[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Next Article in Journal
Three-Dimensional Distribution of Arctic Aerosols Based on CALIOP Data
Previous Article in Journal
Regulatory Impacts of the Three Gorges Dam on Long-Term Terrestrial Water Storage Anomalies in the Three Gorges Reservoir Area: Insights from GRACE and Multi-Source Data
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 17
Issue 5
10.3390/rs17050902
remotesensing-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
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

A Novel Photon-Counting Laser Point Cloud Denoising Method Based on Spatial Distribution Hierarchical Clustering for Inland Lake Water Level Monitoring

by
Xin Lv
1,2,
Xiao Wang
3,*,
Xiaomeng Yang
2,4,
Junfeng Xie
2,
Fan Mo
2,
Chaopeng Xu
5 and
Fangxv Zhang
6
1
College of Geoscience and Surveying Engineering, China University of Mining & Technology—Beijing, Beijing 100083, China
2
Land Satellite Remote Sensing Application Center, Ministry of Natural Resources of China, Beijing 100048, China
3
Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, China
4
College of Geodesy and Geomatics, Shandong University of Science and Technology, Qingdao 266590, China
5
College of Resource Environment and Tourism, Capital Normal University, Beijing 100048, China
6
School of Earth Sciences and Engineering, Hohai University, Nanjing 210024, China
*
Author to whom correspondence should be addressed.
Remote Sens. 2025, 17(5), 902; https://doi.org/10.3390/rs17050902
Submission received: 9 February 2025 / Revised: 27 February 2025 / Accepted: 28 February 2025 / Published: 4 March 2025
(This article belongs to the Section Remote Sensing in Geology, Geomorphology and Hydrology)
Browse Figures
Figure 1
<p>Topographic map of Hubei Province showing the locations of the five reservoirs and five lakes, along with the ICESat-2 orbital tracks. (<b>A</b>) Danjiangkou Reservoir. (<b>B</b>) Zhanghe Reservoir. (<b>C</b>) Fushui Reservoir. (<b>D</b>) Shuibuya Reservoir. (<b>E</b>) Bailianhe Reservoir. (<b>F</b>) Honghu Lake. (<b>G</b>) Liangzi Lake. (<b>H</b>) Futou Lake. (<b>I</b>) Longgan Lake. (<b>J</b>) Daye Lake.</p> "> Figure 2
<p>Schematic diagram of ICESat-2 footprints.</p> "> Figure 3
<p>Schematic map of the location of Danjiangkou Reservoir evaporation station.</p> "> Figure 4
<p>Signal photon extraction process and water level extraction process.</p> "> Figure 5
<p>Water body mask diagram for lakes and reservoirs. (<b>A</b>) Danjiangkou Reservoir. (<b>B</b>) Zhanghe Reservoir. (<b>C</b>) Fushui Reservoir. (<b>D</b>) Shuibuya Reservoir. (<b>E</b>) Bailianhe Reservoir. (<b>F</b>) Honghu Lake. (<b>G</b>) Liangzi Lake. (<b>H</b>) Futou Lake. (<b>I</b>) Longgan Lake. (<b>J</b>) Daye Lake.</p> "> Figure 6
<p>Distribution of ATL03 photons. (<b>a</b>) Signal photon distribution. (<b>b</b>) Partial zoom of the signal photon distribution. (<b>c</b>,<b>d</b>) Difference in the distribution of signal photons and noise photons.</p> "> Figure 7
<p>Schematic diagram of the density distribution differences of photons in the horizontal and vertical directions. (<b>a</b>) Euclidean distance and vertical distance of signal photons. (<b>b</b>) Differences between Euclidean distance and vertical distance.</p> "> Figure 8
<p>A spatial distribution-based hierarchical clustering for photon-counting laser altimeter. (<b>a</b>) Photon data after coarse denoising. (<b>b</b>) Minimum spanning tree generated using Euclidean distance. (<b>c</b>) Schematic of minimum spanning tree construction based on photon density differences in the vertical direction. (<b>d</b>) Hierarchical structure generation. (<b>e</b>) Noise edge filtering using 3 standard deviations and 2 times the interquartile range. (<b>f</b>) Schematic of denoising results.</p> "> Figure 9
<p>Comparison of denoising results for the gt2r beam of ATL03_20200303195727_10400606_006_01.h5 in the Danjiangkou Reservoir. (<b>a</b>) Original signal photons. (<b>b</b>) ATL08 signal photons. (<b>c</b>) Signal photons extracted by SD-HCPLA. (<b>e</b>) Zoomed-in view of ATL08 (<b>f</b>) Zoomed-in view of SD-HCPLA.</p> "> Figure 10
<p>Comparison of denoising results for the gt1l beam of ATL03_20210909052406_11851202_006_02.h5 in the Danjiangkou Reservoir (<b>a</b>) Original signal photons. (<b>b</b>) ATL08 signal photons. (<b>c</b>) Signal photons extracted by SD-HCPLA. (<b>e</b>) Zoomed-in view of ATL08. (<b>f</b>) Zoomed-in view of SD-HCPLA.</p> "> Figure 11
<p>Trends in lake and reservoir water level changes in relation to precipitation variations.</p> "> Figure 12
<p>Trends in lake and reservoir water level changes in relation to surface temperature variations.</p> "> Figure 13
<p>Trends in lake and reservoir water level changes in relation to variations in evapotranspiration.</p> "> Figure 14
<p>Schematic diagram showing the relationship between the east–west length of the water body and the number of effective water level data days obtained.</p> ">
Versions Notes

Abstract

:
Inland lakes and reservoirs are critical components of global freshwater resources. However, traditional water level monitoring stations are costly to establish and maintain, particularly in remote areas. As an alternative, satellite altimetry has become a key tool for lake water level monitoring. Nevertheless, conventional radar altimetry techniques face accuracy limitations when monitoring small water bodies. The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2), equipped with a single-photon counting lidar system, offers enhanced precision and a smaller ground footprint, making it more suitable for small-scale water body monitoring. However, the water level data obtained from the ICESat-2 ATL13 inland water surface height product are limited in quantity, while the lake water level accuracy derived from the ATL08 product is relatively low. To overcome these challenges, this study proposes a Spatial Distribution-Based Hierarchical Clustering for Photon-Counting Laser altimeter (SD-HCPLA) for enhanced water level extraction, validated through experiments conducted at the Danjiangkou Reservoir. The proposed method first employs Landsat 8/9 imagery and the Normalized Difference Water Index (NDWI) to generate a water mask, which is then used to filter ATL03 photon data within the water body boundaries. Subsequently, a Minimum Spanning Tree (MST) is constructed by traversing all photon points, where the vertical distance between adjacent photons replaces the traditional Euclidean distance as the edge length, thereby facilitating the clustering and denoising of the point cloud data. The SD-HCPLA algorithm successfully obtained 41 days of valid water level data for the Danjiangkou Reservoir, achieving a correlation coefficient of 0.99 and an average error of 0.14 m. Compared with ATL08 and ATL13, the SD-HCPLA method yields higher data availability and improved accuracy in water level estimation. Furthermore, the proposed algorithm was applied to extract water level data for five lakes and reservoirs in Hubei Province from 2018 to 2023. The temporal variations and inter-correlations of water levels were analyzed, providing valuable insights for regional ecological environment monitoring and water resource management.

1. Introduction

Lakes are critical freshwater storage bodies on the Earth’s surface, forming an essential component of the global water cycle. Lake water levels are not only key indicators for assessing the health of lakes, but also reflect the dynamic changes in regional water resources [1]. Therefore, monitoring lake water levels provides crucial information for water resource management by relevant nations or organizations and can, to some extent, indicate the impacts of climate change. Similarly, at the same time, reservoirs play an indispensable role in water resource management [2]. They are not only used for storing freshwater, but also serve critical functions in flood control, irrigation, power generation, and water supply. By regulating reservoir water levels, it is possible to effectively mitigate flood risks, ensure water availability for agricultural irrigation, support hydropower generation, and provide drinking water for urban and rural populations [3].
In addition, reservoirs play a significant role in maintaining ecological balance and regulating regional climates. Therefore, monitoring the water level variations of lakes and reservoirs is critically important [4]. However, traditional water level monitoring primarily relies on in situ gauging stations, which require high costs for construction, maintenance, and manpower. Moreover, in remote areas, the harsh natural conditions and limited accessibility present significant challenges for the deployment and monitoring of these stations [5].
With the continuous advancement of remote sensing technology, altimetry satellites have gradually become critical tools for monitoring inland lake water levels, providing high-precision and reliable data, and serving as essential means for observing water level variations in large lakes [6]. Early altimetry satellites primarily utilized radar altimetry, including satellites such as TOPEX/Poseidon, Jason-1, ERS-1, ERS-2, CryoSat-2, and Sentinel-3. These satellites offer all-weather observation capabilities, penetrating cloud cover to enable long-term monitoring of large lakes. For example, the TOPEX/Poseidon satellite provided water level time series data for China’s Dongting Lake from 1993 to 1999 [7]; the Jason satellite series estimated water level variations for Hulun Lake from 2002 to 2017 [8]; ERS-1 was capable of monitoring lakes with surface areas greater than 500 square kilometers [9]; CryoSat-2 observed lakes with areas larger than 9 square kilometers and monitored water level changes in 200 lakes on the Tibetan Plateau from 2010 to 2019 [10]; and data from the Sentinel-3 satellite supported monthly and quasi-monthly water level variation monitoring of lakes on the Tibetan Plateau [11]. Although radar altimeter satellites have been widely used for monitoring inland lake water levels, their larger radar beam widths [12] limit their ability to provide sufficiently detailed and accurate water level data for smaller lakes and reservoirs. However, Deidda C et al. effectively monitored river section water levels using Sentinel 3A, Sentinel 3B, and Jason 2/3 satellites [13]. Meanwhile, Sentinel-6’s fully focused synthetic aperture radar (FF-SAR) data significantly improve along-track resolution, making them widely applied in river water level estimation. Shanmu Ma et al. successfully obtained water level data for six river sections with different characteristics in the middle and upper reaches of the Yangtze River using Sentinel-6 data [14], achieving centimeter-level accuracy when compared with in situ measurements. However, due to the limited surface coverage of these data, they have certain limitations in observing water levels of smaller lakes and also restrict the number of lakes that can be observed.
The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) is equipped with the Advanced Topographic Laser Altimeter System (ATLAS), a photon-counting lidar system. Compared with its predecessor, ICESat, ICESat-2 features a multi-beam design, with a smaller laser footprint and higher sampling density, enabling it to cover more inland lakes and making it suitable for monitoring smaller lakes and reservoirs. ICESat-2’s altimetry data products have been widely applied in Earth surface observations, including ice sheet elevation monitoring [15,16,17,18,19], forest canopy height measurement [20,21,22,23], coastal shallow water bathymetry [24,25,26,27], and inland lake water level monitoring [28,29,30,31,32]. The commonly used ICESat-2 altimetry products include global geolocated photon data (ATL03), land and vegetation height data (ATL08), and inland water surface height data (ATL13). Currently, lake water level measurements primarily rely on ATL08 and ATL13 [33] products for monitoring water level dynamics at regional, national, or global scales. For example, Sarah W. Cooley et al. used the Global Surface Water Occurrence (GSWO) product to filter water surface photons and applied ATL09 data on calibrated backscatter profiles and atmospheric characteristics to filter cloud photons for global lake water level monitoring [34]; Mashanmu et al. used ATL08 data and GSWO water masks to monitor water level dynamics in 473 lakes larger than 1 km2 on the Tibetan Plateau [35]; Lijuan Song et al. combined ATL13 data with Global Lakes Analysis and Discovery (GLAD) lake mask data to study the seasonal water level variations of lakes larger than 1 km2 in China, using machine learning algorithms to predict the water levels of lakes not observed by ATL13 [36]; Weixiao Han et al. monitored water level changes in Qinghai Lake using ATL13 data [37]. However, studies directly using ATL03 altimetry data to extract signal photons for dynamic lake monitoring are relatively rare, and these studies often focus on individual lakes. Additionally, ATL08 is primarily used for land and forest canopy height measurements, resulting in lower accuracy for water level monitoring, while the amount of valid water level data from ATL13 is limited. Currently, no systematic study has compared and analyzed the accuracy and frequency of ATL08, ATL13, and ATL03 for water level observations in water bodies.
Due to the low photon energy for transmission and reception by ICESat-2/ATLAS and its use of the 532 nm wavelength, the measurements are susceptible to interference from solar background noise, the instrument’s own dark noise, and atmospheric scattering noise. This results in a large number of noise photons in the ATL03 data, rendering it unsuitable for direct use in water level extraction without effective denoising. Currently, spaceborne photon-counting denoising algorithms fall into three main categories. The first type converts 2D laser photon data into raster images and then removes noise using image processing techniques [38]. However, this conversion can lead to the loss of valid information, resulting in significant errors. The second type involves using machine learning algorithms to extract signal photon features and remove noise, but this approach relies on high-quality labeled data [39]. The manual annotation of signal photons from ICESat-2/ATLAS is extremely labor-intensive, and the lack of annotated data limits the widespread application of this method. The third type of algorithm leverages the spatial density differences between signal and noise photons for denoising. Density-based spatial clustering algorithms and local statistical methods are widely adopted due to their simplicity and efficiency. However, local statistical methods are highly sensitive to threshold selection, and density-based clustering algorithms such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise) and OPTICS (Ordering Points to Identify the Clustering Structure) are similarly sensitive to parameter settings. When water surface signal photons are unevenly distributed or noise photons exhibit high local density of noise photons is high, the denoising performance of these algorithms may not be optimal.
To address the aforementioned challenges, this study proposes a Spatial Distribution-Based Hierarchical Clustering for Photon-counting Laser Altimeter (SD-HCPLA) to improve the accuracy of water level monitoring. First, to tackle the issue of uneven distribution of water surface signal photons that results in suboptimal denoising performance, this study fully utilizes the spatial distribution features of these photons and develops a Spatial Distribution-Based Hierarchical Clustering for Photon-counting Laser Altimeter (SD-HCPLA), grounded in spatial distribution and hierarchical clustering principles. Second, a denoising experiment was conducted at the Danjiangkou Reservoir, where this method was applied to extract water level data. This study also explored the accuracy and data acquisition capabilities of ICESat-2 elevation products (ATL08 and ATL13) in dynamic water level monitoring. A comprehensive evaluation of the performance of the SD-HCPLA method and ICESat-2 elevation products in monitoring inland lake water levels was then carried out. Subsequently, the SD-HCPLA method was employed to extract water level data for five lakes and five reservoirs in Hubei Province between October 2018 and December 2023, providing a valuable reference for using ATL03 altimetry data to monitor inland lake water level changes across the country. Finally, by analyzing the water level variations in the lakes and reservoirs, this study investigated the causes of these changes from the perspectives of natural factors and human activities, offering scientific evidence for watershed water resource management and ecological protection.

2. Study Areas and Data Sources

2.1. Study Areas

Hubei Province, located in central China, spans geographical coordinates from 29°01′53″N to 33°06′47″N latitude and 108°21′42″E to 116°07′50″E longitude. Known as the “Province of a Thousand Lakes”, Hubei is characterized by a dense river network and abundant water resources, making it a critical source of water for China and a net exporter of water resources. Therefore, monitoring water resources in Hubei Province is of great importance. This study selected several key lakes and reservoirs within Hubei Province as the experimental research area. Figure 1 depicts the spatial distribution of five lakes and five reservoirs, alongside the river network and other water bodies across the province. Additionally, Figure 1 shows the ground tracks of the ICESat-2/ATLAS satellite over these lakes and reservoirs between October 2018 and December 2023.
Table 1 summarizes the main physical characteristics of these lakes and reservoirs, with the lengths and widths of the lakes estimated using Google Earth Pro. The Danjiangkou Reservoir (Figure 1A) is the main water source for China’s South-to-North Water Diversion Project’s middle route, covering a total area of 846 square kilometers, with an annual average inflow of 39.48 billion cubic meters, ensuring water supply for cities such as Beijing, Tianjin, Henan, and Hebei. The Zhanghe Reservoir (Figure 1B), located in central Hubei, controls a catchment area of 2212 square kilometers and is an important artificial reservoir. The Fushui Reservoir (Figure 1C), situated in the Tongshan and Yangxin counties of Hubei, serves multiple functions, including flood control, power generation, and irrigation. The Shuibuya Hydropower Station (Figure 1D) is primarily focused on power generation, with a storage capacity of 4.58 billion cubic meters. The Bailianhe Reservoir (Figure 1E) is the third-largest reservoir in Hubei, with a total storage capacity of 1.104 billion cubic meters. Honghu Lake (Figure 1F) is the largest lake in Hubei, covering an area of approximately 350 square kilometers, with a well-preserved ecosystem. Liangzi Lake (Figure 1G) is the second-largest lake in Hubei, with an area of 370 square kilometers. Futou Lake (Figure 1H), the third-largest lake in the province, spans 161 square kilometers and is an important aquatic conservation area. Longgan Lake (Figure 1I), a key flood retention area in the middle reaches of the Yangtze River, covers 420 square kilometers. Daye Lake (Figure 1J), located southeast of Daye City, and has an average water level of about 16 m.

2.2. ICESat-2 Altimetry Products Data

ICESat-2, the successor to the ICESat mission, is primarily designed to observe ice sheet elevation, sea ice thickness, sea level height, and vegetation canopy height, while also supporting altimetry of inland water bodies. ICESat-2 is equipped with the Advanced Topographic Laser Altimeter System (ATLAS), which measures by emitting 532 nm green laser pulses. Figure 2 illustrates the ground footprints of its strong and weak beams. In this study, we primarily use the Level-2 product ATL03 and the Level-3 products ATL08 and ATL13 for lake water level monitoring. ATL03 provides information on the latitude, longitude, and ellipsoidal height of each photon, and employs a histogram-based denoising method to classify the photons. ATL08, based on ATL03, uses an adaptive nearest neighbor search method to classify land and vegetation photons. ATL13, also based on ATL03 and combined with inland water body mask data, is specifically tailored for monitoring the water levels of inland water bodies.

2.3. Landsat-8/9 Remote Sensing Imagery Data

The Landsat series, operated by NASA (the United States National Aeronautics and Space Administration), is the longest-running Earth observation satellite mission, beginning with the launch of Landsat-1 in 1972 and continuing with the launch of Landsat-9 in 2021. These satellites provide invaluable long-term remote sensing datasets, which are crucial for global change research and are widely used in fields such as agriculture, forestry, meteorology, and wildfire monitoring. The ICESat-2 satellite was launched in 2018, while Landsat-8 and Landsat-9 were launched on 11 February 2013, and 27 September 2021, respectively. In this study, we primarily utilized Landsat-8/9 Collection 2 Level-2 remote sensing imagery to extract water bodies and generate water body masks. The multispectral bands of Landsat-8/9 have a spatial resolution of 30 m. The Landsat-8/9 data used in this study were obtained from the USGS platform (https://earthexplorer.usgs.gov/, accessed on 1 January 2025).

2.4. Danjiangkou Reservoir Water Level Data

This study uses reservoir and lake water level data provided by the Hubei Provincial Department of Water Resources as validation data. The Department maintains daily records of water levels, inflow, and total outflow for large and medium-sized reservoirs and lakes in Hubei Province. These data were accessed from the department’s official portal (http://slt.hubei.gov.cn/sjfb/) accessed on 1 May 2023. The water level data for Danjiangkou Reservoir since 22 September 2021 can now be accessed through the following website: https://hbwater.wetruetech.com/water/portal/wx_station_info?stationCode=61802700&stationType=RR (accessed on 1 January 2025).
The water level data used in this study were sourced from the Danjiangkou Reservoir Evaporation Station, which is operated by the Hanjiang Bureau of China, as shown in Figure 3. The station was established in December 2013 and is located in Danjiangkou City, Hubei Province (111°30′E, 32°34′N). The station monitors various parameters, including evaporation, precipitation, water temperature, air temperature, air pressure, sunshine, total radiation, wind speed, wind direction, ground temperature, and soil moisture. All data are collected in real-time and automatically transmitted. Figure 3 also presents a schematic of the ICESat-2 laser footprints over the Danjiangkou Reservoir in 2023, with the location of the Danjiangkou Dam indicated.

2.5. Related Climate Data

The precipitation and evapotranspiration data used in this study are derived from the 1 km resolution annual precipitation dataset for China (1982–2022) [40] and the 1 km monthly potential evapotranspiration dataset for China (1990–2021) [41], both provided by the National Earth System Science Data Center. The 1 km resolution annual precipitation dataset (1982–2022) is generated by summing the monthly precipitation data from the 1 km resolution dataset (1901–2022) provided by Professor Peng Shouzhang. This dataset consists of multiple TIF files, each representing the annual accumulated precipitation for a given year, with precipitation values measured in millimeters (mm). The 1 km monthly potential evapotranspiration dataset for China (1990–2021) is derived using the Hargreaves potential evapotranspiration formula, and the data are stored as int16 type in NetCDF (NC) files. The dataset is available for access and download through the National Earth System Science Data Center via the following link: https://www.geodata.cn/data/datadetails.html?dataguid=113786088533256 (accessed on 15 January 2025).
The monthly average temperature dataset used in this study was sourced from the 1 km resolution monthly mean temperature dataset for China (1901–2023) [42] provided by the National Earth System Science Data Center, National Science & Technology Infrastructure of China (http://www.geodata.cn, accessed on 15 January 2025).

3. Method

First, the Landsat 8/9 imagery was used to extract water body masks for the lakes, and these masks were then applied to filter the ATL03, ATL08, and ATL13 data within the water body areas. Since ATL08 and ATL13 are NASA’s officially processed Level 3a data products, the data filtered through the water body masks can be directly used for water level calculations. For the ATL03 data filtered by the water body masks, a Spatial Distribution-Based Hierarchical Clustering for Photon-counting Laser Altimeter (SD-HCPLA) was applied to extract signal photons before proceeding with water level calculations. The overall workflow is illustrated in Figure 4.

3.1. Lake Water Body Masking Extraction

In this study, the Normalized Difference Water Index (NDWI) is employed to extract water bodies from Landsat 8/9 Collection 2 Level-2 imagery. NDWI, a widely used index for water body extraction from Landsat data [43], is defined as:
N D W I = N I R − R E D N I R + R E D
In this study, NIR represents the near-infrared band, RED represents the visible red band, and NDWI is a dimensionless index. The data used are Landsat 8/9 Collection 2 Level-2 imagery, which has been geometrically and atmospherically corrected. After clipping the images, NDWI values were calculated using Band 3 and Band 5 to generate a binary map. Morphological operations, including erosion and dilation [44], were then applied to separate reservoirs or lakes from smaller river features. This resulted in a binary mask that isolated the largest lake polygon, forming the water body mask. In this study, the NDWI values ranged from 0 to 0.3. A 3 × 3 kernel was used for both erosion and dilation, with erosion iterated three times and dilation iterated twice, providing a buffer around the lake boundaries.
Landsat 8/9 images acquired between 2018 and 2023, with cloud cover below 20%, were processed to generate water body masks. Lakes obscured by cloud cover were excluded from the dataset. A partial diagram of the water body masks for some lakes is shown in Figure 5.

3.2. Steps and Principles of the SD-HCPLA Algorithm

3.2.1. Data Preprocessing

Before applying the SD-HCPLA algorithm, data preprocessing was performed. Since multiple ATL03 tracks need to be processed, a quality assessment of the ATL03 data is first conducted to identify the segments with signal photon distributions. Initially, the number of histogram bins was determined using Equation (1), after which the elevation data were segmented, and the number of photons in each segment was counted. The data quality was then evaluated using Equation (2).
b i n s _ n u m b e r = c o u n t s
Q u a l i t y = m a x _ c o u n t m e d i a n _ c o u n t
In this context, b i n s _ n u m b e r represents the number of histogram bins, and c o u n t s refers to the number of photons above the water and denotes rounding to the nearest integer. Q u a l i t y is the quality evaluation metric, m a x ⁡ _ c o u n t is the maximum frequency in the histogram, and m e d i a n _ c o u n t is the median frequency. The photons filtered by the water body mask undergo quality evaluation. When the Q u a l i t y value is less than 4, the photon distribution is considered uniform, with no signal photons, and no further processing is required. When the Q u a l i t y value exceeds 4, the photons are segmented every 20 m, and data quality is evaluated for each segment. If the quality evaluation value still exceeds 4, the photons from the segment with the highest frequency are selected as the preliminary denoised result. Subsequently, the SD-HCPLA algorithm is applied to extract the signal photons.

3.2.2. SD-HCPLA Algorithm Principle

As shown in Figure 6, there was a significant difference in the density between signal photons and noise photons in water body areas. Upon close inspection, it can be observed that the distribution of signal photons is relatively uneven in the horizontal direction, while in the vertical direction, the signal photons are spaced more closely together. This phenomenon is primarily due to the fact that the ICESat-2/ATLAS laser reflection points are located on the surface of the terrain. Since ICESat-2 achieves centimeter-level precision in the vertical direction, signal photons are closely spaced vertically, making it easier to distinguish them from noise photons. However, in the horizontal direction, due to the characteristics of the terrain surface, atmospheric conditions, and instrument errors, the spacing between signal photons is larger and their distribution more irregular.
As shown in Figure 6c,d, noise photons in the water body area were farther from the signal photons, and the distribution of signal photons was uneven in the horizontal direction. Furthermore, as illustrated in Figure 7a, some noise photons (such as point E) were far from the signal photons, and the distance between locally dense noise photons was also relatively large. During hierarchical clustering, these noise photons may form small noise clusters that are far from the signal photons (such as point G). When the signal photons are unevenly distributed, the distance between signal photons becomes larger, and in some cases, noise photons may be closer to the signal photons than the signal photons are to each other. As shown in Figure 7b, Euclidean distance failed to effectively distinguish between noise photons and signal photons in the water body area. However, by replacing the Euclidean distance with the vertical distance between photons, it becomes possible to distinguish noise photons from signal photons more effectively. This is because the vertical distance of photons in the water area can more accurately reflect the actual density relationship between photons, thus avoiding misjudgment caused by uneven distribution of signal photons. This replacement of distance measurement makes it possible to more clearly identify the dense areas where signal photons are located during the hierarchical clustering process, and correctly classify noise photons into independent small clusters or noise clusters, thereby improving the accuracy and robustness of the clustering results.
To fully utilize the distribution characteristics of surface signal photons, a Spatial Distribution-Based Hierarchical Clustering for Photon-counting Laser Altimeter (SD-HCPLA) was developed.
  • First, a KD-tree was constructed to enhance the indexing efficiency between photons. The Euclidean distance is then used to calculate the distance between points, resulting in a matrix of the minimum distances between each pair of points. In a Cartesian coordinate system, ( x 1 , y 1 ) and ( x 2 , y 2 ) represent the coordinates of points p o i n t 1 and p o i n t 2 , respectively.
    d p o i n t 1 , p o i n t 2 = ( x 1 − x 2 ) 2 + ( y − y 2 ) 2
  • The Prim algorithm is then used to construct the Minimum Spanning Tree (MST), where the edges of the MST represent the Euclidean distances between adjacent nearest-neighbor points, as illustrated in Figure 8b.
  • Replace the Euclidean distance edges in the minimum spanning tree with the vertical distances between the connected points to form a new minimum spanning tree based on the closest vertical distances, as illustrated in Figure 8c.
  • The hierarchical clustering structure is constructed using a Minimum Spanning Tree (MST). The edges of the MST are first sorted in ascending order. Iteratively, edges connecting points are grouped into clusters. If an edge does not belong to any previously formed cluster, it is assigned to a new one. Edges that connect multiple clusters serve as bridges, merging smaller clusters into larger ones. This process ultimately results in a hierarchical clustering tree, as shown in Figure 8d.
  • As shown in Figure 8e, when extracting signal photons, the focus is primarily on removing outliers. Therefore, the edges of the clustering tree were divided into edges that connect points within the same cluster and edges that connect different clusters. The edges of the minimum spanning tree were statistically analyzed to obtain thresholds based on 3 times the standard deviation and 2 times the interquartile range (IQR) from a box plot. For edges within the same cluster, those exceeding 2 times the interquartile range were filtered to remove outliers connected to those edges. For edges connecting different clusters, those exceeding 3 times the standard deviation were filtered, and the sizes of the two connected clusters were calculated. If the size of a cluster was less than 5, it was considered noise photons.

3.3. Water Surface Elevation Extraction Using ATL08 and ATL13 Data

3.3.1. Water Level Data Extraction Using ATL08

ATL08 adopts the advanced Differential Regression Gaussian Adaptive Nearest Neighbor (DRAGANN) algorithm. The DRAGANN algorithm in ATL08 generates histograms based on the nearest neighbor distance, where the histogram consists of noise Gaussian distribution and signal Gaussian distribution. It searches for the intersection of the two Gaussian distributions, selects the optimal search radius, and uses this radius to identify signal photons. This method effectively extracts signal photons and is a general algorithm used by NASA for extracting signal photons from ATL03 data.
Since ATL08 does not provide the elevation or geographic coordinates (latitude and longitude) of photons, it is necessary to link its data with ATL03 through their respective data dictionaries. First, the “ph_segment_id” from ATL08 is matched with the “segment_id” from ATL03 to identify the corresponding 20 m segments in ATL03 for each segment in ATL08. Next, the photon index “classed_pc_indx” from ATL08 is used to associate the photons within each segment with their counterparts in ATL03. Finally, ATL08 photons with “classed_pc_flag” equal to 1 are selected to obtain the corresponding land photons in ATL03. Elevation conversion is then performed using Equation (5), and the water body mask is applied to filter out signal photons over the water. The elevation of the extracted water signal photons is averaged to obtain the water level elevation for the corresponding date.

3.3.2. Water Level Data Extraction Using ATL13

The ATL13 data product contains photons over water that have already been filtered using a global water mask and have undergone denoising. However, since the lake boundaries change with rainfall, temperature, and seasonal variations, the water body mask from Section 3.1 was applied to filter the ATL13 photon data. Elevation conversion was performed using Equation (5), and the average water surface elevation was calculated to obtain the water level elevation for the corresponding date.

3.4. Methodology for Converting WGS84 Elevation to the Wusong Elevation System

The ATL03 and ATL13 datasets record the longitude, latitude, and ellipsoidal height of each photon. Since ICESat-2 uses the WGS 1984 elevation system and the coordinate system for the lake water level stations in Hubei Province is the Wusong Datum, this study converts the ICESat-2 elevations to the Wusong Datum system using the following formula, as referenced in [45].
H W L = H I C E S a t − 2 − N + 1.717 − 0.32
where H I C E S a t − 2 represents the photon elevation recorded by ICESat-2, N is the geoid undulation, which is stored in the “gtx/geophys_corr/geoid” directory in ATL03 and in the “gtx/segment_geoid” directory in ATL13, and H W L is the water level elevation relative to the Wusong elevation system.

3.5. Preprocessing of Related Climate Data

This study utilized 1 km resolution annual precipitation TIFF data, the China 1 km monthly potential evapotranspiration dataset, and the China 1 km monthly average temperature dataset.
The data were preprocessed to convert the monthly data into annual averages. Precipitation, evapotranspiration, and temperature data were extracted from the central points of each water body to represent the annual trends in precipitation, evapotranspiration, and temperature for each water body. Due to the unavailability of 2023 annual precipitation data, the average annual precipitation for Hubei Province in 2023 was used as a proxy for the annual precipitation for each water body.

4. Results

4.1. Denoising Experiment for the Danjiangkou Reservoir

Taking the ATL03_20200303195727_10400606_006_01.h5 gt2r beam from 3 March 2020, and the ATL03_20210909052406_11851202_006_02.h5 gt1l beam from 9 September 2021, for the Danjiangkou Reservoir as examples, the water body mask data generated in Section 3A were used to filter ATL03 and ATL08 photon data in the lake area. The raw data are shown in Figure 9a and Figure 10a. Figure 9a presents the original ATL03 photon data from the Danjiangkou Reservoir area, where background noise is relatively low. In contrast, Figure 10a shows the original ATL03 photon data in the same area with significantly more background noise, making it a representative case. After filtering the ATL08 data and matching the segment IDs between ATL08 and ATL03, the photon indices with a “classed_pc_flag” of 1 were extracted, and the corresponding signal photon data from ATL03 were identified using the filtered ATL08 photon indices within each segment. These are shown as the red portions in Figure 9b and Figure 10b, representing the ATL08 signal photons in the Danjiangkou Reservoir area.
The processing of ATL03 data first involved a comprehensive data quality assessment of the photon data in the reservoir area based on the data quality evaluation method proposed in Section 3.3.1. The photon data were then segmented in 20 m intervals, and each segment underwent quality assessment and preprocessing. After preprocessing, the photon data were processed using the SD-HCPLA algorithm to calculate the minimum Euclidean distance matrix between photon points. Once the minimum spanning tree was generated, the Euclidean distance edges were replaced with vertical distance edges between photons, and a hierarchical clustering structure was constructed based on this tree. The edges of the minimum spanning tree were analyzed, and signal photons were selected using the preset thresholds. The results are shown in Figure 9c and Figure 10c.
A comparative analysis of Figure 9b,c and Figure 10b,c demonstrates that both the ATL08 data product and the SD-HCPLA algorithm effectively extracted a sufficient number of signal photons. A closer examination of the signal photons extracted by ATL08 and SD-HCPLA, as shown in Figure 9e,f and Figure 10e,f, reveals that ATL08 extracted more sparse photon data reflected from below the water surface, whereas SD-HCPLA accurately extracted the signal photons reflected from the water’s surface. As illustrated in Figure 9e,f, in the zoomed-in view of the gt2r beam on 3 March 2020, where there were fewer noise photons, the ATL08 data product primarily extracted signal photons reflected from below the water’s surface, while SD-HCPLA effectively filtered out photons farther below the water’s surface and accurately extracted the most densely distributed photons from the water’s surface. Figure 10e,f display zoomed-in views of the gt1l beam from 9 September 2021, where more noise photons were present. It can be seen that, although both ATL08 and SD-HCPLA were able to extract signal photons reflected from the water’s surface, ATL08 extracted more densely distributed photons from below the water surface, whereas SD-HCPLA successfully filtered out most of the densely distributed underwater noise photons.
The average elevation of signal photons extracted by ATL08 was computed, while for those extracted by SD-HCPLA, the median elevation was calculated for each segment, and the median elevations of all tracks were averaged. Similarly, elevation data from ATL13, filtered using the Danjiangkou Reservoir mask, were averaged to derive the reservoir’s water level. The water level data from ATL08, ATL13, and SD-HCPLA were then converted to the same elevation system and compared with in situ water level measurements for accuracy analysis, as shown in Table 2. The analysis results indicate that ATL13 data achieved the highest water level accuracy, with the lowest mean error, standard deviation, and root mean square error (RMSE), but only covered 17 days of water level data. In contrast, ATL08 data had lower accuracy, but covered 34 days of water level data. Specifically, ATL13 had a mean error of 0.1031 m, SD-HCPLA had a mean error of 0.1401 m, and ATL08 had a mean error of 0.8018 m. Overall, while the water level accuracy of SD-HCPLA was slightly lower than that of ATL13, it was significantly better than that of ATL08 and provided the most valid water level data, covering 41 days of observations. The comprehensive assessment shows that SD-HCPLA not only offers high accuracy, but also provides the largest amount of valid water level data.
Thus, integrating ATL03 data with the SD-HCPLA denoising algorithm enables both accurate water level estimation and maximized acquisition of valid water level observations.

4.2. Lake and Reservoir Water Level Changes

The SD-HCPLA algorithm was applied to extract water level data for five representative lakes and reservoirs in Hubei Province from 2018 to 31 December 2023. As indicated in Table 3, the Danjiangkou Reservoir provided the most extensive water level dataset, with 52 days of records, while the Shuibuya Reservoir had the fewest, with only 4 days of data.
This study utilized 1 km resolution annual precipitation TIFF data, the China 1 km monthly potential evapotranspiration dataset, and the China 1 km monthly average temperature dataset. As shown in Figure 11, with the exception of Zhanghe Reservoir and Shuibuya Reservoir, which had too few valid water level data points to draw conclusions on periodic variations, other water bodies exhibited seasonal and periodic water level fluctuations. In general, the reservoirs exhibited larger fluctuations, with the water levels of the Danjiangkou, Zhanghe, and Shuibuya Reservoirs varying by over 10 m. In contrast, the lakes exhibited relatively smaller fluctuations. For example, the difference between the maximum and minimum water levels in Honghu Lake and Futou Lake was approximately 2 m, while the difference in Daye Lake was only 5 m. Figure 11 and Figure 13 reveal that precipitation and evapotranspiration significantly influence lake and reservoir water levels, with 2020 being the year of the highest precipitation and the lowest evapotranspiration from 2018 to 2023, resulting in the highest water levels in 2020 for Fushui Reservoir, Bailianhe Reservoir, and the five lakes. As shown in Figure 12, there was no evident correlation between surface temperature and water level changes. Overall, Figure 11, Figure 12 and Figure 13 underscore that precipitation and evapotranspiration directly impact water levels, whereas surface temperature does not appear to have a significant effect.
The Danjiangkou Reservoir experienced the largest water level fluctuations, peaking at 170.1 m on 27 October 2021, and reaching the lowest value of 150.2 m on 5 May 2023, a difference of 19.8 m. As shown in Figure 11, the Danjiangkou Reservoir had already reached its peak water level by the end of 2021. Records show that it reached full capacity at 170 m on 10 October 2021. In October 2023, flooding in the Han River basin caused the water level of the Danjiangkou Reservoir to rise, consistent with the water level trend shown in Figure 11. Between 2019 and 2023, the water level of the Danjiangkou Reservoir significantly decreased, nearing 150 m, while in 2020 and 2021, the fluctuations were much smaller. This stability can be attributed to the sufficient rainfall during these years, which ensured an ample water supply, coupled with effective reservoir management, preventing significant fluctuations in water levels.
The Danjiangkou Reservoir’s water level typically rises at the beginning of each year, likely due to water storage operations in the autumn. Apart from the Danjiangkou Reservoir, 2020 was the year with the highest annual precipitation between 2018 and 2023, and the lowest evapotranspiration, resulting in the highest water levels for Fushui Reservoir, Bailianhe Reservoir, and the five lakes in 2020. A comprehensive analysis of the water level changes across the five lakes and five reservoirs revealed that lakes are more sensitive to precipitation variations. Unlike reservoirs, lakes exhibit seasonal fluctuations where water levels rise during the summer and drop in the winter, while reservoir levels decrease in the summer and increase in the winter. The periodic variations in reservoirs were weaker than those in lakes, likely due to the role of reservoirs in flood control and water flow regulation. Specifically, the Danjiangkou Reservoir plays a key role in supplying water to Henan, Hebei, Beijing, and Tianjin as part of the South-to-North Water Diversion Project. In the late-autumn, the reservoir stores water, while in the summer, it releases water to accommodate upstream inflows, primarily for flood control.

5. Discussion

ICESat-2 data have been extensively used for monitoring inland lake water levels, providing a cost-effective means of obtaining high-precision water level time series. The ICESat-2 data products ATL08 and ATL13 are directly applicable for monitoring water levels in inland water bodies. For example, Cooley et al. used the ATL08 product for global water level monitoring [34], and Lijuan Song applied the ATL13 product to monitor the water levels of Chinese lakes, achieving a mean error of 0.07 m [35]. Zhang et al. used the secondary ATL13 data from ICESat-2 to observe the water levels of Qinghai Lake [46], and Xu et al. employed ATL13 data to remove low-quality data and noise, thereby obtaining water level data for Australian lakes in 2019 [47]. However, research on the direct use of ATL03 data for lake water level monitoring remains limited, with most studies focusing on the ATL08 and ATL13 data products from ICESat-2, which, after error removal, yield higher accuracy in water level measurements. Furthermore, ATL03 requires preprocessing for denoising. Xie et al. applied a two-step statistical histogram-based denoising method to process photon data and extract water levels at Miyun Reservoir. Although the error between the denoised water levels and measured water levels was not explicitly mentioned, the Root-Mean-Square Error (RMSE) of the inverted water levels using denoised data was 0.553 m [48]. The denoising method proposed in this study, the Spatial Distribution-Based Hierarchical Clustering for Photon-counting Laser Altimeter (SD-HCPLA), effectively utilizes the spatial distribution characteristics of photon-counting LiDAR to extract signal photons over water bodies by leveraging both horizontal and vertical distribution differences above the water surface. The water levels extracted from the Danjiangkou Reservoir using the SD-HCPLA method had an RMSE of 0.302 m compared with measured water levels.
However, the algorithm requires stringent coarse denoising. During signal extraction, methods such as standard deviation and quartile-based noise removal are applied. If a significant amount of noise photons remain after coarse denoising and the number of noise photons exceeds that of signal photons, excessive noise may be retained, adversely affecting water level extraction results. When using the SD-HCPLA algorithm, it achieves superior accuracy compared with ATL08, with improved performance in terms of mean error, standard deviation, and RMSE. This is because ATL08, primarily designed for monitoring forest canopies and land elevations, adopts a conservative photon extraction approach to retain canopy and ground surface signals. This conservatism results in reduced accuracy when monitoring lake water levels, as illustrated in Figure 9b and Figure 10b. ATL13 demonstrated the highest accuracy in water level extraction, with a mean error of only 0.1 m, and standard deviation and RMSE of 0.13 m. Moreover, applying error removal techniques to exclude poorly estimated lake water level data further improves accuracy, although this reduces the availability of water level data. From 2018 to April 6, 2023, only 17 days of water level data were available for the Danjiangkou Reservoir, indicating limited data availability. This limitation may be attributed to the conservative water body mask data and cautious signal photon selection in ATL13, resulting in fewer data points but higher accuracy.
The SD-HCPLA algorithm was applied to extract water level data for five representative lakes and reservoirs in Hubei Province from 2018 to 31 December 2023. The east–west and north–south lengths of each lake and reservoir were estimated using Google Earth, as shown in Table 4. The results indicate that the Danjiangkou Reservoir provided the most water level data, while the Shuibuya Reservoir provided the least. This disparity is largely due to the relatively short east–west span of the Shuibuya Reservoir, resulting in fewer ATL03 data tracks passing over it—only 27 tracks. In contrast, the Danjiangkou Reservoir, with an east–west length of 60 km, is the longest water body in the study area, yielding a larger number of tracks (101 tracks), which ultimately provided 52 days of water level data. Cooley et al. [34] and Lijuan Song [35] noted that lake area affects the availability of valid water level data. In this study, Figure 14 further confirms a significant positive correlation between the number of effective water level observation days and the east–west length of the lakes, suggesting that longer east–west dimensions correspond to a greater number of effective water level observations.
Although the repeat orbit cycle of ICESat-2 is 91 days, theoretically allowing for quarterly water level monitoring, actual observations are heavily influenced by weather conditions and instrumental limitations. In particular, under thick cloud cover, the laser photons from ICESat-2/ATLAS struggle to penetrate clouds, leading to missing water level data. Therefore, for larger water bodies, the more frequently the water body is covered by tracks, the higher the probability of obtaining valid water level data. This highlights the importance of analyzing the spatial geometric characteristics of lakes and reservoirs, particularly their east–west extension, in order to increase the frequency of water level data acquisition.
When utilizing the SD-HCPLA algorithm, more effective water level data can be extracted. For larger lakes and reservoirs, the algorithm can capture data reflecting changes in water bodies, and for smaller water bodies, it can retrieve as much water level data as possible. However, the SD-HCPLA algorithm has a relatively long runtime. Although a data quality score is employed to determine the presence of signal photons, the algorithm lacks the ability to differentiate cloud photons. Therefore, when applying the algorithm, a 3-sigma rule must be used to remove outliers. Additionally, in areas where there are no signal photons but noise anomalies exist, misjudgments may occur. To expand the application of this algorithm to nationwide lake water level monitoring and obtain more comprehensive water level data, it is imperative to develop a simpler and more efficient denoising algorithm in the future.
As illustrated in Figure 11, the water level variation trends of lakes and reservoirs differed significantly. This is because lakes are naturally formed water bodies and part of the natural ecosystem, whereas reservoirs are artificially constructed storage facilities primarily used for water resource management, including irrigation, water supply, power generation, flood control, and regulating river flows. The Danjiangkou Reservoir is not only a water storage facility, but also serves as a key water supply source for the South-to-North Water Diversion Project, which requires stricter reservoir regulation. It plays a crucial role in flood control and ensures the safety of water supply for the diversion project.
Before the onset of the flood season each year, reservoirs like the Danjiangkou Reservoir systematically release water to ensure sufficient storage capacity, mitigate flood peaks in the Han River, and protect downstream cities. As the dry season approaches, these reservoirs begin to store water, raising the water levels to secure a sufficient supply for future needs. Consequently, the Danjiangkou Reservoir exhibits a reversed water level cycle compared with lakes: its water levels decrease during the rainy season and increase during the dry season. However, the water levels in reservoirs are also influenced by upstream precipitation and river flow. From late-2021 to mid-2023, the water level of the Danjiangkou Reservoir consistently declined, likely due to drought conditions in the Han River basin, which prevented the reservoir from being replenished.
In contrast, water level changes in other reservoirs are less periodic than those in the Danjiangkou Reservoir. This may be due to the limited availability of effective water level data, making the changes less noticeable. Another possible reason is that reservoir water levels are adjusted based on local water storage needs or flood peak mitigation requirements. On the other hand, lake water levels exhibit strong cyclic and seasonal variation. Unlike reservoirs, lakes experience water level increases during the flood season and decreases during droughts, showing a strong correlation with precipitation.
As shown in Figure 11, the precipitation in 2019 was the highest compared with that in 2018 and 2023, leading to the highest lake water levels occurring from 2019 to 2023. The conclusion from Figure 11 is that, although lakes exhibit strong cyclic variations, their water level changes are relatively gradual and stable. This is because lake water levels are typically determined by natural processes, such as precipitation, runoff, and evaporation. Consequently, lake water levels are directly influenced by precipitation and evaporation, making them more reflective of changes in precipitation. Additionally, the seasonal fluctuations in lake water levels can reflect the local climate characteristics [49].
Reservoir water levels, in contrast, are strongly influenced by human intervention, reflecting demands for local agriculture, urban water supply, industrial use, and flood control. As a result, reservoir water level changes can partially indicate the economic activities in the region. Persistent water level decreases or increases in reservoirs also signal issues with upstream water availability. Therefore, monitoring lake water level changes in a region can more effectively reflect local ecological and environmental changes. Moreover, lake water levels can, to some extent, indicate the region’s economic development.

6. Conclusions

To improve the accuracy of water level monitoring and increase the availability of effective data days, this study proposes the Spatial Distribution-Based Hierarchical Clustering for photon-Counting Laser Altimeter (SD-HCPLA) algorithm. The method was experimentally validated and analyzed at the Danjiangkou Reservoir, and the results were compared with water levels derived from the ATL08 and ATL13 data products. Subsequently, the SD-HCPLA method was employed to extract water level data for five lakes and five reservoirs in Hubei Province, leading to the following conclusions:
(1)
The application of the SD-HCPLA method for water level extraction at the Danjiangkou Reservoir systematically evaluated the accuracy and applicability of ATL08 and ATL13 data for water level monitoring. The results indicate that the SD-HCPLA algorithm significantly improved measurement accuracy compared with the direct use of ATL08, with an average water level error of 0.8018 m for ATL08, while the error for the SD-HCPLA algorithm was reduced to 0.1401 m. Furthermore, the SD-HCPLA method considerably increased the number of effective data acquisition days, obtaining 41 days of valid water level data, compared with only 17 days for ATL13. However, the algorithm has the limitation of a longer runtime during the denoising process. Future research should focus on optimizing the denoising algorithm to reduce runtime, thereby enhancing its efficiency for nationwide lake water level monitoring. Additionally, subsequent studies could apply the SD-HCPLA algorithm to bare land and glaciers, optimizing the method to improve elevation accuracy and monitoring precision of glacial changes using ICESat-2/ATLAS.
(2)
Using the SD-HCPLA algorithm, water level data from 2018 to 2023 were obtained for five lakes and five reservoirs in Hubei Province. This study shows significant differences in the water level variation trends between lakes and reservoirs: lake water levels exhibited distinct seasonal changes and periodic fluctuations, remaining relatively stable overall. In contrast, reservoir water levels were significantly influenced by artificial regulation, with higher levels typically observed during dry seasons and lower levels during rainy seasons, reflecting the role of reservoirs in water resource management. The water level variations in lakes primarily reflect the seasonal fluctuations of natural ecosystems, influenced by precipitation, runoff, and evaporation, whereas those in reservoirs are more indicative of the demands of artificial regulation, including flood control, water supply, power generation, and irrigation. Thus, monitoring lake water levels can more directly reflect local climate and ecological changes, while reservoir level fluctuations to some extent mirror changes in regional economic activities and the status of water resource management. In the future, integrating water level variation data from lakes and reservoirs could provide a more comprehensive reference for regional ecological monitoring and water resource management.

Author Contributions

Conceptualization, X.L. and X.W.; Data curation, X.L. and X.Y.; Formal analysis, X.L. and J.X.; Methodology, X.L.; Software, X.L., C.X. and F.Z.; Supervision, X.W. and J.X.; Validation, X.L. and F.M.; Visualization, X.L.; Writing—original draft, X.L.; Writing—review and editing, X.W. and J.X. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (42371391); in part by the Natural Science Foundation of China (Youth Program, Grant No. 42201428).

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Woolway, R.I.; Kraemer, B.M.; Lenters, J.D.; Merchant, C.J.; O’Reilly, C.M.; Sharma, S. Global lake responses to climate change. Nat. Rev. Earth Environ. 2020, 1, 388–403. [Google Scholar] [CrossRef]
  2. Tranvik, L.J.; Downing, J.A.; Cotner, J.B.; Loiselle, S.A.; Striegl, R.G.; Ballatore, T.J.; Dillon, P.; Finlay, K.; Fortino, K.; Knoll, L.B.; et al. Lakes and reservoirs as regulators of carbon cycling and climate. Limnol. Oceanogr. 2009, 54, 2298–2314. [Google Scholar] [CrossRef]
  3. Iqbal, K.; Limantara, L.M.; Soetopo, W.; Andawayanti, U. Multi-reservoirs with inter-reservoir water transfer operation rules. IOP Conf. Ser. Earth Environ. Sci. 2020, 437, 012021. [Google Scholar] [CrossRef]
  4. Lee, J.; Shin, H. Assessment of future climate change impact on an agricultural reservoir in South Korea. Water 2021, 13, 2125. [Google Scholar] [CrossRef]
  5. Shaikh, M.; Birajdar, F. Advancements in remote sensing and GIS for sustainable groundwater monitoring: Applications, challenges, and future directions. Int. J. Res. Eng. Sci. Manag. 2024, 7, 16–24. [Google Scholar]
  6. Chen, J.; Liao, J. Monitoring lake level changes in China using multi-altimeter data (2016–2019). J. Hydrol. 2020, 590, 125544. [Google Scholar] [CrossRef]
  7. Zhang, J.; Xu, K.; Yang, Y.; Qi, L.; Hayashi, S.; Watanabe, M. Measuring water storage fluctuations in Lake Dongting, China, by Topex/Poseidon satellite altimetry. Environ. Monit. Assess. 2006, 115, 23–37. [Google Scholar] [CrossRef]
  8. Yuan, C.; Gong, P.; Liu, C.; Ke, C. Water-volume variations of Lake Hulun estimated from serial Jason altimeters and Landsat TM/ETM+ images from 2002 to 2017. Int. J. Remote Sens. 2019, 40, 670–692. [Google Scholar] [CrossRef]
  9. Berry, P.A.M.; Garlick, J.D.; Freeman, J.A.; Mathers, E.L. Global inland water monitoring from multi-mission altimetry. Geophys. Res. Lett. 2005, 32. [Google Scholar] [CrossRef]
  10. Jiang, L.; Nielsen, K.; Andersen, O.B.; Bauer-Gottwein, P. A bigger picture of how the Tibetan lakes have changed over the past decade revealed by CryoSat-2 altimetry. J. Geophys. Res. Atmos. 2020, 125, e2020JD033161. [Google Scholar] [CrossRef]
  11. Jiang, L.; Nielsen, K.; Andersen, O.B. Improvements in mountain lake monitoring from satellite altimetry over the past 30 years–lessons learned from Tibetan lakes. Remote Sens. Environ. 2023, 295, 113702. [Google Scholar] [CrossRef]
  12. Zhang, G.; Yao, T.; Xie, H.; Yang, K.; Zhu, L.; Shum, C.; Bolch, T.; Yi, S.; Allen, S.; Jiang, L.; et al. Response of Tibetan Plateau lakes to climate change: Trends, patterns, and mechanisms. Earth-Sci. Rev. 2020, 208, 103269. [Google Scholar] [CrossRef]
  13. Deidda, C.; De Michele, C.; Arslan, A.N.; Pecora, S.; Taburet, N. Accuracy of copernicus altimeter water level data in Italian rivers accounting for narrow river sections. Remote Sens. 2021, 13, 4456. [Google Scholar] [CrossRef]
  14. Ma, S.; Liao, J.; Chen, J.; Guo, Y. An Improved Adaptive Multi-Scale Peak Detection Retracker for River Level Estimation Based on Sentinel-6 Fully Focused SAR Data. Remote Sens. 2025, 17, 791. [Google Scholar] [CrossRef]
  15. Moholdt, G.; Nuth, C.; Hagen, J.O.; Kohler, J. Recent elevation changes of Svalbard glaciers derived from ICESat laser altimetry. Remote Sens. Environ. 2010, 114, 2756–2767. [Google Scholar] [CrossRef]
  16. Cheng, C.; Du, W.; Li, J.; Bao, A.; Ge, W.; Wang, S.; Ma, D.; Pan, Y. Spatiotemporal variations of glacier mass balance in the tomur peak region based on multi-source altimetry remote sensing data. Remote Sens. 2023, 15, 4143. [Google Scholar] [CrossRef]
  17. Mahoney, C.; Hopkinson, C.; Kljun, N.; Van Gorsel, E. Estimating canopy gap fraction using ICESat GLAS within Australian forest ecosystems. Remote Sens. 2017, 9, 59. [Google Scholar] [CrossRef]
  18. Smith, B.; Fricker, H.A.; Holschuh, N.; Gardner, A.S.; Adusumilli, S.; Brunt, K.M.; Csatho, B.; Harbeck, K.; Huth, A.; Neumann, T.; et al. Land ice height-retrieval algorithm for NASA’s ICESat-2 photon-counting laser altimeter. Remote Sens. Environ. 2019, 233, 111352. [Google Scholar] [CrossRef]
  19. Brunt, K.M.; Neumann, T.A.; Smith, B.E. Assessment of ICESat-2 ice sheet surface heights, based on comparisons over the interior of the Antarctic ice sheet. Geophys. Res. Lett. 2019, 46, 13072–13078. [Google Scholar] [CrossRef]
  20. Li, W.; Niu, Z.; Shang, R.; Qin, Y.; Wang, L.; Chen, H. High-resolution mapping of forest canopy height using machine learning by coupling ICESat-2 LiDAR with Sentinel-1, Sentinel-2 and Landsat-8 data. Int. J. Appl. Earth Obs. Geoinf. 2020, 92, 102163. [Google Scholar] [CrossRef]
  21. Zhu, X.; Nie, S.; Wang, C.; Xi, X. The performance of ICESat-2’s strong and weak beams in estimating ground elevation and forest height. In Proceedings of the IGARSS 2020-2020 IEEE International Geoscience and Remote Sensing Symposium, Waikoloa, HI, USA, 26 September–2 October 2020; IEEE: Piscataway, NJ, USA; pp. 6073–6076. [Google Scholar]
  22. Zhu, X.; Nie, S.; Wang, C.; Xi, X.; Lao, J.; Li, D. Consistency analysis of forest height retrievals between GEDI and ICESat-2. Remote Sens. Environ. 2022, 281, 113244. [Google Scholar] [CrossRef]
  23. Narine, L.L.; Popescu, S.; Neuenschwander, A.; Zhou, T.; Srinivasan, S.; Harbeck, K. Estimating aboveground biomass and forest canopy cover with simulated ICESat-2 data. Remote Sens. Environ. 2019, 224, 1–11. [Google Scholar] [CrossRef]
  24. Forfinski, N.; Parrish, C. ICESat-2 bathymetry: An empirical feasibility assessment using MABEL. In Proceedings of the Remote Sensing of the Ocean, Sea Ice, Coastal Waters, and Large Water Regions 2016, Edinburgh, UK, 26–29 September 2016; SPIE: Redondo Beach, CA, USA; Volume 9999, pp. 21–28. [Google Scholar]
  25. Hsu, H.-J.; Huang, C.-Y.; Jasinski, M.; Li, Y.; Gao, H.; Yamanokuchi, T.; Wang, C.-G.; Chang, T.-M.; Ren, H.; Kuo, C.-Y.; et al. A semi-empirical scheme for bathymetric mapping in shallow water by ICESat-2 and Sentinel-2: A case study in the South China Sea. ISPRS J. Photogramm. Remote Sens. 2021, 178, 1–19. [Google Scholar] [CrossRef]
  26. Nguyen, V.A.; Ren, H.; Huang, C.Y.; Tseng, K.H. Bathymetry derivation in shallow water of the South China Sea with ICESat-2 and Sentinel-2 data. J. Appl. Remote Sens. 2021, 15, 044513. [Google Scholar] [CrossRef]
  27. Parrish, C.E.; Magruder, L.A.; Neuenschwander, A.L.; Forfinski-Sarkozi, N.; Alonzo, M.; Jasinski, M. Validation of ICESat-2 ATLAS bathymetry and analysis of ATLAS’s bathymetric mapping performance. Remote Sens. 2019, 11, 1634. [Google Scholar] [CrossRef]
  28. Luo, S.; Song, C.; Zhan, P.; Liu, K.; Chen, T.; Li, W.; Ke, L. Refined estimation of lake water level and storage changes on the Tibetan Plateau from ICESat/ICESat-2. Catena 2021, 200, 105177. [Google Scholar] [CrossRef]
  29. Zhang, G.; Chen, W.; Xie, H. Tibetan Plateau’s lake level and volume changes from NASA’s ICESat/ICESat-2 and Landsat Missions. Geophys. Res. Lett. 2019, 46, 13107–13118. [Google Scholar] [CrossRef]
  30. Ma, Y.; Xu, N.; Sun, J.; Wang, X.H.; Yang, F.; Li, S. Estimating water levels and volumes of lakes dated back to the 1980s using Landsat imagery and photon-counting lidar datasets. Remote Sens. Environ. 2019, 232, 111287. [Google Scholar] [CrossRef]
  31. Chen, T.; Song, C.; Luo, S.; Ke, L.; Liu, K.; Zhu, J. Monitoring global reservoirs using ICESat-2: Assessment on spatial coverage and application potential. J. Hydrol. 2022, 604, 127257. [Google Scholar] [CrossRef]
  32. Xu, N.; Ma, Y.; Zhang, W.; Wang, X.H.; Yang, F.; Su, D. Monitoring annual changes of lake water levels and volumes over 1984–2018 using Landsat imagery and ICESat-2 data. Remote Sens. 2020, 12, 4004. [Google Scholar] [CrossRef]
  33. Huang, X.; Cheng, F.; Wang, J.; Duan, P.; Wang, J. Forest canopy height extraction method based on ICESat-2/ATLAS data. IEEE Trans. Geosci. Remote Sens. 2023, 61, 5700814. [Google Scholar] [CrossRef]
  34. Cooley, S.W.; Ryan, J.C.; Smith, L.C. Human alteration of global surface water storage variability. Nature 2021, 591, 78–81. [Google Scholar] [CrossRef]
  35. Shanmu, M.; Fuping, G.; Huaichun, W. ICESat-2 data-based monitoring of 2018–2021 variations in the water levels of lakes in the Qinghai-Tibet Plateau. Remote Sens. Nat. Resour. 2022, 34, 164–172. [Google Scholar]
  36. Song, L.; Song, C.; Luo, S.; Chen, T.; Liu, K.; Zhang, Y.; Ke, L. Integrating ICESat-2 altimetry and machine learning to estimate the seasonal water level and storage variations of national-scale lakes in China. Remote Sens. Environ. 2023, 294, 113657. [Google Scholar] [CrossRef]
  37. Han, W.; Huang, C.; Gu, J.; Hou, J.; Zhang, Y.; Wang, W. Water level change of Qinghai Lake from ICESat and ICESat-2 laser altimetry. Remote Sens. 2022, 14, 6212. [Google Scholar] [CrossRef]
  38. Meng, W.; Li, J.; Tang, Q.; Xu, W.; Dong, Z. ICESat-2 laser data denoising algorithm based on a back propagation neural network. Appl. Opt. 2022, 61, 8395–8404. [Google Scholar] [CrossRef] [PubMed]
  39. Zheng, X.; Hou, C.; Huang, M.; Ma, D.; Li, M. A density and distance-based method for ICESat-2 photon-counting data denoising. IEEE Geosci. Remote Sens. Lett. 2023, 20, 6500405. [Google Scholar] [CrossRef]
  40. Peng, S.; Ding, Y.; Liu, W.; 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]
  41. Ren, J.Y.; Peng, S.Z.; Cao, Y.; Huo, X.Y.; Chen, Y.M. Spatiotemporal distribution characteristics of climate change in the Loess Plateau from 1901 to 2014. J. Nat. Resour. 2018, 33, 621–633. [Google Scholar]
  42. Peng, S.; Ding, Y.; Wen, Z.; Chen, Y.; Cao, Y.; Ren, J. 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]
  43. Xu, H. Modification of normalised difference water index (NDWI) to enhance open water features in remotely sensed imagery. Int. J. Remote Sens. 2006, 27, 3025–3033. [Google Scholar] [CrossRef]
  44. Gil, J.Y.; Kimmel, R. Efficient dilation, erosion, opening, and closing algorithms. IEEE Trans. Pattern Anal. Mach. Intell. 2002, 24, 1606–1617. [Google Scholar] [CrossRef]
  45. Zhang, J.T.; Liu, C.L. Water level change monitoring of Danjiangkou reservoir based on laser altimetry data. Sci. Surv. Mapp. 2021, 46, 20–24. [Google Scholar]
  46. Zhang, Z.; Bo, Y.; Jin, S.; Chen, G.; Dong, Z. Dynamic water level changes in Qinghai Lake from integrating refined ICESat-2 and GEDI altimetry data (2018–2021). J. Hydrol. 2023, 617, 129007. [Google Scholar] [CrossRef]
  47. Xu, N.; Ma, Y.; Zhang, W.; Wang, X.H. Surface-water-level changes during 2003–2019 in Australia revealed by ICESat/ICESat-2 altimetry and Landsat imagery. IEEE Geosci. Remote Sens. Lett. 2020, 18, 1129–1133. [Google Scholar] [CrossRef]
  48. Xie, J.; Li, B.; Jiao, H.; Zhou, Q.; Mei, Y.; Xie, D.; Wu, Y.; Sun, X.; Fu, Y. Water level change monitoring based on a new denoising algorithm using data from Landsat and ICESat-2: A case study of Miyun Reservoir in Beijing. Remote Sens. 2022, 14, 4344. [Google Scholar] [CrossRef]
  49. Zhang, X.; Yang, H.; Xu, J.; Wang, Y.; Liu, P.; Xu, C.Y. Increasing the available water diversion volume of water source project through flood resource utilization: A case study of the middle route of the South-to-North water diversion project in China. Reliab. Eng. Syst. Saf. 2025, 253, 110530. [Google Scholar] [CrossRef]
Remotesensing 17 00902 g001
Figure 1. Topographic map of Hubei Province showing the locations of the five reservoirs and five lakes, along with the ICESat-2 orbital tracks. (A) Danjiangkou Reservoir. (B) Zhanghe Reservoir. (C) Fushui Reservoir. (D) Shuibuya Reservoir. (E) Bailianhe Reservoir. (F) Honghu Lake. (G) Liangzi Lake. (H) Futou Lake. (I) Longgan Lake. (J) Daye Lake.
Figure 1. Topographic map of Hubei Province showing the locations of the five reservoirs and five lakes, along with the ICESat-2 orbital tracks. (A) Danjiangkou Reservoir. (B) Zhanghe Reservoir. (C) Fushui Reservoir. (D) Shuibuya Reservoir. (E) Bailianhe Reservoir. (F) Honghu Lake. (G) Liangzi Lake. (H) Futou Lake. (I) Longgan Lake. (J) Daye Lake.
Remotesensing 17 00902 g001
Remotesensing 17 00902 g002
Figure 2. Schematic diagram of ICESat-2 footprints.
Figure 2. Schematic diagram of ICESat-2 footprints.
Remotesensing 17 00902 g002
Remotesensing 17 00902 g003
Figure 3. Schematic map of the location of Danjiangkou Reservoir evaporation station.
Figure 3. Schematic map of the location of Danjiangkou Reservoir evaporation station.
Remotesensing 17 00902 g003
Remotesensing 17 00902 g004
Figure 4. Signal photon extraction process and water level extraction process.
Figure 4. Signal photon extraction process and water level extraction process.
Remotesensing 17 00902 g004
Remotesensing 17 00902 g005
Figure 5. Water body mask diagram for lakes and reservoirs. (A) Danjiangkou Reservoir. (B) Zhanghe Reservoir. (C) Fushui Reservoir. (D) Shuibuya Reservoir. (E) Bailianhe Reservoir. (F) Honghu Lake. (G) Liangzi Lake. (H) Futou Lake. (I) Longgan Lake. (J) Daye Lake.
Figure 5. Water body mask diagram for lakes and reservoirs. (A) Danjiangkou Reservoir. (B) Zhanghe Reservoir. (C) Fushui Reservoir. (D) Shuibuya Reservoir. (E) Bailianhe Reservoir. (F) Honghu Lake. (G) Liangzi Lake. (H) Futou Lake. (I) Longgan Lake. (J) Daye Lake.
Remotesensing 17 00902 g005
Remotesensing 17 00902 g006
Figure 6. Distribution of ATL03 photons. (a) Signal photon distribution. (b) Partial zoom of the signal photon distribution. (c,d) Difference in the distribution of signal photons and noise photons.
Figure 6. Distribution of ATL03 photons. (a) Signal photon distribution. (b) Partial zoom of the signal photon distribution. (c,d) Difference in the distribution of signal photons and noise photons.
Remotesensing 17 00902 g006
Remotesensing 17 00902 g007
Figure 7. Schematic diagram of the density distribution differences of photons in the horizontal and vertical directions. (a) Euclidean distance and vertical distance of signal photons. (b) Differences between Euclidean distance and vertical distance.
Figure 7. Schematic diagram of the density distribution differences of photons in the horizontal and vertical directions. (a) Euclidean distance and vertical distance of signal photons. (b) Differences between Euclidean distance and vertical distance.
Remotesensing 17 00902 g007
Remotesensing 17 00902 g008
Figure 8. A spatial distribution-based hierarchical clustering for photon-counting laser altimeter. (a) Photon data after coarse denoising. (b) Minimum spanning tree generated using Euclidean distance. (c) Schematic of minimum spanning tree construction based on photon density differences in the vertical direction. (d) Hierarchical structure generation. (e) Noise edge filtering using 3 standard deviations and 2 times the interquartile range. (f) Schematic of denoising results.
Figure 8. A spatial distribution-based hierarchical clustering for photon-counting laser altimeter. (a) Photon data after coarse denoising. (b) Minimum spanning tree generated using Euclidean distance. (c) Schematic of minimum spanning tree construction based on photon density differences in the vertical direction. (d) Hierarchical structure generation. (e) Noise edge filtering using 3 standard deviations and 2 times the interquartile range. (f) Schematic of denoising results.
Remotesensing 17 00902 g008
Remotesensing 17 00902 g009
Figure 9. Comparison of denoising results for the gt2r beam of ATL03_20200303195727_10400606_006_01.h5 in the Danjiangkou Reservoir. (a) Original signal photons. (b) ATL08 signal photons. (c) Signal photons extracted by SD-HCPLA. (e) Zoomed-in view of ATL08 (f) Zoomed-in view of SD-HCPLA.
Figure 9. Comparison of denoising results for the gt2r beam of ATL03_20200303195727_10400606_006_01.h5 in the Danjiangkou Reservoir. (a) Original signal photons. (b) ATL08 signal photons. (c) Signal photons extracted by SD-HCPLA. (e) Zoomed-in view of ATL08 (f) Zoomed-in view of SD-HCPLA.
Remotesensing 17 00902 g009
Remotesensing 17 00902 g010
Figure 10. Comparison of denoising results for the gt1l beam of ATL03_20210909052406_11851202_006_02.h5 in the Danjiangkou Reservoir (a) Original signal photons. (b) ATL08 signal photons. (c) Signal photons extracted by SD-HCPLA. (e) Zoomed-in view of ATL08. (f) Zoomed-in view of SD-HCPLA.
Figure 10. Comparison of denoising results for the gt1l beam of ATL03_20210909052406_11851202_006_02.h5 in the Danjiangkou Reservoir (a) Original signal photons. (b) ATL08 signal photons. (c) Signal photons extracted by SD-HCPLA. (e) Zoomed-in view of ATL08. (f) Zoomed-in view of SD-HCPLA.
Remotesensing 17 00902 g010
Remotesensing 17 00902 g011
Figure 11. Trends in lake and reservoir water level changes in relation to precipitation variations.
Figure 11. Trends in lake and reservoir water level changes in relation to precipitation variations.
Remotesensing 17 00902 g011
Remotesensing 17 00902 g012
Figure 12. Trends in lake and reservoir water level changes in relation to surface temperature variations.
Figure 12. Trends in lake and reservoir water level changes in relation to surface temperature variations.
Remotesensing 17 00902 g012
Remotesensing 17 00902 g013
Figure 13. Trends in lake and reservoir water level changes in relation to variations in evapotranspiration.
Figure 13. Trends in lake and reservoir water level changes in relation to variations in evapotranspiration.
Remotesensing 17 00902 g013
Remotesensing 17 00902 g014
Figure 14. Schematic diagram showing the relationship between the east–west length of the water body and the number of effective water level data days obtained.
Figure 14. Schematic diagram showing the relationship between the east–west length of the water body and the number of effective water level data days obtained.
Remotesensing 17 00902 g014
Table 1. Physical features of lakes and reservoirs.
Table 1. Physical features of lakes and reservoirs.
CharacteristicsLatitudeLongitudeTypesMax Length (km)Max Width (km)
Danjiangkou 32 ° 31 ′ 40 ″ – 33 ° 01 ′ 55 ″ 111 ° 03 ′ 27 ″ – 111 ° 42 ′ 36 ″ Reservoir6057
Zhanghe 30 ° 56 ′ 57 ″ – 31 ° 04 ′ 06 ″ 111 ° 58 ′ 45 ″ – 112 ° 06 ′ 24 ″ Reservoir1113
Fushui 29 ° 39 ′ 59 ″ – 29 ° 42 ′ 57 ″ 114 ° 39 ′ 14 ″ – 114 ° 52 ′ 53 ″ Reservoir2015
Shuibuya 30 ° 25 ′ 07 ″ – 30 ° 26 ′ 16 ″ 110 ° 16 ′ 42 ″ – 110 ° 19 ′ 57 ″ Reservoir63
Bailianhe 30 ° 33 ′ 58 ″ – 30 ° 41 ′ 07 ″ 115 ° 26 ′ 52 ″ – 115 ° 36 ′ 46 ″ Reservoir1512
Honghu 29 ° 44 ′ 07 ″ – 29 ° 58 ′ 04 ″ 113 ° 13 ′ 19 ″ – 113 ° 28 ′ 20 ″ Lake2428
Liangzi 30 ° 05 ′ 17 ″ – 30 ° 25 ′ 26 ″ 114 ° 21 ′ 51 ″ – 114 ° 45 ′ 11 ″ Lake2832
Futou 29 ° 57 ′ 27 ″ – 30 ° 07 ′ 04 ″ 114 ° 09 ′ 27 ″ – 114 ° 19 ′ 17 ″ Lake1617
Longgan 29 ° 52 ′ 48 ″ – 30 ° 02 ′ 53 ″ 115 ° 59 ′ 34 ″ – 116 ° 16 ′ 23 ″ Lake2719
Daye 30 ° 04 ′ 16 ″ – 30 ° 08 ′ 33 ″ 114 ° 59 ′ 38 ″ – 115 ° 11 ′ 1 ″ Lake198
Table 2. Comparison of water level accuracy for ATL08, ATL13, and SD-HCPLA methods.
Table 2. Comparison of water level accuracy for ATL08, ATL13, and SD-HCPLA methods.
MethodMean Error (m)Standard Deviation (m)Correlation Coefficient (m)RMSE (m)Effective Data Days
SD-HCPLA0.14010.29200.99790.302141
ATL080.80181.24130.96091.445134
ATL130.10310.13000.99960.128917
Table 3. Table of extreme water levels for lakes and reservoirs (2018–2023.12.31).
Table 3. Table of extreme water levels for lakes and reservoirs (2018–2023.12.31).
Water Body TypeWater Body NameMAX Height (m)MAX
Date		MIN Height (m)MIN
Date		MAX-MINValid Days
ReservoirDanjiangkou Reservoir170.12021.10.27150.2092023.05.0519.80052
ReservoirZhanghe Reservoir121.0482021.10.23109.8192019.03.1211.22911
ReservoirFushui Reservoir54.1872020.12.2348.8402022.03.095.34624
ReservoirShuibuya Reservoir390.6682022.01.01376.1672020.05.1814.5004
ReservoirBailianhe Reservoir103.2672020.12.1995.6162022.11.037.65017
LakeHong Lake25.8372020.09.1823.7862023.03.152.05024
LakeLiangzi Lake20.4302020.07.2816.9712019.04.173.45928
LakeFutou Lake21.9292020.10.2619.7682019.10.292.16011
LakeLonggan Lake15.9412020.08.1711.9402019.05.084.00134
LakeDaye Lake20.2262020.07.2415.0642021.01.085.16218
Table 4. Number of days with reservoir water level data (2018–2023.12.31).
Table 4. Number of days with reservoir water level data (2018–2023.12.31).
Type of Water BodyName of the Water BodyNumber of OrbitsValid DaysLength in the
East–West
Direction (km)		Length in the North–South Direction (km)
ReservoirDanjiangkou Reservoir101526057
ReservoirZhanghe Reservoir55111113
ReservoirFushui Reservoir64242015
ReservoirShuibuya Reservoir27463
ReservoirBailianhe Reservoir45171512
LakeHong Lake63242428
LakeLiangzi Lake60282832
LakeFutou Lake50111617
LakeLonggan Lake64342719
LakeDaye Lake5218198
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

Lv, X.; Wang, X.; Yang, X.; Xie, J.; Mo, F.; Xu, C.; Zhang, F. A Novel Photon-Counting Laser Point Cloud Denoising Method Based on Spatial Distribution Hierarchical Clustering for Inland Lake Water Level Monitoring. Remote Sens. 2025, 17, 902. https://doi.org/10.3390/rs17050902

AMA Style

Lv X, Wang X, Yang X, Xie J, Mo F, Xu C, Zhang F. A Novel Photon-Counting Laser Point Cloud Denoising Method Based on Spatial Distribution Hierarchical Clustering for Inland Lake Water Level Monitoring. Remote Sensing. 2025; 17(5):902. https://doi.org/10.3390/rs17050902

Chicago/Turabian Style

Lv, Xin, Xiao Wang, Xiaomeng Yang, Junfeng Xie, Fan Mo, Chaopeng Xu, and Fangxv Zhang. 2025. "A Novel Photon-Counting Laser Point Cloud Denoising Method Based on Spatial Distribution Hierarchical Clustering for Inland Lake Water Level Monitoring" Remote Sensing 17, no. 5: 902. https://doi.org/10.3390/rs17050902

APA Style

Lv, X., Wang, X., Yang, X., Xie, J., Mo, F., Xu, C., & Zhang, F. (2025). A Novel Photon-Counting Laser Point Cloud Denoising Method Based on Spatial Distribution Hierarchical Clustering for Inland Lake Water Level Monitoring. Remote Sensing, 17(5), 902. https://doi.org/10.3390/rs17050902

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