Next Article in Journal
Tradeoffs between Stand Volume and Understory Vegetation Diversity in Quercus wutaishanica Forests under Climate Change
Previous Article in Journal
Study on Nondestructive Detection Imaging Method of Log Knot Based on Judging the Shortest Path of Stress Wave Propagation
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 15
Issue 10
10.3390/f15101749
forests-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

Estimation of Water Demand for Riparian Forest Vegetation Based on Sentinel-2 Data: A Case Study of the Kokyar River Basin

by
Xianhe Liu
1,
Yilinuer Alifujiang
1,2,
Abdugheni Abliz
1,2,*,
Halidan Asaiduli
1,
Panqing Ye
1 and
Buasi Nurahmat
1
1
College of Geography and Remote Sensing Sciences, Xinjiang University, Urumqi 830046, China
2
Xinjiang Key Laboratory of Oasis Ecology, Xinjiang University, Urumqi 830017, China
*
Author to whom correspondence should be addressed.
Forests 2024, 15(10), 1749; https://doi.org/10.3390/f15101749
Submission received: 1 September 2024 / Revised: 26 September 2024 / Accepted: 2 October 2024 / Published: 4 October 2024
(This article belongs to the Topic Remote Sensing in Water Resources Management Models, 2nd Volume)
Browse Figures
Versions Notes

Abstract

:
In recent years, due to the shortage of water resources and the fragile ecological environment in arid areas, the relationship between vegetation and water resources has been relatively close. The unreasonable allocation of water resources and the excessive demand for ecological water use have led to ecological and environmental problems such as river interruption, land desertification, and the extensive withering of vegetation in arid areas; therefore, rapid, accurate estimation of the vegetation ecological water demand has become a hot research topic in related fields. In this study, we classified the land use types in the lower reaches of the Kokyar River Basin based on Sentinel-2A data and calculated the water requirements of each type of vegetation using a combination of the area quota method and improved Penman–Monteith (PM) based on different vegetation coverage levels. The results revealed that in 2020, the water demand of planted woodlands within 0–2 km of the watershed will be the highest, and the water demand of naturally growing arboreal woodlands will be the lowest, and the water demand of the surrounding desert riparian vegetation forests will be very small in relation to the ecological base flow and will not affect the downstream water use for agriculture, industry, and domestic use for the time being. The ecological water demand of the vegetation in the study area can be accurately estimated using Sentinel-2A data, and the research results provide technical support and a theoretical basis for rapid estimation of the ecological water demand of vegetation in typical riparian forests in arid areas and for the allocation of water resources.

1. Introduction

Plants are the most important components of eco-environments, and they are also the most important indicator of the growing impact on the environment. Ensuring the necessary water demand to support normal vegetation growth is crucial for maintaining an ecosystem’s functions, rational allocation of water resources, and coordinated development between ecological water demand and economic water demand in river basins [1]. In recent years, the task of balancing economic development with ecological environmental protection in western arid regions has become increasingly challenging [2]. Therefore, determining the types and distribution areas of natural vegetation within a basin, estimating their actual water demands, and providing important supporting data for basin management, protection of natural vegetation resources in rivers and lakes, sustainable utilization of water resources, and effective management are essential. This approach improves the water resource utilization efficiency, effectively avoids the inefficient use of water resources, and vigorously promotes basin ecological protection efforts [3,4]. Therefore, for the Kokyar River Basin, which is located in a typical arid region, it is essential to provide timely and reasonable protection of forested areas based on the estimation of the vegetation water demand.
Currently, in both China and internationally, research on the theory and calculation methods of the ecological water demand is still in the exploratory stage. Due to differences in the research areas, methods, and subjects [5], many scholars have gained varying understandings of the concept of ecological water demand, and there is no unified standard definition of ecological water demand. Gleick first proposed the concept of the basic ecological water requirement (BEWR) in 1998 [6]. In simple terms, the BEWR is the minimum amount of water resources needed to maintain the normal ecological functions of an ecosystem. He elaborated on the BEWR in both broad and narrow senses. Broadly, it refers to the minimum water consumption required to maintain the balance of water, heat, salt, and biological equilibrium in an ecosystem. Narrowly, it refers to the water consumption required to maintain basic vegetation growth and to perform protective functions [2].
The commonly used research methods for the vegetation water demand mainly include the area quota, potential evapotranspiration, water balance, and improved Penman–Monteith (PM) methods [7]. The area quota method is suitable for artificial vegetation and thoroughly studied natural vegetation. It determines the ecological water demand of a specific type of vegetation based on its ecological water quota. For example, Liu et al. divided the area downstream of the Da Xi Lake Reservoir to Tai Tema Lake in the Tarim River into sparse forest land, forest land, low-coverage grassland, and high-coverage grassland [8]. The minimum water requirements for the different types of vegetation were calculated based on the research conducted by Wang et al., and the water demand results were determined accordingly [9]. The advantage of the area quota method is that it involves fewer parameters, and the data are relatively easy to obtain. However, its drawback is that the water requirement quota standards need further verification, and it does not consider the influences of precipitation and groundwater on the water requirements of the vegetation in the ecological environment [10]. The potential evapotranspiration method refers to the process by which the water in the soil is transferred to the atmosphere through evaporation, also known as soil evaporation. Typically, potential evapotranspiration calculations can be achieved through mathematical models, laboratory experiments, and field observations. Common methods for calculating the potential evapotranspiration [11,12,13] include the energy balance method, evaporimeter method, weighing method, and empirical formula methods.
Ye et al. calculated the total ecological water demand of natural vegetation in the downstream area of the Tarim River Basin based on the areas of different types of vegetation and the different depths of the groundwater [14]. Gong et al. used remote sensing and geographic information system (GIS) technology to calculate the surface evapotranspiration in the Zhalong Wetland Protection Area, and they obtained the current ecological water demand in 2002, 2010, and 2016 [15].
The improved PM method is based on crop evapotranspiration, combined with crop coefficients and soil moisture limitation factors, and it calculates the vegetation’s ecological water demand [16]. Bai et al. used the PM formula and Hargreaves-Samani (HS) formula as the basis for quantitatively converting weather forecast information to predict the reference crop water demand in a typical area, i.e., Wuhan City [17]. The water balance method considers the vegetation ecosystem as a whole. By calculating the evapotranspiration and soil moisture content at the end of a period, the ecological water demand of the vegetation in that period can be obtained. Lu et al. established the correspondence between remote sensing data and water balance elements, selected reasonable regional evapotranspiration data, and decomposed the total evapotranspiration into precipitation evapotranspiration and irrigation evapotranspiration using the field water balance [18]. Then, they estimated the reasonable range of irrigation water consumption and irrigation water demand.
While research on the vegetation water demand (VWD) has matured and produced many results, future research should consider the impact of vegetation coverage on water demand calculations. Research in the Tuhai Basin in Xinjiang is limited, and there are few available references. In addition, relying solely on one method to estimate the water demand lacks persuasiveness. The Kokyar River Basin contains mixed growth of forests, shrubs, and grasslands and both artificial and natural forests. This region is characterized by a patchy vegetation distribution and high heterogeneity, which make it difficult to ensure the accuracy of land cover classification using medium- to low-resolution satellite imagery [19]. In recent years, high-resolution satellites covering the globe have emerged, such as the Sentinel and Gaofen (GF) satellites, providing favorable conditions for regional-scale vegetation coverage estimation [20,21]. Among them, Sentinel-2 data are capable of reflecting comprehensive land cover information and providing spatial data about land vegetation growth [22], soil coverage, regional environment [23,24], and even emergency assistance when necessary. The unique feature of the Sentinel-2A data containing three bands in the vegetation red edge range gives it an advantage in extracting and classifying natural vegetation in a region [25]. The high-resolution and multispectral characteristics of Sentinel-2A data greatly facilitate the extraction and classification of surface vegetation.
If it is possible to quickly and directly calculate the water demand of vegetation in a certain area, it will be more beneficial to the protection of the ecological environment, the allocation of water resources, and the reduction of water wastage in arid zones. The aim of this study was to utilize the significant advantages of Sentinel-2A data in monitoring terrestrial vegetation growth, including a short revisit period and rich spectral information. Using an object-oriented approach for natural vegetation classification, we extracted the main types of natural vegetation within a 2 km range along the Kokyar River bank area. Based on the resulting data, we employed the area quota method and potential evapotranspiration method to estimate the ecological water demand of the vegetation in this region. Many researchers have partially explored but have not yet completed a proper systematic approach. This paper intends to calculate the water demand of riparian forest vegetation in the Kokyar River Basin and to verify the feasibility of the methodology in the study area. The results of this study can assist decision-makers in providing a basis for water resource allocation and management to maximize the water resource utilization efficiency and to optimize the effectiveness of the governance in the Kokyar River Basin. Ultimately, the aim of this study was to improve the efficiency of freshwater resource utilization in the oases in Xinjiang and to achieve water conservation goals in arid regions.

2. Materials and Methods

2.1. Study Area

The Kokyar River is located in Shanshan County, Xinjiang Uygur Autonomous Region, China. It originates from the southern slope of the eastern Tianshan Mountains and flows from north to south. It is formed by the confluence of two main tributaries, the Kuokuerwu’er River and Qiongkeshi Lake. The river is 45.6 km long (89°58′–90°19′ E, 43°05′–43°25′ N latitude) (Figure 1). The Kokyar River Basin is bordered on the west by the Ertangou Basin and on the north by the Tianshan Mountains, which connect to the Mubi River Basin. It is classified as a mountain stream river and is mainly replenished by rainfall and snowmelt. The watershed area above the river’s mouth is approximately 707 km2, and the average annual runoff is 1.123 × 108 m3. The Kokyar River receives 85.29% of its annual runoff from May to August, with peaks during four flood seasons. Specifically, the summer months (June to August) contribute 68.1% of the annual runoff, the spring months (March to May) contribute 18.9%, and the autumn months (September to November) contribute 11.4%. The winter months (December to February) contribute minimally, accounting for only 1.6% of the annual runoff. In the downstream area, the Kokyar River is intercepted by the Kokyar Reservoir, which serves as a water supply for industrial, urban, and agricultural irrigation activities.
The main natural vegetation types and dominant species along the banks of the Kokyar River Basin include Tamarix ramosissima Ldb, Salix matsudana Koidz, Chondrella piptocoma, and Ulmus pumila L. In addition to the natural vegetation, there are areas of artificial forests, dominated by Populus nigra var. thevestina Populus nigra var. thevestina, which are typically found in this region. The vegetation coverage along the riverbanks ranges from approximately 60% to 90% (Figure 2). During our taxonomic and field surveys, we focused on the natural vegetation within 0–2 km of the banks of the first 10 km of the river in the Kokyar River Basin.

2.2. Data Source and Preprocessing

Based on the local climatic conditions, September provides optimal solar radiation levels for distinct differentiation of the land cover types on remote sensing imagery. Therefore, in this study, Sentinel-2A multispectral satellite imagery data acquired on 22 September 2020, were selected and obtained from the European Space Agency’s Copernicus Data Hub (https://dataspace.copernicus.eu, accessed on 1 September 2024). The selected date featured clear weather conditions within the study area, and the images are free of cloud cover and other natural factors that could obscure visibility. As a result, the various land cover types were clearly discernible. The false color composite results of Sentinel-2A imagery data bands B8 (red), B4 (green), and B3 (blue) for the study area are depicted in Figure 3.
To ensure the correct utilization of the remote sensing imagery and optimal representation of the target features, it was necessary to perform data preprocessing on the remote sensing image data. The most commonly used preprocessing tasks include remote sensing data format conversion, multi-band synthesis, radiometric and geometric corrections, image mosaic processing, and image clipping. The preprocessing steps used in this study included using the SNAP version 9.0 software to convert the Sentinel-2A data from the FTDR format to the TIFF format, which is compatible with the ENVI version 5.6 software, and generating 13 bands with a spatial resolution of 10 m. To enhance the land use information for the study area, the blue band (B2), green band (B3), red band (B4), and near-infrared band (B8) were selected and synthesized. The image was then clipped using the vector boundaries of the 2 km range from the Kokyar River Basin to obtain an image of the study area for land use information extraction. The specific preprocessing workflow is illustrated in Figure 4.

2.3. Research on Segmentation and Merging of Images

Image segmentation refers to the process of merging adjacent and homogeneous pixels and separating non-adjacent and heterogeneous pixels [26]. The rationality of segmentation directly affects the classification results, making image segmentation the most critical step in object-oriented classification. In this study, the contrast between adjacent pixels, spectral homogeneity, and heterogeneity thresholds were used as references to segment and merge the image pixels to determine the optimal scale. The obtained results are shown in Figure 5.

2.4. Establelishing Classification Hierarchy and Rules

Classification rules refer to the principles specified for the classification process based on the spectral characteristics of each classification system and the actual situation. According to the feature space and defined samples of each land cover type, as well as the national remote sensing image classification system standards for land cover categories [27], in this study, we designed a classification system that included eight types of land cover: grassland, shrubland, woodland, plantations, water bodies, bare land, residential area, and roads. The interpretation symbols for each land use type in the Kokyar River Basin were also determined (Table 1).

2.5. Research Methods

The vegetation coverage can be determined using high-resolution remote sensing imagery data, such as satellite or aerial imagery [28,29]. This involves extracting vegetation information using methods such as image classification and pixel-based analysis. Subsequently, the percentage of the vegetation coverage is calculated by dividing the number of pixels representing vegetation by the total number of pixels. Based on different distance ranges from the river, namely, 0–1 km and 1–2 km from the banks of the Kokyar River, the vegetation coverage areas of the different types of vegetation coverage are calculated.
The vegetation coverage can be determined using a pixel binary model, which assumes that each pixel on the ground consists only of vegetation and non-vegetation parts. This model incorporates the normalized difference vegetation index (NDVI) and can be expressed as Formular (1) [30]:
f c = N D V I − N D V I s o i l N D V I v e g − N D V I s o i l
where NDVI is the NDVI value of each pixel; NDVIsoil is the NDVI value when there is no vegetation cover; and NDVIveg is the NDVI value when there is full vegetation cover.
The ecological water demand of vegetation refers to the amount of water needed for plants to maintain normal growth in a specific environment. It can be understood as the water required by plants in a particular ecological setting. It describes the quantity of water necessary for the growth of a specific type of vegetation in a given geographic area [31]. Plants absorb water for metabolic and growth processes, and their water requirements vary under different environmental conditions. Therefore, the concept of the ecological water demand for vegetation involves calculating the water needed by plants based on different vegetation types, ecological conditions, and climate characteristics, providing insights for irrigation and water resource management practices.
Given the availability of abundant reference data, the direct calculation method can be chosen. The calculation Formular (2) [32] is as follows:
W = ∑ i = 1 n W i = ∑ i = 1 n A i r i
where Wi is the water demand of vegetation type i, A i is the area of vegetation type i, and r i is the ecological water demand quota of vegetation type i.
In the absence of reliable direct reference data, the evapotranspiration of plants can be used to conduct the estimation using the potential evapotranspiration method. The Formular (3) [32] for this method is as follows:
E w = S v ⋅ ET 0 · K c K s
where Ew is the ecological water consumption for maintaining a certain area of vegetation (m3); Sv is the area of natural vegetation to be maintained (hm2); Kc is the vegetation coefficient, which varies with the vegetation growth period; and Ks is the soil moisture limiting coefficient (also known as wilting coefficient).
(1)
Evapotranspiration calculation
The calculation of evapotranspiration ( E T 0 ) of reference crops is based on the improved Penman Monteith formula proposed by Penman [33]. The specific formula is as Formular (4):
E T 0 = 0.408 Δ R n − G + r 900 T + 273 u 2 e s − e a Δ + r 1 + 0.34 u 2
In the formula, E T 0 is the reference crop evapotranspiration (mm/d), Δ is the slope of the saturated water vapor pressure curve (kPa/°C), G is the soil heat flux (MJ m−2 d−1), r is the hygrometer constant (kPa/°C), u 2 is the wind speed at a height of 2 m (m/s), e s is the saturated water vapor pressure (kPa), e a is the actual water vapor pressure (kPa), and R n is the surface net radiation (MJ m−2 d−1).
The reference crop evapotranspiration data for this study comes from the 1 km China monthly reference crop evapotranspiration E T 0 dataset created by scholar Peng Shouzhang using meteorological data [34,35,36,37,38].
(2)
Soil moisture limitation analysis
The soil moisture limitation coefficient (also known as wilting coefficient) is mainly influenced by soil texture. This study used the soil influencing factor Ks at 42 stations in Xinjiang to obtain the Hami station data from the annual average data [39], and 0.35 was taken as the Ks in the Tuha Basin region.
(3)
Vegetation coefficient
The vegetation coefficient is a function of vegetation coverage, vegetation height, and leaf area index, which varies with the growth period of vegetation. The vegetation coefficient of different vegetation in Shanshan County is determined based on the ratio of leaf area index in the early and late stages of growth to leaf area index in the middle stage of growth. This article refers to the research of relevant scholars [33,34,35,36,37,38,39,40], and the vegetation growth season in various regions of Xinjiang is from April to October. The vegetation growth stages are divided into early growth stage (April), growth development stage (May–June), mid growth stage (July–September), and late growth stage (October). Therefore, this article determines the growth period rules and plant coefficients of each vegetation type (Table 2).

2.6. Research Process

After preprocessing the satellite data, the next step involves using the object-oriented classification method to determine the segmentation and merging scale, set up sample categories, and conduct classification. The object-oriented classification method takes into account the geometric information, spatial topology relationships, and texture features of the image, and it segments the remote sensing image into meaningful and non-overlapping classification units to improve the classification accuracy [41]. In this paper, the support vector machine (SVM) method is used to conduct object-oriented classification. Using the contrast, spectral homogeneity, and heterogeneity thresholds between adjacent pixels as a reference, segment and merge image pixels to determine the optimal scale. In the SVM classification method parameter settings, the kernel type options are linear, polynomial, radial basis function, and sigmoid. In this paper, we choose polynomial and set the maximum value of the number of times the core polynomial (degree of kernel polynomial) to 4; we control the balance between the sample error and the extension of classification rigidity parameter penalty parameter with a set value of 90; we set the hierarchical processing level for SVM training and classification process parameter pyramid levels to use the default value of 0; and the classification probability threshold for the classification sets the probability domain value, i.e., if a pixel calculates the probability of matching all the rules to be less than this value, the pixel will not be classified and the range is 0~1, then this experiment chooses 0.1. After multiple experiments, it was found that setting the scale level value in the segmentation setting to 35 and the merge level value in the merge setting to 85 had the best effect. Even though the number of samples is limited and the number of training samples is small, the SVM can still achieve good classification results [42,43]. After training using the training samples, the preliminary classification results are obtained. Then, the initial classification results are further processed via clustering and filtering to obtain the final classification results [44]. Subsequently, a vegetation classification thematic map is drawn based on the accuracy test. If the results do not meet the quality standards, the classification method is adjusted, and the classification is performed again. The vegetation area is calculated based on the vegetation classification thematic map, and the vegetation water demand within 0–2 km of the Kokyar River Basin is estimated using direct or indirect calculation methods. Finally, conclusions are drawn based on discussion and demonstration. The specific research workflow is shown in Figure 6.

3. Results and Discussion

3.1. Land Use Information and Vegetation Coverage Area

The object-oriented classification method used for extracting the land use information for the Kokyar River Basin yielded good visual effects and classification results. The results meet the research requirements and provide favorable supporting data for calculating the regional ecological water demand. The accuracy evaluation based on the confusion matrix is presented in Table 3.
The research results indicate that within the 2 km of the basin, bare land covered 93% of the area, accounting for the highest proportion, and was evenly distributed along the upstream and downstream areas of the river. The next largest land use type was artificial forest land (2.4%), which was primarily distributed in the downstream area. The forest land occupied the smallest area, accounting for only 0.13% of the total area, and was mainly distributed in the downstream area. The other land cover types were concentrated along both sides of the riverbank, and the residential areas exhibited the most scattered distribution. These findings further confirm the low vegetation coverage and typical characteristics of water scarcity in the Kokyar River Basin. The area and percentage of each land cover type are summarized in Table 4.
The calculation results (Table 5) revealed that the artificial forest land had a high coverage (>75% coverage) and occurred as large contiguous areas, consistent with the characteristics of artificial forests. By contrast, the other three types of natural vegetation were scattered. The shrubland had the highest distribution when the coverage was less than 45%, and the overall coverage increased with decreasing distance from the river channel. By calculating the total area based on different coverage levels, it was found that the artificial forest land covered the largest area, with a total of 1.17 km2, followed by the forest land (0.06 km2), shrubland (0.33 km2), and grassland (0.11 km2). The specific results are illustrated in Figure 7 and Figure 8 and are displayed on the map in Figure 9.

3.2. Calculation of Vegetation Water Demand

Referring to the preliminary estimation of the ecological water consumption in Xinjiang 31, based on the field investigation, the main species in the forested areas were desert riparian forests composed of willow and elm trees, and there was little difference in the water requirements of the two species. Among them, the willow trees accounted for the majority. The evapotranspiration of the willow trees was calculated to be 2000.4 m3/hm2, and the water consumption was estimated using the indirect calculation method. For the artificial forest land, the ecological water requirement quotas were determined based on various local forestry standards in Xinjiang (Table 6) [45], and the direct calculation method was selected. In the shrubbery areas, the main species was Haloxylon ammodendron, with the majority being Calligonum mongolicum Turcz. The evapotranspiration of the Haloxylon ammodendron was 3637.5 m3/hm2, and the indirect calculation method was selected. For the grasslands, which mainly consisted of floodplain meadows, the ecological water requirement quotas were 2625 m3/hm2 in southern Xinjiang and 1470 m3/hm2 in northern Xinjiang, and the direct calculation method was selected. The total area of the vegetation and the estimated vegetation water requirements for the different vegetation types along the banks of the Kokyar River were calculated based on their coverages (Table 7).

3.3. Analysis of Classification Results

Although different land cover types exhibit distinct spectral properties, human visual interpretation introduces subjective factors, and the geographical phenomena and processes are complex. After initial classification, some land cover types may exhibit a salt-and-pepper look [41]. As shown in Figure 10 (in the figure, orange represents grassland, dark green represents tree forest land, green represents shrub forest land, light green represents artificial forest land, blue represents water bodies, pink represents bare land, and yellow represents roads; misclassified forest land is on the left, misclassified roads are in the middle, and misclassified grassland is on the right), the errors and omissions in the classification were adjusted and corrected using the ENVI software and the land use information classified after by human observation. Roads have a prominent elongated feature, so in the land use information extraction process, spatial elongation rules can be set to remove road information [46]. However, due to the unique geographical, topographical, and environmental factors in the study area, the elongation method for roads may not fully correct the road information, which in turn may affect the information about the other land use types. In this study, we utilized the online LocaSpace Viewer software Version 3.2.9 to modify the misclassified and omitted land use information.
In response to the land use information for the Kokyar River Basin in 2020, in this study, we utilized the online LocaSpace Viewer platform to conduct manual visual interpretation. Several validation samples were uniformly and randomly selected for verification. The distribution of the verification points is shown in Figure 11.

3.4. Analysis of Vegetation Water Demand Results

The accuracy of the land use classification and the precision of the vegetation classification directly impact the accuracy of the calculations of the vegetation water requirements. Sparse forest ecosystems, which contain complex interwoven vegetation, present challenges in vegetation remote sensing monitoring and remain the land cover type with the lowest classification accuracy globally [47,48]. For the overall results obtained in this study, the overall classification accuracy and kappa coefficient are 86.74% and 0.84, respectively. Among them, the classification accuracies of the water bodies, bare land, and residential areas exceed 90%, indicating that the object-oriented classification method has an excellent classification performance for these three land cover types. However, the mapping accuracy of the grassland in the entire classification evaluation is 47%, suggesting difficulties in classifying the grassland in the study area and a low visual effectiveness. This could be due to several reasons. It is important to choose the spatial distribution of the classification samples in the image. Ideally, the samples should be evenly distributed throughout the image to avoid classification bias due to too many or too few samples in a localized area, but most of the vegetation samples within the watershed of the Kokoa River are clustered and distributed and interspersed with each other, which affects the accuracy of classification. Remote sensing images appear to contain a variety of features within a single image element, which can lead to the spectral features of the image element not representing any one pure feature class, thus increasing the difficulty of classification. The existence of mixed image elements makes it difficult for the classifier to accurately classify the image elements into the correct category because the classifier usually distinguishes based on spectral features, while the spectral features of mixed image elements are the superposition of multiple features, which will lead to a decrease in classification accuracy, and in this paper, we adopt the mixed image decomposition and end-element extraction can alleviate such effects. Additionally, the study area is located in a typical arid region, and the presence of many small and short channels in the basin can easily lead to misclassification. The scattered distribution of the residential areas in the study area may result in underrepresentation if the training samples are not uniformly selected, leading to misclassification. Finally, the spectral information in the images is affected by highways, roads, and areas in the shadows cast by buildings in the study area, all of which can decrease the classification accuracy.
The Kokyar River Basin is located in an extremely arid region where the groundwater depth exceeds 30 m. Currently, the most suitable methods for estimating the ecological water demand are the area quota method and the potential evapotranspiration method. The ecological water demand of plants is not only related to natural factors such as climatic conditions and soil substrates but also greatly influenced by community types and plant species.
In this context, it is feasible to estimate the ecological water demand of the entire system based on the main tree species in the different forest stands [2]. Based on the calculation results, it was found that the artificial forest land covered the largest area and had the highest water demand. It heavily relied on river water for irrigation and was located between two reservoirs where regular human interventions for planting and maintenance were carried out. The shrub forest land consisted of the dominant native species in the area and required less water for evapotranspiration compared with the forest land. The shrubs could still grow beyond 1 km from the riverbanks. By contrast, less area was covered by deciduous forest land due to its higher natural water demand. It could only grow around the riverbanks without human intervention, resulting in a smaller distribution range. Most of the grasslands in the area had a high coverage, which was related to their natural characteristics. The abundant grassland cover effectively reduced evapotranspiration and withstood intense sunlight. Similar to the deciduous forest land, the grasslands were highly dependent on water, resulting in a smaller distribution area. Therefore, in the Kokyar River Basin, the artificial forest land had the highest water demand, and the deciduous forest land had the lowest water demand.
We compare other water demand calculation studies, such as Benjamin D. Goffin’s use of this method in the Maipo River Basin of central Chile, combining NDVI, Landsat-8 Operational Land Imager (OLI), and Terra Moderate Resolution Imaging Spectroradiometer (MODIS) to develop a regional model of water scarcity and validate the R2 between 0.77 and 0.92 for maize during 2018–2022. This method is recognized as effective in calculating water demand in arid areas with limited observations [49] and in combination with Liangliang Jiang’s research in the Ailik Lake Basin in Xinjiang [50]. In this paper, after verifying the water demand of the woodland in the river valley, we found that there is a gap with the ecological recharge value considering that the local annual precipitation in the study area is only 15–30 mm and the evapotranspiration can reach 3000 mm. We modified the value and found that it is within a reasonable range. Dengkai Chi used the Penman–Monteith method with GIS and remote sensing techniques to dynamically calculate the vegetation coefficient and introduced the soil moisture limitation coefficient to calculate the vegetation coefficient of Ergot. This coefficient is used to calculate the vegetation water demand for four vegetation types (coniferous forest; broad-leaved forest; steppe; and meadow coniferous forest, broad-leaved forest, grassland, or meadow) in the Ergune River basin, which proved the effectiveness of this method [1]. Jaouad El Hachimi, in the Tadla irrigated perimeter in central Morocco, calculated the local water demand of agricultural vegetation. The study area is located in a similar climate to that of the study area of this paper, and both are in arid and water-scarce areas, with the highest water demand in the months of March and April and the highest water demand for cereals, sugar beets, alfalfa, citrus fruits, and olive trees; vegetation was calculated and obtained better results [51]. Therefore, the model has a good rationality to be used in this study area. Dong X calculated the average monthly water requirement quota of all kinds of vegetation in the central region of the Eurasian continent. The monthly water requirement quota for all types of vegetation over many years (forest land, brushland, grassland, desert grassland, farmland, flood land) [52] is different from those presented in this paper. After careful analysis, it was found that due to the difference in vegetation cover, the limiting parameter Kc is different, which affects the results differently. In conclusion, this method is suitable for use in this study area and has been verified to perform well in other study areas, but there is a slight difference in the results of some papers because of the difference in the vegetation cover and the influence of parameters such as the value of evapotranspiration.
In this study, we utilized both direct and indirect calculation methods based on the area quota approach. According to the calculation results, the water demand of the vegetation on both banks of the Kokyar River, downstream of the Kokyar and Kokyar II reservoirs, accounted for only 0.71% of the annual runoff. However, the peak water demand of the vegetation primarily occurred in June, July, and August. Based on the measured data (Figure 12), the water demand of the vegetation accounted for a very small proportion at only 1.9% of the total outflow water volume. This suggests that there is still significant potential for further development of the vegetation on both riverbanks. Given the significant burden posed by industrial water use and agricultural irrigation in the downstream area, decision-makers should more reasonably allocate water resources based on the actual situation.
According to various calculation methods for ecological base flow assurance [53,54,55], for example, the Tennant method, also known as the Montana method, is one of the most commonly used ecological flow calculation methods, generally using 10% to 30% of the annual average flow as the ecological base flow. The Texas method generally takes 30% to 40% of the monthly average flow rate with a 50% guarantee rate. In this study, the ecological water demand of vegetation on both sides was only 1.9%, far below the threshold that can affect ecological base flow. We calculated the minimum amount of water required for the river vegetation ecosystem, the amount of water needed to maintain the basic ecological function of the river without damage, and the amount of water required for each period of the year without depletion and breakage of the flow and other phenomena. Based on hydrological parameters, using the 7Q10 method [56], using a 90% guarantee rate of the deadliest month of the average amount of water for seven consecutive months as the minimum flow design value of the Ning Li River derives the basic ecological water requirement of the river as the river protection threshold [57]. The flow curve can meet the water demand of the vegetation, and the surrounding vegetation did not expand because of the climate influence and the small watershed area.

4. Conclusions

In this study, we utilized Sentinel-2A data and employed the support vector machine (SVM) method based on object-oriented classification to investigate the water demand of the vegetation in the Kokyar River Basin in Shanshan County. The research findings reveal that within a 2 km radius of the basin, bare land covered 93% of the total area as evenly distributed along the upstream and downstream sections of the river and accounted for the highest proportion. Regarding the types of vegetation, the artificial forests covered the largest area, accounting for 2.4% of the total area, and were mainly distributed in the downstream section of the river. Conversely, the natural woodland covered the smallest area, accounting for 0.12% of the total area, and was primarily distributed in the downstream section of the river. The other land types were concentrated on both sides of the riverbanks, and no arable land was present due to the natural conditions. The water bodies covered 1.26 km2, accounting for 1.67% of the total area. The residential areas were scattered, with a total area of 0.8 km2, accounting for 1.05% of the total area. These areas consisted of local residences and workplace sites along the riverbanks. The roads occupied 0.23% of the total area.
According to the calculations based on the different coverage levels, the areas of the different types of vegetation within 0–2 km on both banks of the Kokyar River were as follows: the grassland, shrubland, woodland, and artificial forest-covered areas of 0.114 km2, 0.329 km2, 0.594 km2, and 1.166 km2, respectively.
Using the area quota method and the potential evapotranspiration method, the water demand of the vegetation within 0–2 km on both banks of the Kokyar River was calculated. The results are as follows: The water demand of the grassland was 1.6778 × 104 m3, accounting for a small proportion (2.76%) of the total water demand of the vegetation. The water demand of the shrubland was 1.1958 × 105 m3, accounting for 19.66% of the total water demand of the vegetation. The water demand of the woodland was 1.1883 × 104 m3, accounting for 1.95% of the total water demand of the vegetation, i.e., a lower water demand. The artificial forest had the highest water demand (4.5999 × 105 m3), accounting for 75.63% of the total vegetation water demand. The total water demand was 6.8234 × 105 m3. By considering the total water demand in relation to the base flow, it was found that it accounted for only 1.9%, indicating that the current base flow did not significantly affect the development of the surrounding vegetation, suggesting the potential for further development.
The utilization of Sentinel-2A satellite data significantly improved the accuracy of the vegetation classification. Our research indicates that integrating high- to medium-resolution satellite imagery allows for rapid and precise calculation of vegetation water requirements. This approach was found to be highly effective in arid regions, and the results can facilitate both the utilization and conservation of water resources.

Author Contributions

X.L. designed and wrote this paper, performed the experiments, and conducted the fieldwork; Y.A. performed data collection and analysis, and conceptualization; A.A. conceived and supervision this study; H.A., P.Y. and B.N. was responsible for manuscript proofreading and data analysis. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by the Third Xinjiang Scientific Expedition Program (2022xjkk1100) and founded by the 2024 Intramural Cultivation Program of Philosophy and Social Sciences (Commissioned by the Institute of Central Asian Studies) (No. 24FPY002).

Data Availability Statement

Data are contained within the article.

Acknowledgments

Authors wish to thank the referees for providing helpful suggestions to improve this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Chi, D.; Wang, H.; Li, X.; Liu, H.; Li, X. Estimation of the ecological water requirement for natural vegetation in the Ergune River basin in Northeastern China from 2001 to 2014. Ecol. Indic. 2018, 92, 141–150. [Google Scholar] [CrossRef]
  2. Su, X.; Liu, Y.; Jiang, T.; Hai, W.; Xuan, L.; Xiao, L. Prediction of future ecological water demand of terrestrial ecosystem in Northwest China. JWARP 2023, 39, 9–18+78. [Google Scholar]
  3. Ling, H.; Guo, B.; Xu, H.; Fu, J. Configuration of water resources for a typical river basin in an arid region of China based on the ecological water requirements (EWRs) of desert riparian vegetation. Glob. Planet. Change 2014, 122, 292–304. [Google Scholar] [CrossRef]
  4. Alifujiang, Y.; Lu, N.; Feng, P.; Jiang, Y. China’s Urban Water Utilization Based on the Water Footprint Methodology. Water 2024, 16, 462. [Google Scholar] [CrossRef]
  5. Shen, X.; Yang, X.; Wei, T. Research status and review of vegetation ecological water demand in arid regions. Plateau Agric. 2018, 2, 544–548+575. [Google Scholar]
  6. Gleick, P.H. Water in crisis: Paths to sustainable water use. Ecol. Appl. 1998, 8, 571–579. [Google Scholar] [CrossRef]
  7. Hao, B.; Su, X.; Ma, X.; Zhang, D. Research progress of vegetation ecological water requirements in arid area. J. Water Res. Water Eng. 2009, 20, 1–5. [Google Scholar]
  8. Liu, X.; Xu, H.; Ling, H.; Bai, Y.; Fu, J. Ecological water requirements in the lower reaches of the tarim river. Chin. J. Desert Res. 2013, 33, 1198–1205. [Google Scholar]
  9. Wang, R.; Song, Y.; Fan, Z.; Ma, Y. Estimation on Ecological Water Demand Amount in Four Sources and One Main Stream Area of Tarim Basin. J. Soil. Water Conserv. 2001, 15, 19–22. [Google Scholar]
  10. Liu, Z.; Zhang, M.; Pan, L. Discussion on calculation methods of ecological water demand. Anhui Agric. Sci. 2008, 36, 13921–13922. [Google Scholar]
  11. Xiang, K.; Li, Y.; Horton, R.; Feng, H. Similarity and difference of potential evapotranspiration and reference crop evapotranspiration–a review. Agric. Water Manag. 2020, 232, 106043. [Google Scholar] [CrossRef]
  12. Pereira, L.S.; Allen, R.G.; Smith, M.; Raes, D. Crop evapotranspiration estimation with FAO56: Past and future. Agric. Water Manag. 2015, 147, 4–20. [Google Scholar] [CrossRef]
  13. Hargreaves, G.H.; Samani, Z.A. Reference crop evapotranspiration from temperature. Appl. Eng. Agric. 1985, 1, 96–99. [Google Scholar] [CrossRef]
  14. Ye, Z.; Chen, Y.; Li, W. Ecological water demand of natural vegetation in the lower Tarim River. J. Geogr. Sci. 2010, 20, 261–272. [Google Scholar] [CrossRef]
  15. Gong, Z.; Lu, L.; Jin, D.; Qiu, H.; Zhang, Q.; Guan, H. Remote sensing estimation of evapotranspiration and ecological water demand in Zhalong wetland under land use/cover change. Acta Ecol. Sin. 2021, 41, 3572–3587. [Google Scholar]
  16. Monteith, J.L. Environmental Control of Plant Growth; Academic Press: New York, NY, USA, 1963; pp. 95–112. [Google Scholar]
  17. Bai, Y.; Lu, M.; Cheng, H.; Yan, D.; Sun, H. Research on the reference crop evapotranspiration forecast based on weather forecast information bai yi-wen. China Rural. Water Hydropower 2021, 10, 175–179+184. [Google Scholar]
  18. Lu, S.; Zhao, H.; Jiang, Y.; Hao, Z.; Zhang, X.; Chen, G. Analysis of irrigation water based on multi-source remote sensing data and water balance principle. J. Hydraulic. Eng. 2021, 52, 1126–1135. [Google Scholar]
  19. Aghababaei, M.; Ebrahimi, A.; Naghipour, A.A.; Asadi, E.; Verrelst, J. Vegetation Types Mapping Using Multi-Temporal Landsat Images in the Google Earth Engine Platform. Remote Sens. 2021, 13, 4683. [Google Scholar] [CrossRef]
  20. Samat, A.; Yokoya, N.; Du, P.; Liu, S.; Ma, L.; Ge, Y.; Issanova, G.; Saparov, A.; Abuduwaili, J.; Lin, C. Direct, ECOC, ND and END Frameworks—Which One Is the Best? An Empirical Study of Sentinel-2A MSIL1C Image Classification for Arid-Land Vegetation Mapping in the Ili River Delta, Kazakhstan. Remote Sens. 2019, 11, 1953. [Google Scholar] [CrossRef]
  21. Li, X.; Tian, X.; Duan, T.; Tian, Q.; Wang, Y.; Verrelst, J.; Mu, X.; Gu, X. Remote sensing estimation of woody and herbaceous vegetation coverage in temperate sparse forest grassland using integrated drone and satellite imagery. J. Remote Sens. 2023, 27, 2139–2152. [Google Scholar]
  22. Herrmann, I.; Pimstein, A.; Karnieli, A.; Cohen, Y.; Alchanatis, V. LAI assessment of wheat and potato crops by VENμS and Sentinel-2 bands. Remote Sens. Environ. 2011, 115, 2141–2151. [Google Scholar] [CrossRef]
  23. Mahdianpari, M.; Salehi, B.; Mohammadimanesh, F.; Gill, E. The first wetland inventory map of newfoundland at a spatial resolution of 10 m using sentinel-1 and sentinel-2 data on the google earth engine cloud computing platform. Remote Sens. 2018, 11, 43. [Google Scholar] [CrossRef]
  24. Miletić, B.R.; Matović, B.; Orlović, S.; Gutalj, M.; Đorem, T.; Marinković, G.; Simović, S.; Dugalić, M.; Stojanović, D.B. Quantifying Forest Cover Loss as a Response to Drought and Dieback of Norway Spruce and Evaluating Sensitivity of Various Vegetation Indices Using Remote Sensing. Forests 2024, 15, 662. [Google Scholar] [CrossRef]
  25. Tian, Y.; Chen, Z.; Hui, F.; Cheng, X.; Ouyang, L. ESA Sentinel-2A/Bsatellite: Characteristics and applications. J. Beijing Norm. Univ. (Nat. Sci.) 2019, 55, 57–65. [Google Scholar]
  26. Zhu, X.; Ning, X.; Wang, H.; Zhang, H. Land use classification for optimization segmentation based on high-precision land cover data. Bull. Surv. Mapp. 2021, 46, 140–149. [Google Scholar]
  27. Khatami, R.; Mountrakis, G.; Stehman, S.V. A meta-analysis of remote sensing research on supervised pixel-based land-cover image classification processes: General guidelines for practitioners and future research. Remote Sens. Environ. 2016, 177, 89–100. [Google Scholar] [CrossRef]
  28. Jia, K.; Yao, Y.; Wei, X.; Tian, Q.; Wang, Y.; Verrelste, J.; Mu, X.; Gu, X. Research progress on remote sensing estimation of vegetation coverage. Adv. Atmos. Sci. 2013, 28, 774–782. [Google Scholar]
  29. Wang, X.; Zhang, Y.; Zhang, K. Automatic 10 m Forest Cover Mapping in 2020 at China’s Han River Basin by Fusing ESA Sentinel-1/Sentinel-2 Land Cover and Sentinel-2 near Real-Time Forest Cover Possibility. Forests 2023, 14, 1133. [Google Scholar] [CrossRef]
  30. Covich, A. Water in Crisis: A Guide to the World’s Fresh Water Resources; Oxford University Press: New York, NY, USA, 1993. [Google Scholar]
  31. Jia, B.; Ci, L. The primary estimation of water demand by the eco-environment in Xinjiang. Acta Ecol. Sin. 2000, 20, 243–251. [Google Scholar]
  32. Hu, G.; Zhao, W. Review on calculation methods of ecological water demand for vegetation in arid and semi-arid regions. Acta Ecol. Sin. 2008, 28, 6282–6291. [Google Scholar]
  33. Liu, Y.; Pereira, L.S. Validation of the Crop Coefficient Calculation Method Recommended by FAO. J. Agric. Eng. 2000, 16, 26–30. [Google Scholar]
  34. Peng, S. 1 km Monthly Potential Evapotranspiration Dataset in China (1901–2022); National Tibetan Plateau/Third Pole Environment Data Center: Yangling, China, 2022. [Google Scholar]
  35. Peng, S.Z.; Ding, Y.X.; Wen, Z.M.; Chen, Y.M.; Cao, Y.; Ren, J.Y. Spatiotemporal change and trend analysis of potential evapotranspiration over the Loess Plateau of China during 2011–2100. Agric. For. Meteorol. 2017, 233, 183–194. [Google Scholar] [CrossRef]
  36. Ding, Y.X.; Peng, S.Z. Spatiotemporal trends and attribution of drought across China from 1901–2100. Sustainability 2020, 12, 477. [Google Scholar] [CrossRef]
  37. Ding, Y.X.; Peng, S.Z. Spatiotemporal change and attribution of potential evapotranspiration over China from 1901 to 2100. Theor. Appl. Clim. 2021, 145, 79–94. [Google Scholar] [CrossRef]
  38. Peng, S.Z.; Ding, Y.X.; Liu, W.Z.; Li, Z. 1 km monthly temperature and precipitation dataset for China from 1901 to 2017. Earth Syst. Sci. Data 2019, 11, 1931–1946. [Google Scholar] [CrossRef]
  39. Liao, Y. Research on Vegetation Ecological Water Demand in Xinjiang Region Based on GIS; Northwest A&F University: Yangling, China, 2012. [Google Scholar]
  40. Chen, Y. Ecological Protection and Sustainable Management of Tarim River Basin in Xinjiang; Science Press: Beijing, China, 2015. [Google Scholar]
  41. Chang, C.; Zhao, G.; Wang, L.; Zhu, X.; Gao, Z. Land use classification based on RS object-oriented method in coastal spectral confusion region. J. Agric. Eng. 2012, 28, 226–231. [Google Scholar]
  42. Tzotsos, A.; Argialas, D. Support vector machine classification for object-based image analysis. In Object-Based Image Analysis: Spatial Concepts for Knowledge-Driven Remote Sensing Applications; Springer: Berlin/Heidelberg, Germany, 2008; pp. 663–677. [Google Scholar]
  43. Li, H.; Gu, H.; Han, Y.; Yang, J. Object-oriented classification of high-resolution remote sensing imagery based on an improved colour structure code and a support vector machine. Int. J. Remote Sens. 2010, 31, 1453–1470. [Google Scholar] [CrossRef]
  44. Wang, L.; Guo, Y.; He, J.; Wang, L.; Zhang, X.; Liu, T. Classification Method by Fusion of Decision Tree and SVM Based on Sentinel-2A Image. Trans. Chin. Soc. Agric. Mach. 2018, 49, 146–153. [Google Scholar]
  45. Keyimu, M.; Halik, Ü.; Rouzi, A. Relating water use to tree vitality of Populus euphratica Oliv. in the lower Tarim River, NW China. Water 2017, 9, 622. [Google Scholar] [CrossRef]
  46. Fu, F. Research on Urban Green Space Extraction Based on Sentinel-1A and Sentinel-2A Data. Zhejiang Normal University: Jinhua, China, 2020. [Google Scholar]
  47. Chen, Q.; Li, X.; Xiu, X.; Yang, G. Large scale shrub coverage mapping of sandy land at 30m resolution based on Google Earth Engine and machine learning. Acta Ecol. Sin. 2019, 39, 4056–4069. [Google Scholar]
  48. Yue, J.; Mu, G.; Tang, Z.; Yang, X.; Lin, Y.; Xu, L. Remote sensing estimation models for vegetation coverage in desert regions of Xinjiang based on NDVI. Arid Land Geogr. 2020, 43, 153–160. [Google Scholar]
  49. Goffin, B.D.; Thakur, R.; Carlos, S.D.C.; Srsic, D.; Williams, C.; Ross, K.; Neira-Román, F.; Cortés-Monroy, C.C.; Lakshmi, V. Leveraging remotely-sensed vegetation indices to evaluate crop coefficients and actual irrigation requirements in the water-stressed Maipo River Basin of Central Chile. Sustain. Horiz. 2022, 4, 100039. [Google Scholar] [CrossRef]
  50. Jiang, L.; Jiao, J.; Cheng, Y.; Wang, X.; Liu, J.; Li, Y. Study on ecological water requirement for the lake and natural vegetation in the Alice Lake basin. Arid Zone Res. Environ. 2016, 30, 108–114. [Google Scholar]
  51. El Hachimi, J.; El Harti, A.; Lhissou, R.; Ouzemou, J.-E.; Chakouri, M.; Jellouli, A. Combination of Sentinel-2 Satellite Images and Meteorological Data for Crop Water Requirements Estimation in Intensive. Agriculture 2022, 12, 1168. [Google Scholar] [CrossRef]
  52. Dong, X.; Hu, S.; Yuan, Q.; Kou, Y.; Li, S.; Deng, W.; Ren, P. Research on Vegetation Ecological Security in Arid Region Mountain Front River Valleys Based on Ecological Water Consumption and Water Demand. Land 2023, 12, 1642. [Google Scholar] [CrossRef]
  53. Chen, A.; Wu, M. Practice on ecological flow and adaptive management of hydropower engineering projects in China from 2001 to 2015. Water Policy 2018, 20, 336–354. [Google Scholar]
  54. Li, Q.; Wang, Q.; Chen, H.; Qin, Y.; Zhang, M.; Liu, S. Progress and perspectives on ecological flow assessment methods in China. Chin. J. Ecol. 2024, 44, 36–46. [Google Scholar]
  55. Yue, S.; Li, H.; Cheng, B.; Gao, Z. The value of environmental base flow in water-scarce basins: A case study of Wei River Basin, Northwest China. Water 2018, 10, 848. [Google Scholar] [CrossRef]
  56. Li, Y.; Liu, X.; Li, P.; Chen, L.; Zhu, Q.; Wang, T. Analysis of ecological water requirement of rivers for different purposes: The case of Liuli River. J. Ecol. 2018, 38, 4393–4403. [Google Scholar]
  57. Yao, J.; Wang, G.; Yu, R.; Su, J.; A, Y.; Zhang, X.; Wang, L.; Fang, Q. Investigating the regional ecological environment stability and its feedback effect on interference using a novel vegetation resilience assessment model. Sci. Total Environ. 2024, 930, 172728. [Google Scholar] [CrossRef]
Forests 15 01749 g001
Figure 1. Diagram of the area where the study area is located (China in the upper left corner, Xinjiang Uygur Autonomous Region in the lower left corner, and the Kokyar River Basin on the right).
Figure 1. Diagram of the area where the study area is located (China in the upper left corner, Xinjiang Uygur Autonomous Region in the lower left corner, and the Kokyar River Basin on the right).
Forests 15 01749 g001
Forests 15 01749 g002
Figure 2. Vegetation pictures created based on the field survey observations (September 2020).
Figure 2. Vegetation pictures created based on the field survey observations (September 2020).
Forests 15 01749 g002
Forests 15 01749 g003
Figure 3. Remote sensing image of the Kokyar River Basin from the Sentinel-2 satellite.
Figure 3. Remote sensing image of the Kokyar River Basin from the Sentinel-2 satellite.
Forests 15 01749 g003
Forests 15 01749 g004
Figure 4. Data preprocessing results. (a) B4, B3, and B2 true color. (b) B4, B3, and B2 false color. (c) 2 km buffer zone clipping.
Figure 4. Data preprocessing results. (a) B4, B3, and B2 true color. (b) B4, B3, and B2 false color. (c) 2 km buffer zone clipping.
Forests 15 01749 g004
Forests 15 01749 g005
Figure 5. (a) Merging of images of the study area. (b) Segmentation of images of the study area.
Figure 5. (a) Merging of images of the study area. (b) Segmentation of images of the study area.
Forests 15 01749 g005
Forests 15 01749 g006
Figure 6. Research workflow.
Figure 6. Research workflow.
Forests 15 01749 g006
Forests 15 01749 g007
Figure 7. Proportion of land use in the 0–2 km area of the Kokyar basin.
Figure 7. Proportion of land use in the 0–2 km area of the Kokyar basin.
Forests 15 01749 g007
Forests 15 01749 g008
Figure 8. Vegetation coverage.
Figure 8. Vegetation coverage.
Forests 15 01749 g008
Forests 15 01749 g009
Figure 9. Distribution of vegetation coverage on both banks of the Kokyar River at 0–1 km, 1–2 km, and 0–2 km.
Figure 9. Distribution of vegetation coverage on both banks of the Kokyar River at 0–1 km, 1–2 km, and 0–2 km.
Forests 15 01749 g009
Forests 15 01749 g010
Figure 10. Misclassified features.
Figure 10. Misclassified features.
Forests 15 01749 g010
Forests 15 01749 g011
Figure 11. Map showing the distribution of the verification points.
Figure 11. Map showing the distribution of the verification points.
Forests 15 01749 g011
Forests 15 01749 g012
Figure 12. The amount of water discharged from the Kokyar River in 2020.
Figure 12. The amount of water discharged from the Kokyar River in 2020.
Forests 15 01749 g012
Table 1. Land use classification system for the Kokyar River Basin.
Table 1. Land use classification system for the Kokyar River Basin.
CategoryInterpretation of Ground ObjectsInterpretation Signs
GrasslandIrregular plots dominated by growing herbs and frequently uneven in shape; marked in light red.Forests 15 01749 i001
ShrublandAn area consisting of wild or cultivated shrubs, characterized by its large area; marked in bright red.Forests 15 01749 i002
WoodlandA forest belt or forest composed of arbor species, characterized by its large area; marked in deep red.Forests 15 01749 i003
PlantationsA forest belt formed by artificial construction, with a regular shape and large area; marked in light red.Forests 15 01749 i004
Water bodiesThe land occupied by water bodies with a specified use; marked in blue.Forests 15 01749 i005
Residential areaA place where people gather and settle characterized by regular shapes and clusters of plotsForests 15 01749 i006
RoadsA strip of land for the passage of trackless vehicles and pedestriansForests 15 01749 i007
Table 2. Regularity of plant growth period and crop coefficient of each vegetation type.
Table 2. Regularity of plant growth period and crop coefficient of each vegetation type.
Vegetation TypeEarly Growth
(April)		Growth and Development Period
(May–June)		Mid-Growth
(July–September)		Late Growth
(October)
High grassland0.230.440.550.45
Medium-cover grass0.150.200.300.25
Low-cover grass0.110.180.250.15
Shrub land0.190.330.580.60
Arbor forest land0.200.520.910.78
Artificial forest land0.531.041.130.97
Table 3. Accuracy evaluation of land use information extraction results for the Kokyar River Basin.
Table 3. Accuracy evaluation of land use information extraction results for the Kokyar River Basin.
Land Cover TypeMapping Accuracy (Pixels)User Accuracy (Pixels)Mapping Accuracy (%)User Accuracy
(%)		Overall Accuracy (%)Kappa Coefficient
Grassland18/2818/2364.2978.2686.740.84
Shrub land94/11294/12983.9372.87
Arbor forest land47/10047/4847.0097.92
Artificial forest land119/139119/16785.6171.26
Waters area62/6862/6891.1891.18
Residential area375/380375/41598.6890.36
Cultivated land74/8274/8090.2492.50
Roads212/245212/22486.5394.64
Table 4. Statistics of land cover/land use types and areas in the Kokyar River Basin (Unit: km2).
Table 4. Statistics of land cover/land use types and areas in the Kokyar River Basin (Unit: km2).
Land Cover 0–1 km1–2 km0–2 kmProportion (%)
Grassland0.180.000.180.25
Shrub land0.610.050.660.88
Arbor forest land0.090.000.090.13
Artificial forest land1.530.011.542.03
Waters area1.220.031.261.67
Residential area32.7038.8071.5193.76
Cultivated land0.210.590.801.05
Roads0.030.130.160.23
Cultivated land0.000.000.000.00
Table 5. Vegetation coverage area within 0–2 km of both banks of the Kokyar River (unit: km2).
Table 5. Vegetation coverage area within 0–2 km of both banks of the Kokyar River (unit: km2).
Serial NumberCategoryCoverage
> 75%		Coverage
60%~75%		Coverage 45%~60%Coverage
< 45%		Area
(km2)
1Grassland0.0934210.0173880.0167160.0527850.114133
2Shrub land0.2162410.0522830.0555550.3336610.328742
3Arbor forest land0.051590.0077980.0072860.0229940.059404
4Artificial forest land1.1666090.0874480.0733440.2120081.166018
Table 6. Ecological water quota for oasis artificial forests (unit: m3/hm2).
Table 6. Ecological water quota for oasis artificial forests (unit: m3/hm2).
TypeProtective ForestTimber ForestFuelwoodGarden Forest
Tree speciesArrowhead poplar Arrowhead poplarArrowhead poplarApple trees, etc.
Irrigation quota (m3)3945573057305250
Table 7. Water demand of vegetation within a range of 0–2 km on both banks of the Kokyar River (unit: m3).
Table 7. Water demand of vegetation within a range of 0–2 km on both banks of the Kokyar River (unit: m3).
NumberTypeArea (hm2)Water Demand (m3)
1Grassland11.41331.6778 × 104 m3
2Shrub land32.87421.1958 × 105 m3
3Arbor forest land5.94041.1883 × 104 m3
4Artificial forest land116.60184.5999 × 105 m3
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

Liu, X.; Alifujiang, Y.; Abliz, A.; Asaiduli, H.; Ye, P.; Nurahmat, B. Estimation of Water Demand for Riparian Forest Vegetation Based on Sentinel-2 Data: A Case Study of the Kokyar River Basin. Forests 2024, 15, 1749. https://doi.org/10.3390/f15101749

AMA Style

Liu X, Alifujiang Y, Abliz A, Asaiduli H, Ye P, Nurahmat B. Estimation of Water Demand for Riparian Forest Vegetation Based on Sentinel-2 Data: A Case Study of the Kokyar River Basin. Forests. 2024; 15(10):1749. https://doi.org/10.3390/f15101749

Chicago/Turabian Style

Liu, Xianhe, Yilinuer Alifujiang, Abdugheni Abliz, Halidan Asaiduli, Panqing Ye, and Buasi Nurahmat. 2024. "Estimation of Water Demand for Riparian Forest Vegetation Based on Sentinel-2 Data: A Case Study of the Kokyar River Basin" Forests 15, no. 10: 1749. https://doi.org/10.3390/f15101749

APA Style

Liu, X., Alifujiang, Y., Abliz, A., Asaiduli, H., Ye, P., & Nurahmat, B. (2024). Estimation of Water Demand for Riparian Forest Vegetation Based on Sentinel-2 Data: A Case Study of the Kokyar River Basin. Forests, 15(10), 1749. https://doi.org/10.3390/f15101749

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