Determination of the Suitable Lake Surface Area of Typical Terminal Lakes in Arid Regions

Abstract

The continuous depletion of global groundwater resources has posed a serious threat to the ecological stability of terminal lakes in arid regions. However, accurate ecological assessment and water resource management of these lakes face a long-term key bottleneck—the determination of an appropriate lake surface area. Previous research has primarily focused on identifying the minimum interannual suitable lake surface area, with limited exploration of the suitable area range for lakes experiencing significant annual surface area fluctuations. Taitema Lake is located in the southeastern Tarim Basin of arid northwest China and serves as the terminal lake for both the Tarim and Cherchen Rivers. This study examines Taitema Lake, a continental terminal lake in an arid region. We developed a comprehensive ecological security evaluation system based on landscape structure, steady-state conditions, and habitat elements to establish the minimum suitable lake surface area threshold. By combining this with the threshold for maximum suitable lake surface area—when ecological water use efficiency peaks—we determined the interannual suitable lake surface area for Taitema Lake to be 33.7–154.4 km². This study employed the MIKE 11 one-dimensional hydrodynamic model. Within the constraints of the lake surface area range determined by ecological water demand, we propose ecological dispatching plans for specific periods. During the green-up period (April to May), water is alternately transferred through either the Wenkuoer River or the old Tarim River at a flow rate of 24 m³/s, with a total conveyance volume of 1.3 × 10⁸ m³. For the sowing period (August to October), a dual-channel approach is used where both rivers transport water simultaneously at 27 m³/s each, resulting in a total conveyance volume of 4.3 × 10⁸ m³. This study offers valuable insights, supported by multi-scale models, for optimizing water resource allocation and ecological protection of lakes in arid areas.

1. Introduction

Terminal lakes in arid regions serve as crucial water sources, vital for maintaining biodiversity and ecological balance, and hold significant ecological, economic, and social value. Research on these lakes provides a scientific foundation for addressing global climate change, promoting environmental protection, and achieving sustainable development [1]. However, these areas face ecological challenges, such as declining lake water levels, deteriorating water quality, and reduced biodiversity, due to global climate change and intense human activities [2]. Persistent hydrological stress leads to the eventual desiccation of the lake, particularly in terminal lake areas of arid regions. Current studies primarily examine factors affecting lake sediments [3], precise protection of lake vegetation [4], salinization of lakes, optimal allocation of lake water resources [5], greenhouse gas emissions from lakes [6], and assessments of suitable water surface conditions [7]. The scientific planning of lake surface areas promotes the rational use of limited water resources, preventing over-exploitation or waste. Furthermore, it helps protect the ecological environment, regulate local climates, improve agricultural irrigation, and mitigate desertification [8]. This approach ensures the sustainable use of water resources, supports long-term economic and social stability, and enhances the environmental adaptability of arid regions.

In recent years, significant progress has been made in researching suitable water surface areas in arid regions [9,10]. The focus has shifted from merely determining the minimum lake surface area to understanding the dual-threshold mechanism that includes both minimum and maximum suitable lake surface areas. For example, studies on lakes in arid regions have established an ecological security assessment system and ecological water use efficiency indicators, proposing lower and upper limits for lake surface areas [11,12]. Additionally, by analyzing the relationship between the water level and area, as well as the vegetation coverage rate and ecological water level, researchers have suggested appropriate lake surface areas for different seasons [13,14]. The assessment system now includes factors such as water resources, environment, ecology, morphology, hydrology, and biology to dynamically adjust lake areas, balancing ecological security and social services [15,16]. However, the existing research has limitations: most evaluation systems lack multi-dimensional comprehensive evaluations and precise analyses of key ecological water use periods throughout the year [17]. Moreover, previous ecological dispatching schemes have remained primarily theoretical. They often lacked quantitative regulation via hydrodynamic models and did not systematically link suitable water surface areas with conveyance schemes. These physical and climatic characteristics make Taitema Lake’s surface area highly sensitive to changes in water recharge, highlighting the necessity of maintaining an appropriate lake surface area through scientific regulation.

This study innovatively proposes a systematic methodology that integrates multi-dimensional ecological security assessment, ecological water-use efficiency analysis, and hydrodynamic modeling. Firstly, this study addresses the shortcomings of previous research by developing a comprehensive ecological security evaluation system for Taitema Lake’s suitable lake surface area. The system encompasses eight indicators across three dimensions: landscape structure, steady-state conditions, and habitat elements. By integrating the ecological water use benefits calculated using the improved ESV, we establish the threshold for the lake’s interannual suitable water surface area. Additionally, employing a one-dimensional hydrodynamic model, we determine water surface regulation targets and propose ecological dispatching schemes for the greening-back and sowing periods. These schemes are designed to meet the dual objectives of efficient water resource utilization and ecological protection, considering the suitable lake surface area during the water conveyance period under ecological water demand constraints. This study aims to offer a multi-scale and model-supported scientific foundation for managing water resources and restoring ecosystems in arid inland lakes.

2. Materials and Methods

2.1. Research Area

The Tarim River Basin (34.20–43.39° N, 71.39–93.45° E) is situated in southern Xinjiang, encompassing the Taklimakan Desert, with a drainage area of about 1.02 million square kilometers. This study focuses on the region from the Daxihaizi Reservoir to Taitema Lake in the lower reaches of the Tarim River (Figure 1). Taitema Lake, located in the southeast of the Tarim Basin, serves as the terminal lake for both the Tarim and Cherchen Rivers. It features a warm temperate continental desert climate. The lake’s terrain is gentle, with an average surface altitude of approximately 805 m. Characterized as a typical shallow lake, its bed is expansive, and the water is shallow. The lake forms in a low-lying area where the Altun Mountains’ plain meets the alluvial plains of the Tarim and Cherchen Rivers, primarily receiving surface recharge from ice and snow meltwater and groundwater. The seasonal and interannual fluctuations in snow meltwater and groundwater recharge are the primary natural drivers of lake surface changes, and their variation patterns form the scientific foundation for determining the appropriate threshold of the lake surface area. This high-altitude area experiences intense evaporation and frequent sandstorms. The annual average temperature is 11.5 °C, with an annual precipitation of just 28.5 mm, while the annual evaporation reaches 2920 mm. These geographical and hydrological characteristics collectively determine the vulnerability of Taitema Lake’s ecosystem, making its water balance highly susceptible to climate fluctuations and human activities. Therefore, maintaining an appropriate lake surface area through scientific water regulation is essential for ensuring the ecological security of this terminal lake.

Taitema Lake once dried up due to excessive water resource exploitation in its upper reaches [18,19,20]. After 2000, the Tarim River Basin Ecological Water Conveyance Project was initiated, conducting 24 ecological water conveyance operations, 18 of which reached Taitema Lake. As water conveyance progressed, the lake’s water area gradually recovered, partially restoring the lake area’s ecosystem. Developing an ecological water conveyance plan based on an appropriate lake surface area is crucial for ensuring the basin’s ecological security and promoting its sustainable economic development.

2.2. Data Sources

This study utilizes multi-source remote sensing data from 2000 to 2023 to examine both annual and monthly changes in lake surface area, with rigorous quality control and harmonization procedures applied to ensure consistency. The Landsat NDVI and land surface temperature (LST) datasets were derived from the harmonized Surface Reflectance Tier 1 collections (Landsat 5/7/8) on the Google Earth Engine (GEE) platform, which have undergone radiometric calibration, atmospheric correction, and cross-sensor normalization. The annual NDVI was generated using the maximum value composite method to minimize cloud and atmospheric effects, while LST data were cloud-masked and temporally matched with NDVI for TVDI calculation. For SPEI calculation, monthly temperature and precipitation data were obtained from the Tieganrik meteorological station, the closest station to Taitema Lake, located within the same arid climatic zone; while this introduces some uncertainty due to local microclimatic differences, the station provides the best available long-term record for regional drought assessment. The annual 30 m resolution land use/cover dataset of China was acquired via GEE. This published dataset, produced using a consistent classification methodology, has a known overall accuracy and ensures comparability in land type area calculations over time. The land use maps were cropped to the study area and imported into Fragstats 4.2 to compute landscape pattern indices (SHDI, PAFRAC, IJI, CONTAG, DIVISION) using an 8-neighbor rule at a consistent 30 m resolution to maintain scale invariance. In the calculation of ecological water consumption, the unit area ecological water consumption data for non-water areas were derived from the CAS/IGSNRR/PML/V2_v017 dataset (filtered on the GEE platform), while the ecological water consumption for water areas was obtained by multiplying the multi-year average evaporation depth (2920.2 mm) by the water area. This approach is based on the assumptions that unit area ecological water consumption is relatively spatially homogeneous and that the evaporation data are reliable; however, the associated uncertainties may have some impact on the final threshold determination. All spatial analyses were conducted within the GEE environment or under a uniform projection to guarantee spatio-temporal consistency across the entire study period.

2.3. Research Methods

2.3.1. Ecological Security Assessment System

The entropy weight method is utilized for assessing ecological security, objectively determining weights based on the variability of the indicator data. To examine the sensitivity of the weights, key indicators (e.g., LA, VA) were adjusted by ±10% and ±20%. The results showed that the comprehensive ecological security scores and the identification of years with “good” or higher security levels remained stable. Although alternative weighting approaches such as equal weighting and expert scoring were considered, the entropy weight method was adopted for its objectivity and its ability to effectively reflect the differential contributions of indicators to ecosystem security in arid lake regions. The assessment indicators (

Table 1. Indicators, purposes, and calculation methods for ecological security assessment.

Indicator	Purpose	Calculation Method
LA	Characterizes the annual surface area of effective lake water bodies.	On the Google Earth Engine platform, the JRC Global Surface Water dataset is used. A pixel is classified as water if it is identified as water in ≥7 months within a given year, and the annual lake area is obtained by summing all such water pixels [23].
VA	Reflects the total area covered by vegetation.	Based on the NDVI, areas with NDVI > 0.1 are considered vegetated; whereas areas with NDVI < 0.1 are regarded as non-vegetated [24]. The vegetation area (VA) is obtained by summing all pixels with NDVI > 0.1.
NDVI	Reflects vegetation cover.	The annual NDVI is derived using the maximum value composite method: NDVI images for all 12 months in a year are collected, and for each pixel, the maximum monthly NDVI value is taken as that pixel’s annual NDVI [25].
TVDI	Used to characterize surface drought conditions.	$\mathrm{TVDI} = {\displaystyle \frac{{\mathrm{T}}_{\mathrm{s}} - {\mathrm{T}}_{\mathrm{smin}}}{{\mathrm{T}}_{\mathrm{smax}} - {\mathrm{T}}_{\mathrm{smin}}}}$ ${\mathrm{T}}_{\mathrm{s}} \mathrm{is} \mathrm{the} \mathrm{land} \mathrm{surface} \mathrm{temperature}, {\mathrm{T}}_{\mathrm{smax} } \mathrm{is} \mathrm{the} \mathrm{maximum} \mathrm{land} \mathrm{surface} \mathrm{temperature} \mathrm{corresponding} \mathrm{to} \mathrm{a} \mathrm{given} \mathrm{NDVI} \mathrm{value} \mathrm{in} \mathrm{the} \mathrm{NDVI}\text{-}\mathrm{Ts} \mathrm{feature} \mathrm{space}, \mathrm{and} {\mathrm{T}}_{\mathrm{smin}}$ is the minimum land surface temperature corresponding to the same NDVI value [26].
SPEI	Reflects the degree to which drought conditions over a given period deviate from the multi-year normal.	Monthly mean precipitation, monthly mean air temperature, and geographic coordinates (latitude and longitude) are imported into Python 3.8.0. The SPEI is then calculated using standard SPEI calculation codes to obtain time series of drought indices.
EQ	Characterizes the overall health status of an ecosystem.	$\mathrm{EQ} = {\displaystyle \frac{\mathrm{Abio}\times \left({0.21 \times \mathrm{A}}_{\mathrm{G} }+ 0.28 \times {\mathrm{A}}_{\mathrm{w}} + 0.04 \times {\mathrm{A}}_{\mathrm{U}} + 0.01 \times {\mathrm{A}}_{\mathrm{N}}\right)}{\mathrm{A}}}$, $\mathrm{where} {\mathrm{A}}_{\mathrm{G}}$$\mathrm{is} \mathrm{grassland} \mathrm{area}, {\mathrm{A}}_{\mathrm{w} }$$\mathrm{is} \mathrm{water} \mathrm{area}, {\mathrm{A}}_{\mathrm{U} }$$\mathrm{is} \mathrm{construction} \mathrm{land} \mathrm{area}, {\mathrm{A}}_{\mathrm{N}}$ is unused land area, A is the total area of the study region, and Abio is the normalization coefficient of the habitat quality index (458.5 for the Taitema Lake region) [27,28].
LH	Evaluates landscape diversity.	$\mathrm{LH} = 0.7 \times \mathrm{SHDI} + 0.3 \times \mathrm{PAFR}\_\mathrm{AC}$, where SHDI is the Shannon diversity index, and PAFR_AC is the area-weighted mean patch fractal dimension index [29,30].
LC	Evaluates landscape connectivity.	$\mathrm{L}\mathrm{C}=0.4\times \mathrm{I}\mathrm{J}\mathrm{I}+0.4\times \mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{A}\mathrm{G}+0.2\times \mathrm{D}\mathrm{I}\mathrm{V}\mathrm{I}\mathrm{S}\mathrm{I}\mathrm{O}\mathrm{N}$, where IJI is the interspersion and juxtaposition index, CONTAG is the contagion index, and DIVISION is the landscape division index [31,32,33].

) are categorized into three aspects: landscape structure, steady-state conditions, and habitat elements. Indicators are assigned weights based on their contribution to ecosystem security. To obtain the comprehensive score for each indicator, the weights are multiplied by the standardized values and then summed [21] (

Table 2. Weights of ecological security evaluation indicators.

Primary Indicator	Secondary Indicator	Weight
Landscape Structure	Landscape Heterogeneity (LH)	0.097
	Landscape Connectivity (LD)	0.083
Steady-State Condition	Lake Area (LA)	0.234
	Vegetation Area (VA)	0.209
	Normalized Difference Vegetation Index (NDVI)	0.112
	Ecological Quality (EQ)	0.106
Habitat Elements	Temperature Vegetation Dryness Index (TVDI)	0.085
	Standardized Precipitation Evapotranspiration Index (SPEI)	0.074

). A higher score indicates a more optimal structure and configuration of ecosystem elements, leading to a higher ecosystem quality and safety. Ecological security levels are classified using the natural breakpoint method as excellent (score > 0.58), good (0.4 < score < 0.58), medium (0.28 < score < 0.4), poor (0.15 < score < 0.28), and very poor (score < 0.15) [22]. The minimum suitable lake surface area threshold is determined based on a lake surface achieving a comprehensive score of good or above. The “good” level (score > 0.4) was chosen as the minimum threshold, balancing ecological protection and water resource feasibility. This level indicates a fundamentally healthy and stable ecosystem. Adopting a lower “medium” level could risk ecosystem degradation, while pursuing a higher “excellent” level might exceed the practical water resource capacity in arid regions, making it unsustainable.

2.3.2. Ecological Water Use Benefits

Through statistical analysis of data including the average price, per-unit yield, and sown area of grain crops in the lower reaches of the Tarim River from 1990 to 2019, the single ESV equivalent factor in this region was determined to be 1881.82 CNY·hm⁻². The equivalent factor is estimated based on regional economic data, assuming spatially uniform ecosystem service values. It overlooks vegetation heterogeneity, is influenced by fluctuations in grain prices, and may undervalue non-market services such as sand fixation. These associated uncertainties affect the absolute values of ESV but have limited impact on the relative threshold determination in this study. Based on this factor, the coefficient table for the service value per unit area of the ecosystem in the lower reaches of the Tarim River (

Table 3. Coefficient of service value per unit area of the ecosystem in the lower reaches of the Tarim River.

Primary Category	Secondary Category	Grassland	Water Body	Unused Land	Construction Land
Supply Services	Food Production	1505.46	0	0	0
	Raw Material Production	1693.64	865.64	0	0
Regulation Services	Gas Regulation	1505.46	38,351.50	56.45	0
	Climate Regulation	3669.55	18.82	37.64	0
	Hydrological Regulation	2465.18	34,211.50	18.82	0
	Waste Treatment	2051.18	4685.73	639.82	0
Supporting Services	Soil Retention	564.55	188.18	18.32	0
	Biodiversity Maintenance	94.09	18.82	0	0
Cultural Services	Aesthetic Landscape Provision	75.27	8167.1	18.82	82.6
Total		13,624.38	86,507.29	790.36	82.6

) was developed.

Three driving factors—VC, HDI, and TVDI—are introduced, with the TVDI calculation method aligning with that used in the ecological security assessment system. HDI, in particular, quantifies the intensity of human activity interference on various land use components within a region. Equation (1) defines the Human Disturbance Intensity (HDI) [34].

$$HDI={\sum }_{k=1}^m{A}_k{P}_k /A$$

(1)

In the formula, m denotes the number of land use types, while A_k represents the total area of land use type k (in km²). The human disturbance intensity parameter for land use type k is indicated by P_k, with specific values assigned as follows: grassland is 0.24, water area is 0.12, unused land is 0.08, and construction land is 0. The total area of land use type is denoted by A (in km²). The Enhanced Vegetation Index (EVI) ranges from −1.0 to 1.0. The parameters carry some uncertainty, though the multi-index comprehensive evaluation has mitigated the sensitivity of the overall results to any individual parameter. Equation (2) defines Vegetation Coverage (VC) [35].

$$VC={\displaystyle \frac{EVI-{EVI}_s}{{EVI}_v-{EVI}_s}}$$

(2)

In this context, VC denotes vegetation coverage, EVI_S refers to the EVI value of bare land, and EVI_v indicates the EVI value of pure vegetation or the maximum EVI value. Equation (3) defines the total Ecosystem Service Value (ESV).

$$ESV=(\sum A_k\times {CV}_k)\times {\lambda }_i\times D_i$$

(3)

In the formula, ESV denotes the total value of ecological services. The term A_k refers to the area of the KTH land use type, while CV_k represents the value coefficient of ecosystem services for various KTH land use types. The symbol λ_i indicates the weight value of the i-th driving factor, and D_i signifies the value of the i-th driving factor.

Ecological water utilization efficiency (EWUE) is defined as the ratio of the total ecological service value of a lake area to its total ecological water consumption, measured in CNY/m³ [36]. The EWUE calculation inherits uncertainties from both ESV and ET estimates, which affect its absolute value. However, since the study relies on relative EWUE trends rather than absolute values, these uncertainties have limited impact on the identification of optimal lake area thresholds. Equation (4) defines the Ecological Water Use Efficiency (EWUE).

$$EWUE={\displaystyle \frac{ESV}{ET}}$$

(4)

In the formula, EWUE denotes the benefits of ecological water use, ESV signifies the value of ecological services, and ET indicates the ecological water consumption.

2.3.3. MIKE 11 Hydrodynamic Model

The MIKE 11 hydrodynamic model is a one-dimensional non-constant flow numerical simulation tool widely used for simulating hydrodynamic processes in rivers, channels, and related hydrological systems [37]. Its core relies on solving the Saint-Venant equations to describe fluid movement in rivers or channels, focusing on dynamic changes in water level and flow rate [38]. This study’s model is constructed using measured flow data from the Aragan and Kurgan sections, followed by model parameter calibration. In the simulation, minor tributaries with negligible impact on results are ignored. The river is simplified as the confluence of the old Tarim River and the Wenkuoer River above the Alagan hydrological section, with its lower reaches connecting to Taitema Lake. For sections with sparse cross sections, the MIKE 11 model’s cross-section interpolation module generates intermediate cross sections. Where measured sections are entirely absent, standard sections are uniformly set based on river regulation project data. Neglecting minor tributaries may slightly underestimate the total inflow to the lake, while using standard cross sections in river reaches lacking measured data may diminish the influence of local topography on water level and flow velocity, thereby reducing the local precision of the simulation in detail. However, these systematic simplification errors have been partially compensated for through parameter adjustment during model calibration and validation. Moreover, the simulated and measured data at key cross sections show good agreement, reliably supporting the analysis of water conveyance-focused dispatch schemes. The model’s upper boundary is the daily discharge flow from the Daxihaizi Reservoir gate, while the lower boundary is the water level at the Kurgan section. Key parameters include the riverbed roughness and the permeability coefficient. To ensure a smooth start-up, the initial water level and flow rate are set to 0.001 m and 0.01 m³/s, respectively.

The data collected from the Aragan and Kugan sections between 29 July 2021 and 22 October 2021 were chosen to verify the model parameters. This period is representative of key water conveyance periods but may not encompass all hydrological year types. Since the focus of this study is on assessing the feasibility of dispatch schemes rather than long-term prediction, the limitations of this calibration period have limited impact on the core findings. For evaluation, the Nash Efficiency coefficient (NSE) and the determination coefficient (R²) were employed [39]. Generally accepted benchmarks indicate that an NSE value above 0.75 reflects good model performance, while an R² value above 0.80 is commonly used as a reference threshold for strong correlation. The formulas used are as follows:

$$\mathrm{N}\mathrm{S}\mathrm{E}=1-{\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{Q}}_{\mathrm{o},\mathrm{i}}-{\mathrm{Q}}_{\mathrm{s},\mathrm{i}})}{\sum _{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{Q}}_{\mathrm{o},\mathrm{i}}-{\mathrm{Q}}_0)}},$$

(5)

$${\mathrm{R}}^2={\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{Q}}_{\mathrm{o},\mathrm{i}}-{\mathrm{Q}}_{\mathrm{o}})({\mathrm{Q}}_{\mathrm{s},\mathrm{i}}-{\mathrm{Q}}_{\mathrm{o}})}{\sqrt{\sum _{\mathrm{i}=1}^{\mathrm{n}}{({\mathrm{Q}}_{\mathrm{o},\mathrm{i}}-{\mathrm{Q}}_{\mathrm{o}})}^2}\sum _{\mathrm{i}=1}^{\mathrm{n}}{({\mathrm{Q}}_{\mathrm{s},\mathrm{i}}-{\mathrm{Q}}_{\mathrm{s}})}^2}}.$$

(6)

In the formulas, ${\mathrm{Q}}_{\mathrm{o},\mathrm{i}}$ denotes the measured value at the i-th time step, ${\mathrm{Q}}_{\mathrm{o},\mathrm{i}}$ signifies the simulated value at the i-th time step, and ${\mathrm{Q}}_0$ represents the table. The mean of the measured values is denoted as ${\mathrm{Q}}_{\mathrm{s}}$, representing the mean of the simulated values, with n indicating the total number of time steps. At the Alagan section, both the NSE and R² are 0.91, whereas at the Kugan section, they are 0.81 and 0.8, respectively. These results suggest that, within a certain error margin, the MIKE model is effective for simulating the ecological water conveyance process in the lower reaches of the Tarim River.

3. Results

3.1. Spatio-Temporal Variation Characteristics of the Surface of Taitema Lake

From 2000 to 2023, the annual surface area of the effective water body in Taitema Lake exhibited an overall trend characterized by an initial expansion followed by contraction (Figure 2a). An expansion phase was observed from the early 2000s to 2017, with the peak area (243.4 km²) recorded in 2017. Subsequently, the lake surface entered a fluctuating contraction phase, decreasing to 62.6 km² by 2023. Notable peaks occurred during this period, such as in 2003 (96 km²) and 2012 (144.8 km²), while notable lows were observed in 2009 (3 km²) and 2014 (20.53 km²).

On a monthly scale, the lake surface area showed a seasonal “increase–decrease–increase” dynamic (Figure 2b). The area typically reached its first annual peak in March (173.7 km²), declined to its lowest point in June (84.4 km²), and rose to a second peak in December (154.9 km²). The period of lowest area coincided with the summer high-temperature season, while the recovery from autumn to early winter corresponded with the timing of ecological water conveyance operations during those months.

Between 2000 and 2023, Taitema Lake’s water body consistently expanded, beginning in the upper left corner of the lake area, moving toward the central region, and ultimately reaching the lower right corner (Figure 3). Concurrently, grasslands extended from the upper right corner to the lake’s center, with scattered grasslands eventually emerging in the lower part of the area. Throughout this period, bare land steadily converted into water bodies and grasslands, indicating an overall expansion of Lake Taitema’s surface. This process represents a long-term net expansion.

3.2. Determination of the Appropriate Surface Area of Lake Taitema

Between 2000 and 2023, the vegetation coverage expanded significantly from 1.6 km² to 93.3 km² (Figure 4a). As the vegetation area and ecological conditions improved, the EQ rose from 0.047 to a peak of 0.268 in 2019. Similarly, the NDVI value climbed steadily from 0.0006, reaching a high of 0.0598 in 2021. This growth suggests that the Taitema Lake ecosystem is becoming richer in both species variety and quantity (Figure 4b). The LH increased from 0.41 to 1.04, and the LC rose from 0.44 to 0.69 (Figure 4c). These increases indicate improved connectivity and diversity in the landscape patches of Taitema Lake. From 2000 to 2023, the water area of Lake Taitema generally trended upward, alleviating drought conditions. The SPEI and TVDI displayed a fluctuating decrease, shifting from severe to mild drought (Figure 4d). Among the indicators of the entropy weight method, the lake area and vegetation area were the two most significant indicators driving the increase in the entropy-weighted score. Their sustained growth directly dominated the long-term improvement trend of the comprehensive score.

From 2000 to 2023, the entropy weight method evaluation score for Lake Taitma exhibited an upward trend, rising from 0.1 in 2000 to 0.78 in 2019, before declining to 0.64 in 2023 (Figure 5a). Using 2011 as a dividing point, the ecological security level from 2000 to 2011 was consistently medium or below (score < 0.4). Post-2011, the level fluctuated between good and excellent (score > 0.4). Among the years with good and excellent results, 2014, 2015, and 2021 had the smallest lake surface areas. The average of these years established the minimum lake surface area threshold at 33.69 km². These three years were selected as a statistically minimal-area sample, aiming to identify the lower limit of water body size required to maintain ecological security, which aligns with the principle of efficient water use in arid regions. If other years meeting the “good” criterion were used, their larger areas would raise the threshold, and the suitable range is inaccurate. While increasing the lake surface area enhances the ecological security, an excessively large area can lead to higher evaporation and water resource wastage. Defining the optimal maximum threshold for lake area based on ecological water utilization efficiency is therefore necessary.

From 2000 to 2023, both the ESV and ET of Lake Taitema initially increased and then decreased (Figure 5b), peaking in 2019 when the ESV reached 33.4 × 10⁸ CNY, and the ET reached 37 × 10⁸ m³. These trends closely mirrored changes in the lake’s surface area. This study adopted quadratic polynomial regression to analyze the relationship between ecological water-use efficiency (EWUE) and lake surface area, with the goodness-of-fit evaluated based on the RMSE, and the results indicated a satisfactory fit (Figure 6a). The efficiency of ecological water utilization generally rose, peaking in 2020. During the early phase of lake surface expansion, this efficiency improved, reaching its highest value at a lake surface area of 154.4 km² (Figure 6b). Beyond this point, the efficiency began to decline, suggesting that further increases in lake area do not necessarily enhance the ecological water utilization efficiency.

Considering parameters like ecological security and water use efficiency comprehensively, the suitable threshold for Taitma Lake is 33.7 km² to 154.4 km². However, this range only reflects the appropriate lake surface area across different years and does not specify the suitable surface area within a single year. Thus, determining the final appropriate lake surface area range requires integrating ecological water conveyance.

3.3. Determination of the Water Supply Plan for the Lower Reaches of the Tarim River

The Daxihaizi Reservoir, located in the lower reaches of the Tarim River, serves as the primary water source for replenishing Taitema Lake. The relationship between the lake’s surface area and its water inflow (Figure 7) was derived from monthly data in 2021 using linear regression. Inflow data came from the measured discharge at downstream hydrological sections, and lake area data were obtained from concurrent Landsat imagery. The analysis assumes a monthly water balance, with channel inflow as the dominant factor, along with water inflow and its surface area. Within the long-term hydrological context (2000–2023), 2021 can be considered a near-average hydrological year. While the use of a single year’s data inherently limits the ability to reflect extreme wet or dry conditions, the relationship derived here is intended not for precise long-term hydrological prediction but rather to establish a practical reference for water-conveyance dispatch planning. Water delivery during the green-up period (April–May) and the sowing period (August–October) supports vegetation germination and sowing in the lake area, promoting the growth of natural vegetation. The historical data on the lake surface area relative to the ecological water demand suggest that the surface area should exceed 40 km² during the green-up period and 120 km² during the sowing period. They are approximate reference targets rather than absolute optimal values. Consequently, water inflow into the lake should be approximately 8 × 10⁶ m³ from April to May and 65 × 10⁶ m³ from August to October.

Using the established MIKE 11 model, the drainage flow of the old Tarim River and the Wenkuoer River was adjusted to simulate the water volume through the Kurgan section. Four key sections—Yingsu, Kardayi, Alagan, and Kouran—were chosen to model water dissipation during the greening-up period (Figure 8). The Daxihaizi Reservoir transferred water to the Wenkuoer River at 24 m³/s, delivering a final volume of 8.01 × 10⁶ m³ to the Kurgan section. For the seed sowing period, six sections—Laoyingsu, Bozikule, Yingsu, Kardayi, Alagan, and Kurgan—were selected to simulate water dissipation (Figure 9). The Daxihaizi Reservoir transferred water to both the Wenkuoer and old Tarim Rivers at 27 m³/s. After convergence, a total volume of 64 × 10⁶ m³ was delivered to the Kurgan section. The water volume at the Kurgan section can approximate the inflow to Lake Taitema. A specific ecological dispatching plan was developed by integrating the lake’s inflow needs during the green-up and sowing periods. From April to May, single-channel water conveyance is used, with the Wenkuoer River flowing at 24 m³/s and a total ecological water conveyance of 130 × 10⁶ m³. From August to October, a dual-channel system is employed, maintaining both the Wenkuoer and old Tarim Rivers at 27 m³/s, with a total ecological conveyance of 430 × 10⁶ m³.

4. Discussion

4.1. New Ideas for Determining the Appropriate Lake Surface Area

Compared to most previous studies that primarily determined a static lake surface area based on interannual scales and a single minimum threshold (e.g., using energy value theory to estimate the area of Dongjuyan Lake [40]), this study achieves two key methodological advances:

First, in terms of the assessment dimension, earlier approaches often overlooked the pronounced intra-annual fluctuations of terminal lakes in arid regions due to insufficient temporal resolution [41,42,43]. By integrating ecological security and ecological water-use efficiency, this study defines a dual-threshold interannual suitable range for the lake (33.7–154.4 km²). This range clearly indicates that an excessively small lake area can destabilize and degrade the ecosystem, while an excessively large area may lead to inefficient use of water resources.

Second, regarding the temporal scale, to address the insufficient representation of intra-annual ecological processes in earlier research, this study identifies two critical water-demand phases—the green-up period and the sowing period—based on vegetation phenology [44,45,46]. Using the area–inflow relationship, the interannual range is further disaggregated into seasonal dispatch targets. This shifts the suitable area from a static value to a process-oriented objective that adjusts dynamically with ecological demands. This approach offers a new perspective for determining the suitable surface area of terminal lakes in arid regions.

4.2. Water Supply Plans for Maintaining the Ecological Security of Lakes

From a vegetation perspective, this study identifies the green-up and sowing periods as key ecological water supply windows. Based on their respective water demands, the optimal lake surface area should exceed 40 km² (requiring 800 × 10⁶ m³ inflow) during green-up and 120 km² (requiring 650 × 10⁶ m³ inflow) during sowing. Using these inflow–area relationships, the required river discharge was inversely derived and simulated with the MIKE 11 model, ultimately leading to the proposal of an ecological dispatching scheme: single-channel at 24 m³/s (130 × 10⁶ m³ total) in green-up and dual-channel at 27 m³/s (430 × 10⁶ m³ total) in sowing. Compared to previous studies, this research achieves methodological integration and breakthroughs in the following aspects: First, in contrast to studies that primarily identify gaps in ecosystem service supply and demand using the “supply–flow–demand” framework [47], this research not only identifies key water-demand periods but also employs a hydrodynamic model to achieve quantitative and spatially explicit derivation from lake ecological targets to upstream river discharge operations, thereby transforming a conceptual framework into an actionable dispatching scheme. Second, compared to research that simulates and optimizes ecological water replenishment schemes using two-dimensional hydrodynamic–water quality coupled models [48], this study focuses on long-term temporal scales and water conveyance processes. By utilizing the one-dimensional MIKE 11 model to efficiently simulate water transmission losses and routing in river channels, it is better suited for ecological scheduling scenarios in arid river corridors, where longitudinal flow dominates, and long-distance water allocation is required. Third, relative to studies examining the impacts of climate change and human activities on lake baseflow, water replenishment, and pollution loads [49,50,51], this research goes beyond analyzing external drivers by further coupling vegetation phenological water-demand rhythms with hydrodynamic processes. It specifies concrete flow and water volume thresholds under different ecological objectives (green-up support, sowing security), ensuring that the dispatching scheme is both ecologically targeted and engineering-feasible.

Given uncertainties in data, models, and future climate variability, the proposed water supply plan should be regarded as an indicative scheme rather than a fixed optimal solution. It provides a quantitative reference framework for ecological dispatching in arid-region lakes, integrating multi-source data and model coupling. However, its implementation should be dynamically adapted based on real-time hydrological monitoring and ecosystem feedback to accommodate long-term environmental changes and management needs. This scheme enhances watershed ecological security and functional performance, thereby transforming water regulation into measurable improvements in ecological functions.

5. Conclusions

This study establishes an appropriate surface area threshold for Taitema Lake. Derived from a comprehensive ecosystem security evaluation incorporating eight key indicators and an analysis of ecological water-use efficiency, this threshold provides a scientific basis for balancing hydrological sustainability and ecological functions in arid terminal lakes. Furthermore, by linking this area range to lake water inflow and coupling it with the MIKE 11 hydrodynamic model, the research develops an ecological water-conveyance dispatching scheme from the lower Tarim River to the lake. The scheme specifies seasonal river discharge rates and total water delivery volumes, thereby translating the theoretical area target into a practical management strategy for ecological restoration and water-resource allocation. Based on Taitema Lake’s specific conditions, this study developed a method that combines multi-indicator assessment and hydrodynamic modeling. It offers a transferable framework for similar arid-region terminal lakes. The main conclusions are as follows:

Between 2000 and 2023, Taitema Lake’s surface area showed an overall trend of increase followed by decline, peaking at 243.4 km² in 2017 after starting near dryness. Seasonally, the area peaks in March and December, with a low in June due to high evaporation. Ecological security was rated good or above in 45.8% of years, supporting a suitable area threshold of 33.7–154.4 km², beyond which water-use efficiency declines. To maintain this range, seasonal water conveyance is applied: single-channel flow at 24 m³/s (130 × 10⁶ m³ total) in April–May and dual-channel flow at 27 m³/s (430 × 10⁶ m³ total) in August–October, ensuring ecological functionality under varying hydrological conditions.

Based on Taitema Lake’s specific conditions, this study developed a method that combines multi-indicator assessment and hydrodynamic modeling. It offers a transferable framework for similar arid-region terminal lakes. This study has some limitations that should be acknowledged. First, the precision of the determined thresholds may be affected by uncertainties in remote sensing data and model parameters. Second, key water quality factors—particularly salinity dynamics, which are central to the degradation of terminal lakes in arid regions—were not explicitly included. Furthermore, the proposed water conveyance scheme is based on historical hydrological conditions, and its adaptability to future climate change requires further validation. Future research could focus on the following three points: first, integrating multi-source data with higher spatio-temporal resolution and conducting multi-model comparisons to strengthen threshold robustness; second, incorporating salinity into an integrated “area–volume–quality” assessment framework; and third, simulating different climate scenarios to support adaptive water-resource management strategies.