using NDVI. Zhang et al. [9] have established wetland water extraction rules based on water indices and vegetation indices and have implemented wetland management measures. These studies provide new ideas for regional ecological environment quality monitoring and have characteristics of real-time monitoring, dynamic changes, and objective evaluation, which are important for achieving regional ecological environment monitoring and governance and regional sustainable development and have become research hotspots in fields such as urban ecology and landscape ecology. However, for complex ecosystems, a single remote sensing monitoring index is difficult to effectively and comprehensively reflect its ecological environment quality [10], and a system that integrates multiple monitoring indices is needed for an objective and comprehensive evaluation.

Over the years, scholars have proposed various methods for comprehensively evaluating the quality of regional ecological environments. Among them, Xu [11] proposed RSEI (remote sensing ecological index), a comprehensive evaluation index based on remote sensing technology that reflects multiple ecological environmental factors. RSEI is capable of fully reflecting the quality of regional ecological environments and has been widely used. For example, Wen et al. [12] conducted an ecological evaluation of the southern suburbs of Beijing based on the remote sensing ecological index. Gou et al. [13] combined the RSEI method with the random forest algorithm to monitor the ecological environment quality of Beijing, China. Wang et al. [14] analyzed the temporal and spatial changes of the ecological redline area in Nanjing, China, using RSEI. Wang et al. [15] analyzed and evaluated the temporal and spatial changes of ecological environment quality in Yanhe River Basin using RSEI. Yueriguli et al. [16] monitored the ecological quality changes of the land reclamation area using RSEI. These studies indicate that as an index mainly relying on remote sensing information, RSEI can carry out timely, objective, and dynamic monitoring and evaluation of the ecological environment quality at different scales, and also has high practical significance and reference value [17,18,19]. However, the regional-scale ecological environment quality evaluation based on RSEI requires extensive preprocessing of remote sensing image data. To address this issue, the Google Earth Engine (GEE) cloud platform developed by Google can be utilized. It provides access to large-scale, long-term time series of remote sensing data and allows users to interactively develop and test algorithms, thereby improving the efficiency of remote sensing image processing [20,21,22,23].

The Yellow River Delta is one of the most important river deltas in China and a core component of the economic development in the northern region of Shandong Province. This area boasts a unique geographical location, abundant natural resources, and immense development potential, making it a subject of domestic and international attention. However, the development and construction process in this area has given rise to many ecological and environmental problems, such as the urban heat island effect [24,25,26], soil pollution [27], and increased disaster risks [28,29,30], which are at odds with the core principles of ecological protection and high-quality, eco-friendly development. Therefore, in order to promote ecological protection and high-quality development in the Yellow River Basin and the delta region, it is essential to accurately, timely, and comprehensively understand the spatiotemporal distribution and evolution trend of the ecological environment quality in the Yellow River Basin. Recently, there has been a growing number of studies conducted by scholars on the Yellow River Delta region. For instance, Su et al. [31] utilized GEE and GIS to analyze multi-scale time series of the Yellow River Delta area. Xu et al. [32] combined RS and GIS to analyze the temporal and spatial characteristics of major artificial objects in the Yellow River Delta. Jiao et al. [33] analyzed the evolution of wetlands in the Yellow River Delta using remote sensing. However, previous studies [34,35,36,37,38] that compared the ecological environment time series only normalized the ecological environment conditions of each year in the research area and ignored the differences and trends in the overall ecological environment quality between different years, resulting in poor comparability of the normalized interannual ecological environment indices. Additionally, few scholars have included land-use and land-cover change (LUCC) and net primary production (NPP) in their analysis of the impact of the Yellow River Delta’s ecological environment.

To fill this gap and further analyze the change processes and drivers of ecological and environmental quality in the Yellow River Delta, this manuscript used Landsat5 (TM) and Landsat8 (OLLI) data as remote sensing data sources to evaluate the ecological and environmental quality of the study area dynamically for 20 years using GEE and RSEI. Additionally, the 20-year vegetation NPP was inverted using the MOD13Q1 dataset and the CASA model, and the correlation between NPP and land use and ecological environment quality was explored using linear fitting method to provide scientific basis for ecological environment planning, ecological environment management, and ecological environment protection and development in the Yellow River Delta region.

Our main scientific contributions are summarized as follows. First, the study focuses on the Yellow River Delta, a region with unique environmental conditions and land use patterns, adding new perspectives and insights to research in this area and providing an important contribution to understanding the relationship between NPP, LUCC, and RSEI in the region. Second, advanced methods of spatial and statistical analysis are used. Through the integrated use of remote sensing data, land use maps, and statistical methods, patterns and trends are revealed that may have been over-looked by traditional methods of analysis. Third, we provide an in-depth analysis of the impacts of land use change on NPP and RSEI, which provides important information for decision making on sustainable land management and environmental protection.

2. Study Area and Data Sources

2.1. Study Area

Yellow River Delta (38°15′23″ N–37°52′21″ N, 118°45′57″ E–119°35′46″ E) is located in the core of Blue and Yellow Economic Belt of Shandong Province, mainly including Dongying District, Hekou District, Kenli District, and Lijin County, with an area of about 26,500 square kilometers, which is an important economic development area in China. With an average annual temperature of 12.9 °C and an average annual precipitation of 551.6 mm, the modern Yellow River Delta has a warm temperate continental monsoon climate. It is the largest, newest, and most comprehensive wetland ecological function reserve in China’s warm temperate region. Figure 1 depicts the study area’s scope.

2.2. Data Source and Processing

• Landsat5 (TM) and Landsat8 (OLI) image data from 2000 to 2020 were used as remote sensing images in this study, and LUCC and NPP data were primarily employed as impact factor data.
• The LUCC data were published by Xin Huang in Wuhan University. Based on 335,709 scenes of Landsat data on Google Earth Engine, Xin Huang constructed spatio-temporal features of land use, combined with a random forest classifier to obtain classification results of land use, and proposed a post-processing method that includes spatio-temporal filtering and logical inference to further improve the spatio-temporal consistency of the annual China Land Cover Dataset (CLCD). The CLCD contained year-by-year land cover information for China from 1985 + 1990–2020. The overall accuracy of the CLCD was 79.31% based on 5463 visually interpreted samples.
• The NPP data were obtained using the CASA model inversion. The relevant data (MOD13Q1 image, MOD11A2 image, monthly solar radiation data, monthly mean temperature data, and monthly precipitation data) used for calculations in this study were provided by NASA and the China Meteorological Science Data Network Shared Services Network.

The data time period was chosen from June to August to avoid the impact of seasonal variations and water bodies on the data processing results. The annual permanent water body data were used to flood and remove the study area’s water bodies to completely remove their impact on RSEI calculations. The GEE cloud platform was used for all picture pre-processing operations, including radiometric correction, geometric correction, algorithmic de-clouding, and water body masking. Using the principle component analysis (PCA) method and the Landsat image dataset hosted on the GEE cloud platform, we were able to map the spatio-temporal distribution of the region using the RSEI data of the Yellow River Delta from 2000 to 2020. Based on these data, we analyzed the dynamic evolution of eco-environmental quality in the Yellow River Delta over 20 years and analyzed the variability associated with the eco-environmental quality of the Yellow River Delta using NPP data generated by LUCC and CASA models. The detailed workflow of this study is shown in Figure 2, and the data sources are shown in

Table 1. Research data sources.

Data Set	Data Source	Resolution	Data Description
Landsat5 (TM)	Google Earth Engine	30
Landsat8 (OLI)	Google Earth Engine	30
China Land Cover Dataset (CLCD)	Xin Huang of Wuhan University	30	The overall accuracy rate for CLCD was 79.31%
MOD13Q1	Google Earth Engine	500
MOD11A2	Google Earth Engine	500
Monthly solar radiation data,	NASA	500
monthly mean temperature data	China Meteorological Science Data Network Shared Services Network	500
monthly precipitation data	China Meteorological Science Data Network Shared Services Network	500

.

3. Research Indicators and Methods

3.1. RSEI Model

RSEI integrates four ecological evaluation indicators of greenness, wetness, heat, and dryness by using the principal component analysis method.

3.1.1. Greenness Index

The normalized difference vegetation index (NDVI) is selected to characterize the greenness index of the study area [39]. The calculation formula is as follows:

$$NDVI=\frac{\left(p_{nir}-p_{red}\right)}{\left(p_{nir}+p_{red}\right)}$$

(1)

where $p_{nir}$ and $p_{red}$ represent the reflectance of the near-infrared band and the red band of the Landsat5 (TM) and Landsat8 (OLI) satellite images, respectively.

3.1.2. Wetness Index

The wetness index is characterized by a wetness component which is denoted by WET based on the tassel cap transformation [40]. The wetness component is extracted based on Landsat5 (TM) and Landsat8 (OLI) data [41]. The formula for calculating the wetness component is as follows:

$$Wet\left(TM\right)=0.0315 b1+0.02 b2+0.3102 b3+0.1594 b4-0.6803 b5-0.6109 b7$$

(2)

$$Wet\left(OLI\right)=0.1511 b1+0.1973 b2+0.3283 b3+0.3407 b4-0.7117 b5-0.4559 b7$$

(3)

where b1, b2, b3, b4, b5, and b7, respectively, represent the reflectance of bands 1, 2, 3, 4, 5, and 7 of Landsat5 (TM) image and the reflectance of bands 2, 3, 4, 5, 6, and 7 of Landsat8 (OLI) image; the wetness components of the Landsat5 (TM) and Landsat8 (OLI) images are, respectively, represented by Wet (TM) and Wet (OLI).

Table 2. Landsat5 (TM) and Landsat8 (OLI) corresponding wavebands.

Symbol	Bands
	Landsat5 (TM)	Landsat8 (OLI)
$b1$	Band1	Band2
$b2$	Band2	Band3
$b3$	Band3	Band4
$b4$	Band4	Band5
$b5$	Band5	Band6
$b7$	Band7	Band7

displays the formula’s parameters.

3.1.3. Heat Index

The land surface temperature (LST) is chosen to characterize the heat index. Heat data is a significant parameter at the interface between the Earth’s surface and the atmosphere, as well as an important parameter for investigating the exchange of materials and energy between the surface and the atmosphere. The LST products in Aqua MOD11A2 with a period of 8 days and a spatial resolution of 1 km are where the heat data for this investigation were derived. The built-in function of GEE was used to resample the LST data for 30 m.

3.1.4. Dryness Index

The normalized difference built-up and soil index (NDBSI) which is synthesized using the index-based built-up index (IBI) and the bare soil index (SI) is selected to characterize the dryness index in this manuscript. The formula is as follows:

$$SI=\frac{\left(p_{SWIR1}+p_{red}\right)-\left(p_{blue}+p_{NIR}\right)}{\left(p_{SWIR1}+p_{red}\right)+\left(p_{blue}+p_{NIR}\right)}$$

(4)

$$IBI=\frac{\frac{2p_{SWIR2}}{p_{SWIR1}+p_{NIR}}-\left[\frac{p_{NIR}}{p_{red}+p_{NIR}}+\frac{p_{green}}{p_{SWIR1}+p_{green}}\right]}{\frac{2p_{SWIR2}}{\left(p_{SWIR1}+p_{NIR}\right)}+\left[\frac{p_{NIR}}{p_{red}+p_{NIR}}+\frac{p_{green}}{p_{SWIR1}+p_{green}}\right]}$$

(5)

$$NDBSI=\frac{SI+IBI}{2}$$

(6)

where $p_{blue }{p}_{green }{p}_{red }{p}_{NIR }{p}_{SWIR1 }{p}_{SWIR2}$ are, respectively, the surface reflectance of the blue band, green band, red band, near-infrared band, shortwave infrared band 1, and shortwave infrared band 2 corresponding to Landsat5 (TM) and Landsat8 (OLI) images.

3.1.5. Normalization of Ecological Indicators

Since the NDVI, NDBSI, and WET calculated by the formula are of different dimensions, they need to be normalized before subsequent analysis to normalize the data of different dimensions to the [0, 1] interval and transform them into dimensionless indicators. The normalization formula is as follows:

$$NIi=\frac{Indicatori-Indicatormin}{Indicatormax-Indicatormin}$$

(7)

where $NIi$ is the value of an indicator after regularization; $dicatori$ is the value of the indicator in image element $i$; $Indicatormax$ is the $maximum$ value of the indicator; and $Indicatormin$ is the $minimum$ value of the indicator. In order to make the results of different years uniform and comparable, the maximum and minimum values of this indicator in the whole research period are taken respectively.

3.1.6. Determination of the Weight of Each Ecological Index Factor

RSEI is composed of four factors coupled NDVI, NDBSI, LST, and WET, and the contribution and relative importance of each factor in RSEI is expressed by its weight. In order to eliminate the influence of human factors in the determination of the weight of each factor, we perform multi-indicator integration based on principal component analysis (PCA) to determine the contribution of each factor automatically and objectively.

3.2. Inversion Estimation of NPP in the Yellow River Delta by CASA Model

NPP refers to the amount of energy that is still available for plant growth and reproduction after deducting the part consumed by plant respiration from the fixed energy of plant photosynthesis in the primary production process. NPP reflects the carbon fixation capacity of terrestrial vegetation and is a key parameter for characterizing terrestrial ecological processes [42]. By studying the relationship between vegetation NPP and its ecological quality in the Yellow River Delta, we can further explore the relationship between the carbon fixation capacity and the ecological environment quality in the Yellow River Delta region.

In this manuscript, we estimated NPP in the Yellow River Delta based on CASA model [43,44,45], which fully considered the physiological and ecological characteristics of the vegetation itself and the environmental conditions for growth, and integrates remote sensing data, meteorological data, and vegetation cover type data to achieve vegetation NPP estimation. The vegetation NPP in the CASA model was calculated from the photosynthetically active radiation absorbed by vegetation and the actual light energy use [43], and the equation is shown as follows:

$$NPP(x,t)=APAR\left(x,t\right)\ast \epsilon \left(x,t\right)$$

(8)

$$APAR\left(x,t\right)=SOL\left(x,t\right)\ast fAPAR\left(x,t\right)\ast 0.5$$

(9)

where $APAR\left(x,t\right)$ is the photosynthetically active radiation absorbed by vegetation at the location of pixel $x$ in month $t$, and the unit is $MJ/\left(m²·month\right)$; $\epsilon \left(x,t\right)$ is the actual light energy utilization at the location of pixel $x$ in month $t$, and the unit is $\frac{gC}{MJ};$ $SOL\left(x,t\right)$ is the total solar radiation at the location of pixel $x$ in month $t$, and the unit is $MJ/\left(m²·month\right)$; $fAPAR\left(x,t\right)$ is the proportion of incident photosynthetically active radiation absorbed by vegetation at the location of pixel $x$ in month $t$, and it is dimensionless; $0.5$ indicates the proportion of solar effective radiation ($\mathrm{wavelength} \mathrm{of} 0.38~0.71 \mathrm{mm}$) utilized by vegetation to the total solar radiation. $fAPAR\left(x,t\right)$ is determined by vegetation type and vegetation cover, and studies have shown that there is a good linear relationship between $fAPAR\left(x,t\right)$ and NDVI values, as shown in the following equation:

$$fAPAR\left(x,t\right)=\min \left[\frac{SR-S{R}_{min}}{S{R}_{max}-S{R}_{min}},0.95\right]$$

(10)

$$SR=\frac{1+NDVI}{1-NDVI}$$

(11)

where $SR$ is the ratio vegetation index, $S{R}_{min}$ is 1.08 according to experience, and $S{R}_{max}$ is 4.14~6.17 depending on the vegetation type.

3.3. Land-Use Dynamic Degree

Land-use dynamic degree refers to the quantity change of land use types in a certain period, which mainly reflects the regional differences of the intensity and rate of land use change. In this manuscript, land-use dynamic degree is used to calculate the change range of a single land type in the Yellow River Delta region. The formula is as follows:

$$K=\frac{U_a-U_b}{U_a}\ast \frac{1}{T}\ast 100\%$$

(12)

where $K$ is the dynamic degree of a certain LUCC type during the study period, $U_a$ and $U_b$ are the area of a certain LUCC type at the beginning and end of the study, respectively, and $T$ is the study time.

4. Results and Analysis

4.1. Indicator Analysis

Many scholars believe that the greater the value of NDVI and WET, the greater the vegetation coverage and surface humidity, and the higher the eco-environmental quality. In contrast, the larger the values of NDBSI and LST, the more severe the hardening degree of soil and building land, the higher the surface temperature, and the worse the eco-environmental quality. Through normalization, the advantages of PC1 are highlighted, which can integrate most of the characteristic information of the four indicators and is consistent with the reality. Therefore, a comprehensive ecological index can be established to analyze the spatio-temporal evolution of eco-environmental quality in the Yellow River Delta region.

Specifically, the index is obtained by linearly combining four individual ecological indicators over five periods, where the weight of each individual indicator is determined by the results of the PCA analysis. As PCA analysis can combine several indicators with high correlation into a new indicator, the remote sensing ecological index can well reflect the quality and change trend of the regional ecological environment. In this paper, the first principal component (PC1) of the five periods was obtained through PCA analysis, as shown in

Table 3. Analysis of components of RSEI.

Year	Mean RSEI	WET	NDVI	NDBSI	LST	PC1 (%)
2000	0.63	0.61	0.47	−0.67	−0.61	70.32
2005	0.47	0.65	0.45	−0.60	−0.53	68.79
2010	0.42	0.73	0.56	−0.49	−0.64	66.69
2015	0.52	0.60	0.48	−0.55	−0.49	74.64
2020	0.51	0.71	0.55	−0.53	−0.54	80.85
mean	0.51	0.66	0.50	−0.57	−0.56	70.32

. The contribution rate of the first principal component (PC1) of the five periods was higher than 66.69%, with the contribution rates of 2000, 2015, and 2020 all exceeding 70%, indicating that PC1 integrated most of the characteristics of the four individual ecological indicators of the five periods. It can well reflect the changes in regional ecological and environmental quality. Meanwhile, by analyzing the explanatory variables of PC1, we can learn that the wetness and greenness indicators are positive in all five periods, while the dryness and heat indicators are negative, indicating that the wetness and greenness indicators have a positive impact on the ecological environment quality, while the dryness and heat indicators have a negative impact on the eco-environmental quality, which is consistent with the actual situation [33]. Therefore, the remote sensing ecological index can provide a more comprehensive assessment of the regional ecological quality and the degree of influence of each ecological index on the ecological environment.

4.2. Variation Characteristics of RSEI in the Yellow River Delta

4.2.1. Analysis of Spatio-Temporal Variation of RSEI in the Yellow River Delta Region

The study area included three districts and one county. In order to better express the spatio-temporal distribution characteristics of eco-environmental quality in the study area, based on the Technical Specification for Assessment of Eco-environmental Status (HJ/T192-2006) issued in 2015, the RSEI values of 2000, 2005, 2010, 2015, and 2020 were divided into five intervals with an interval of 0.2: 0~0.2, 0.2~0.4, 0.4~0.6, 0.6~0.8, and 0.8~1.0, corresponding to five grades of poor, fair, moderate, good, and excellent, respectively. The percentage and distribution of RSEI grades in the Yellow River Delta in each year are shown in Figure 3 and Figure 4.

According to Figure 4, the regions with excellent and good eco-environmental quality are mainly distributed in the vicinity of the Yellow River Basin, and the distribution shape resembles an inclined “Y”. The regions with poor and fair eco-environmental quality are located in the coastal areas of the Yellow River Delta. The eco-environmental quality gradually deteriorates from the center of the Yellow River Delta to the periphery, and in the urban areas, the eco-environmental quality is primarily distributed in two levels: poor and fair. According to the combination of Figure 3 and Figure 4, the eco-environmental quality in 2000 was at its worst, with 45.71% and 18.02% of the area with poor and fair eco-environmental quality, respectively. In 2010, the eco-environmental quality was the best, with 9.57% and 44.62% of the area with excellent and good eco-environmental quality, respectively. From 2000 to 2010, the whole ecology of the Yellow River Delta gradually improved. Compared with 2000, the area with poor and fair ecological quality in 2010 decreased by 67.64% and 69.99%, respectively, indicating a considerable improvement in ecological quality. From 2010 to 2020, there had been a decline in the eco-environmental quality. Compared with 2010, the areas of excellent and good eco-environmental quality in 2020 decreased by 10.03% and 47.51%, respectively, showing a significant decline, which was related to the rapid economic development and urban construction expansion in the Yellow River Delta in the past decade. The rapid economic development had promoted the regional urbanization and the area of forest vegetation decreases, which made the quality of ecological environment in some regions decrease.

4.2.2. Analysis of the Interannual Variation Trend of RSEI in the Yellow River Delta Region

Table 4. Percentage of RSEI grades in different districts and counties in different years.

Year	RSEI Grade	Study Area
		Dong Ying	He Kou	Ken Li	Li Jin
2000	Excellent	13.55%	2.75%	4.84%	0.81%
	Good	18.42%	7.28%	10.14%	3.34%
	Moderate	36.24%	19.13%	24.99%	17.79%
	Fair	29.07%	50.06%	41.76%	55.08%
	Poor	2.63%	20.77%	18.24%	22.97%
2005	Excellent	2.80%	5.92%	15.08%	9.89%
	Good	17.62%	26.63%	32.54%	43.96%
	Moderate	30.50%	34.20%	27.20%	28.78%
	Fair	42.31%	25.02%	20.27%	13.26%
	Poor	6.77%	8.23%	4.86%	4.11%
2010	Excellent	1.99%	7.09%	11.13%	16.17%
	Good	14.78%	45.34%	45.57%	55.47%
	Moderate	28.55%	29.91%	24.34%	18.24%
	Fair	37.58%	12.83%	14.04%	8.11%
	Poor	17.10%	4.83%	4.86%	2.01%
2015	Excellent	1.02%	3.83%	12.00%	2.61%
	Good	14.45%	17.57%	23.55%	26.47%
	Moderate	28.13%	42.09%	32.64%	43.11%
	Fair	40.50%	30.83%	24.20%	21.09%
	Poor	15.89%	5.64%	7.41%	6.71%
2020	Excellent	1.42%	3.56%	15.95%	5.82%
	Good	13.44%	22.29%	26.01%	23.18%
	Moderate	32.17%	33.85%	28.68%	33.30%
	Fair	37.84%	27.63%	23.13%	26.74%
	Poor	15.14%	12.67%	6.10%	10.95%

and Figure 4 and Figure 5 show that the eco-environmental quality in Dongying District was better in 2000 compared to the other three districts and counties, with a high percentage of excellent and good eco-environmental quality grades, 13.55% and 18.42%, respectively. In 2005, the eco-environmental quality of Dongying District gradually decreased by about 36.13%, while that of the other three districts and counties increased significantly, with the eco-environmental quality of Hekou District increasing by 69.19%, Kenli District increasing by 68.54%, and Lijin County increasing by 92.29%. From 2000 to 2005, the eco-environmental quality of Kenli District, Hekou District, and Lijin County in the Yellow River Delta showed a trend of improvement, while Hekou District and the northern part of Lijin County and Dongying District showed a trend of getting worse and a scattered trend of staying the same. In 2010, the eco-environmental quality of Dongying District continued to decline by about 17.87%, while that of the other three districts gradually became better, with a small-scale improvement in eco-environmental quality. From 2005 to 2010, the eco-environmental quality in the central and southern part of He Kou District, the northern part of Lijin County, and the southeastern part of Kenli District all showed a trend of improvement, while the whole area of Dongying District, the southern part of Lijin County, the northern part of Kenli District, and the southwestern part of He Kou District mainly showed a staggered distribution of getting worse and staying the same. In 2015, the overall eco-environmental quality of the Yellow River Delta deteriorated, with a 7.75% decrease in Dongying District, 59.18% decrease in Hekou District, 37.3% decrease in Kenli District, and 59.4% decrease in Lijin County, and the areas with good eco-environmental quality were mainly concentrated in the northeastern and northern parts of the Yellow River Delta. From 2010 to 2015, the eco-environmental quality in most areas of the four districts and counties showed a deteriorating trend, and only the northern and southern part of Lijin County, the southwestern part of Dongying District, and the eastern part of Kenli District showed a better trend, covering an area of about 749.76 km². In 2020, the overall eco-environmental quality in the Yellow River Delta showed little change compared with that of 2015, and the distribution of ecological environment quality was similar to that of 2015, with excellent and good grades appearing only in the northeast of Kenli District and the southeast of Hekou District, with an increase of 15.28%. From 2015 to 2020, the central part of the delta showed a worsening trend, with an area of about 2137.76 km², and only the Dongying District, the southern of the Kenli District, and the central part of the Hekou District showed a tendency of greater ecological quality, with an area of about 491.26 km².

In conclusion, the eco-environmental quality in the Yellow River Delta showed a V-shape change in the past 20 years. From 2000 to 2010, the overall eco-environmental quality gradually became better, among which the eco-environmental quality of Dongying District gradually became worse, and the eco-environmental quality of Hekou District, Kenli District, and Lijin County gradually improved. From 2010 to 2020, the eco-environmental quality gradually deteriorated, with a large area of the eastern, central, and western parts of the delta showing a deterioration of the ecological environment, and the rest of the region showing a staggered deterioration and improvement. From 2010 to 2020, the overall eco-environmental quality of the Yellow River Delta had shown a tendency of improvement. This changing trend was related to the active implementation of policies such as “Regulations of the People’s Republic of China on Nature Reserves, Wetland Protection Measures in Shandong Province, Ecological Protection and High-Quality Development of the Yellow River Delta by the local government in the past 20 years, which had effectively controlled its ecological environment and effectively improved the eco-environmental quality of the Yellow River Delta.

4.3. Characteristics of NPP and LUCC Change in the Yellow River Delta

4.3.1. Interannual Spatio-Temporal Variation of NPP in the Yellow River Delta

In this study, CASA model was used to invert the 20-year vegetation NPP in the Yellow River Delta. The mean distribution of inverted NPP in different years was presented in

Table 5. The mean value of NPP in different years in the Yellow River Delta.

Year	Mean NPP (g C/m2)
2000	136.97
2005	182.43
2010	170.15
2015	186.77
2020	199.19

, and the spatio-temporal distribution of NPP was displayed in Figure 6.

Combining Table 5 and Figure 6, it can be seen that the vegetation NPP in the Yellow River Delta was extremely uneven spatially, showing an increasing trend from sea to land. The vegetation NPP in the Yellow River Delta decreased from the riverbank to the periphery, and the average value of vegetation NPP was 175.23 g C/(m².a), and the vegetation NPP was generally lower in land areas closer to the sea. From 2000 to 2020, the vegetation NPP in the Yellow River Delta showed an overall tendency of “decreasing and then increasing”. It can be divided into two time regions according to this, which were 2000–2010 and 2010–2020. Compared with other years, the vegetation growth was better in the Yellow River Delta river area in 2000, and the NPP was the largest and most widely distributed in this region. In 2005, the vegetation NPP near the Yellow River Channel gradually declined and migrated to the northeast as a whole. In 2010, vegetation NPP decreased significantly in the central part of the estuary and the northern part of Kenli District, and in the western part of the Yellow River Delta, vegetation NPP gradually migrated from the western part of Lijin County to the western part of Hekou District, and 52.1% of all areas showed a decrease in vegetation NPP values compared with 2000. During the period of 2010–2020, the vegetation NPP gradually rebounded in 2015 in the southwestern part of Hekou District and the western part of Lijin County, and also showed an increasing trend in vegetation NPP near the Yellow River channel compared to 2010. In 2020, the vegetation NPP in the Yellow River Delta continued to increase on the whole, and the mean value of NPP was the largest in that year compared to 2010, at 199.18 g C/(m².a), but near the estuary of the Yellow River, the vegetation NPP in the area of 95.36 km² decreased significantly.

In summation, the CASA model’s inversion of the 20-year vegetation NPP in the Yellow River Delta produced the desired results. According to the results, the vegetation NPP in the Yellow River Delta generally showed a trend of “decreasing before growing,” and it also displayed a decreasing pattern from the Yellow River channel area to the coastal area, which was consistent with previous studies [46].

4.3.2. Spatio-Temporal Dynamic Changes of LUCC in the Yellow River Delta

Land use/cover change (LUCC) analysis is crucial to ensuring the security of the ecological environment and enhancing eco-environmental quality because it has a significant impact on both human production activities and the sustainable development of the ecological environment [47].

In this manuscript, the annual China Land Cover Dataset (CLCD) was selected for the Yellow River Delta land use data, which mainly contained nine main land types, including arable land, forest land, shrub land, grassland, water surface, glade, bare land, urban construction land, wetland, and so on. Within this dataset, from 2000 to 2020, the land types in the Yellow River Delta mainly included arable land, forest land, urban construction land, coastal mudflats, and water bodies. In order to further understand the 20-year land use change and the range of land change in the Yellow River Delta, a 20-year land transfer matrix was created, as shown in Figure 7, and land-use dynamic degree was calculated according to Formula (12), which was used to represent the quantitative change of LUCC types within a certain period.

(Note: UBL represents urban and built-up lands, WB represents water body, FL represents forest land, CTF represents coastal tidal flat, Sea represents ocean, and CL represents Cropland).

As illustrated in Figure 7, in the Yellow River Delta, arable land was primarily distributed in the middle of the delta, with no significant land changes and just a minor amount of arable land being shifted to water bodies. Forest land was mainly distributed in the eastern part of the inlet, covering very little area. Coastal mudflats were distributed in the delta area on the sea side, with significant land changes and a primary transfer to town construction land. The urban construction land was mainly distributed in the northwestern, southeastern, and northern regions, and the land changes were not obvious. In the Yellow River Delta, the arable land area decreased by 681.18 km² in the past 20 years, and the LUCC dynamic degree was −1.01%, indicating that the arable land showed a slow decreasing trend in the past 20 years, and the change of land transfer was more significant. The area of forest land decreased by 65.33 km², and the LUCC dynamic degree was −4.51%, with the largest fluctuation of LUCC change, indicating that forest land shrank severely in the past 20 years, with a sharp decline and obvious land transfer change.

The area of coastal mudflats decreased by 882.71 km², with a 20-year land area transfer rate of 83.82% and a LUCC dynamic degree of −4.19%, and the land use change was second only to forest land. The change of urban construction land was the most obvious, with the transfer rate that reached 744.65% from 2015 to 2020. The rapid expansion of cities and towns led to an increase of 859.31 km² in the area of urban construction land, and the dynamic degree of land use was 8.78%, indicating that the cities and towns in the Yellow River Delta had been showing a trend of expansion in the past 20 years, and the expansion rate had gradually increased. The shifting of land had brought complex impacts on both human society and the natural environment, and there was an urgent need to find a balance between land use change and urban development.

4.4. The Relationship between NPP, LUCC and RSEI in the Yellow River Delta

4.4.1. Relationship between NPP and RSEI in the Yellow River Delta

As an important indicator of terrestrial ecological environment, RSEI is closely related to vegetation NPP and land use. As can be seen from

Table 6. Distribution of the correlation between RSEI and NPP in the Yellow River Delta.

Year	Correlation
2000	0.122
2005	0
2010	0
2015	−0.395
2020	−0.397

, the fitting directions of NPP and RSEI in each year were not consistent. The fitting results fluctuated greatly from 2000 to 2020, with the correlation coefficients being 0.12, 0, 0, −0.395, and −0.397 (p < 0.01). This indicated that there was no stable correlation between eco-environmental quality and NPP in the Yellow River Delta region, and the response of eco-environmental quality to NPP was complex. Additionally, it also indicated that there was no stable correlation between carbon fixation capacity and eco-environmental quality in the Yellow River Delta region. In other words, the change of carbon fixation capacity in the Yellow River Delta region did not have irregular changes in the impact of eco-environmental quality in the region.

4.4.2. The Relationship between LUCC and RSEI in the Yellow River Delta

The level of LUCC in the Yellow River Delta has a significant impact on the ecological environment. We used the spearman correlation coefficient approach to fit the RSEI and LUCC change to explore the relationship between LUCC and RSEI in the Yellow River Delta. The fitting results are shown in

Table 7. Correlation analysis between RSEI and LUCC.

Land Use/Cover Types	Spearman Correlation Coefficient
arable land	0.34
forest land	0.08
coastal mudflats	−0.61
urban construction land	0.79

. Table 7 shows that arable land and urban construction land had a positive correlation with eco-environmental quality (R > 0, p < 0.01) in the Yellow River Delta, and the change of the proportion of urban construction land and arable land led to a decrease in eco-environmental quality, which was consistent with the research of many scholars [46,48]. However, there was a weak correlation between arable land and eco-environmental quality (R > 0, p < 0.01), which may be related to the reduced contribution of arable land to eco-environmental quality due to the proximity to the sea in the study area. Within the Yellow River Delta region, forest land and eco-environmental quality did not show correlation, and there was a negative correlation between coastal mudflats and eco-environmental quality (R < 0, p < 0.01), indicating that forest land had less impact on eco-environmental quality in the Yellow River Delta region, while the change of coastal mudflats had a negative impact on eco-environmental quality.

5. Discussion

LUCC change land use change is not only closely related to human productive activities, but also profoundly affects the sustainable development of ecological environment. Our results show that land use change in the Yellow River Delta in the past 20 years will indeed affect the eco-environmental quality, which is basically consistent with previous research trends. For example, Li and Zhao [49] found that the area of natural grazing land, marshland, and bare rocky gravel land decreased, and the area of marshland, river surface, sandy land, and glacier and permanent snow increased by 14.26% and the area of poor and poor grade decreased by 27.9% in the source area of the Party River. Cui and Zhang [50] found that the conversion of dry land into grassland, forest land, and other ecological land is the main factor for the improvement of ecological environment quality using RSEI Qinling area for nearly 40 years. Wang and Tang’s study on Nanchang City area found that conversion of cropland to watershed was the main reason for ecological environment improvement in the study area, and conversion of cropland to construction land was the main reason for ecological environment degradation in the area [51]. However, the correlation between forest land and ecological environment obtained in this manuscript was different from the results of previous studies [47], mainly because the CLCD data from Wuhan University used in this manuscript, whose precision was only 79.31%, and this dataset had a smaller proportion of forest land area in the Yellow River Delta region, with 74.03 km² in 2000, resulting in a land-use dynamic degree of −4.51% and a decrease in forest land area of 65.33 km², and also no obvious response with RSEI.

The production and material cycle of vegetation is one of the important contents of vegetation ecology research. In this manuscript, CASA model was used to estimate the vegetation NPP in the Yellow River Delta over the past 20 years, and the multi-year average value was 175.23 g C/ (m².a) was obtained, which was consistent with the results estimated by others. However, the estimated results of this manuscript were slightly higher than those of MOD17A3 product data, because the latter underestimated the productivity of shrubbery and grassland in the Yellow River Delta [52]. Meanwhile, due to the different study areas, data sources, and parameter processing strategies, the estimated values were slightly lower than the results of Chi et al. [53]. However, according to the existing research results, the vegetation NPP estimation of the Yellow River Delta in this manuscript was still within a reasonable range, and the conclusions obtained were of important reference value for the study of the structure and function of the Yellow River Delta ecosystem, as well as regional ecological environment governance and management.

The quality of the ecosystem is influenced by a variety of natural factors and human activities, of which cloudiness is also an important natural factor. The area studied in this paper is close to the sea and has high cloudiness, which may lead to some impact on the accuracy of cloud-free images synthesized using the cloud removal algorithm for a specific time period. This is because clouds can obscure the ground and affect the acquisition and processing of remote sensing images, thus affecting the accuracy of ecological quality assessment. In contrast, deltas in the lower latitude belts, such as the Ganges Delta [54] and the Pearl River Delta [55], would be more likely to obtain high-quality remote sensing data than the study area, as they are located in drier climatic zones. In addition, the degree of urbanization and level of human activity in the study area will also affect the accuracy of the ecological quality assessment. For example, more urbanized areas may face more pollution and land use pressures, and these factors may have a significant impact on the quality of the ecological environment. Therefore, the study area in this paper has certain limitations and difficulties, and the impact of these factors on the findings needs to be fully considered in the research process. Meanwhile, future research can further improve the accuracy of ecological environment quality assessment by combining more remote sensing data and ground observation data [56]. However, there are still limitations in using only four indicators, namely greenness, humidity, heat, and dryness, to evaluate the ecological environment condition [57], and the model is constructed using only the first principal component, so the RSEI may be one-sided [58]. Due to the influence of water bodies [49] and the inconsistency of the principal component loadings [59], the RSEI may not be comparable, and a more simple and scientific method and comprehensive index for ecological environment quality evaluation should be explored.

In summary, based on the above discussion and analysis, the following recommendations are made for ecological environmental protection in the Yellow River Delta region:

(1)

Land transformation and human disturbance in coastal mudflats in the Yellow River Delta region should be reduced to maintain the balance between the eco-environmental quality and the coastal mudflats.

(2)

When building near the Yellow River basin, attention should be paid to improving the ecological conditions of the coastline area. Take measures to reduce large-scale industrial expansion and construction, return farmland to forests, lakes, and grasses, and enhance the ecological environment.

(3)

Restore some abandoned land in the delta to lakes, forests, and grasslands to stop the spread of arable land. At the same time, a sound urban development plan should be formulated, and ecological construction subsidies should be implemented in urban development areas to reduce its negative impact.

6. Conclusions

Based on the GEE cloud computing platform, our manuscript integrates Landsat and MODIS remote sensing data to construct RSEI through principal component analysis to dynamically assess the eco-environmental quality of the Yellow River Delta from 2000 to 2020, while exploring the effects of land use and NPP on eco-environmental quality. Conclusions are drawn as follows:

(1)

In the Yellow River Delta region, the areas with good eco-environmental quality were mainly located near the Yellow River basin, and the distribution resembles a tilted “Y”, and the areas with poor environmental quality were located in the areas near the seaside edge of the Yellow River Delta. The eco-environmental quality gradually deteriorated from the middle to the edge of the Yellow River Delta, and the eco-environmental quality in urban areas was poor.

(2)

The eco-environmental quality of the Yellow River Delta showed a “V”-shaped fluctuation within 20 years, and the overall quality of the ecological environment was improving. From 2000 to 2010, the eco-environmental quality gradually became better. The eco-environmental quality of Dongying District gradually deteriorated, while Hekou District, Kenli District, and Lijin County gradually improved. From 2010 to 2020, the eco-environmental quality gradually deteriorated, with a large area of the eastern, central, and western parts of the delta experiencing the deterioration of the ecological environment, and the rest of the region showing a staggered deterioration and improvement.

(3)

There was no stable correlation between the eco-environmental quality and NPP in the Yellow River Delta region, indicating that there was no stable correlation between carbon fixation capacity and eco-environmental quality in the Yellow River Delta region. There was a positive correlation between urban construction land and eco-environmental quality among land use types in the Yellow River Delta, with the largest absolute value of land-use dynamic degree of 8.78%, significant land expansion, and obvious changes in land transfer. Arable land and eco-environmental quality showed a weak correlation, and the absolute value of land-use dynamic degree was the smallest, which was −1.01%. Arable land showed a slow decreasing trend, and land transfer changes were more obvious. Forest land and eco-environmental quality did not show correlation, and coastal mudflats showed negative correlation with eco-environmental quality, indicating that coastal mudflats in the Yellow River Delta had an important role in improving eco-environmental quality.