An Elevation Editing Method for Water Bodies in InSAR derived DSM

In the InSAR-DSM, which is the Digital Surface Model (DSM) acquired through Interferometric Synthetic Aperture Radar (InSAR), the presence of significant noise, voids, and elevation anomalies in water bodies is pervasive. These disparities arise from factors such as diminished backscattering due to electromagnetic wave absorption by water, surface-based specular reflection, and the impact of temporal decorrelation. To tackle this challenge, an automated elevation editing method for water bodies based on InSAR-DSM is introduced. Water bodies are automatically categorized as oceans, lakes, or rivers based on their morphological characteristics and location criteria. Distinct algorithms are subsequently deployed for elevation editing within each category. Results demonstrate that after editing water body elevations, all missing values within the original InSAR-DSM for water bodies are effectively addressed. Elevation anomalous pixels within water bodies are reduced by 68.5%, and the elevation standard deviation, which reflects elevation variability in water bodies, is decreased by 47.31%. This process effectively rectifies water body elevations within InSAR-DSM.


Introduction
A Digital Elevation Model (DEM) holds a foundational role in advancing the "Digital China" and broader "Digital Earth" initiatives by supplying essential digital terrain data [1] .InSAR, a microwave remote sensing technique, enables all-weather, continuous surface observations [2] .It offers significant advantages, including high precision, extensive coverage, and cost-effectiveness, making it a preeminent method for global DEM generation.However, it's crucial to note that InSAR produces a DSM rather than a true DEM [3] .Consequently, the DSM requires elevation editing, including tasks such as void filling, filtering building and vegetation heights, and refining water body elevations, to attain an accurate DEM.
Due to electromagnetic wave absorption by water, surface-based specular reflection, and the impact of temporal decorrelation [4,5] , water bodies within the InSAR-DSM often exhibit noise, voids, and elevation anomalies.Therefore, meticulous elevation editing of water bodies is essential to achieve a smooth representation while maintaining elevation accuracy.Presently, a significant gap exists in comprehensive automated methods encompassing the entire workflow, from water body classification to water body elevation editing within existing DEM editing approaches.This deficiency poses challenges in meeting the demands of global-scale DEM production [6,7] .Consequently, this paper proposes a methodology for the automated editing of water body elevations using InSAR-DSM data.This approach not only establishes the theoretical foundation for global DEM production but also aligns with the overarching objectives of advancing the "Digital China" and "Digital Earth" strategic initiatives.

2.
Research Area Overview Considering the necessity of encompassing three distinct water body types-oceans, lakes, and riversthis study focuses on a specific experimental region comprising portions of Macau, Zhongshan City, and Zhuhai City (21.9719°N~22.5240°N,113.2614°E~113.6284°E)as depicted in figure 1 (a).The fundamental parameters of the experimental data are presented in table 1.This experimental region is situated along the southeastern coast of mainland China, on the western bank of the Pearl River Delta.It represents a convergence point between mainland China and the South China Sea, characterized by numerous watercourses.As illustrated in figure 1 (b), the InSAR-DSM corresponding to this experimental area exhibits a significant presence of voids (depicted as blank areas), noise, and elevation anomalies within its water bodies.Therefore, further elevation editing procedures are urgently required.The schematic representation of the automated elevation editing method for water bodies based on InSAR-DSM is depicted in figure 2. To accommodate the diverse characteristics of oceans, lakes, and rivers, as well as their unique production requirements [8] : (1) Geoid undulation height assignments are imperative for oceans.
(3) Dynamic rivers, reflecting variability, require varying elevations to capture their dynamic nature.Consequently, the water body mask extracted from the InSAR-DSM is categorized into these three classifications: ocean, lake, and river.Distinct algorithms are subsequently deployed for elevation editing within each category.A smoothing transition process is applied to ensure seamless integration and address potential elevation discontinuities at the boundaries.

Water Body Classification
Water body classification entails categorizing water bodies into oceans, lakes, and rivers based on morphological, area, and geographical criteria.The following steps are executed: (1) The ocean mask is extracted for the experimental region using high-resolution global coastline vector data.Subsequently, water body masks are classified into oceans based on this ocean mask.
(2) Lakes typically exhibit circular or nearly circular shapes, whereas rivers possess elongated forms.Therefore, criteria such as the aspect ratio of the minimum bounding rectangle for connected components within the water body mask [9,10] are utilized to classify the remaining water body mask into lakes and rivers.
The high-resolution global coastline vector data utilized in this study derives from the GSHHG (Global Self-consistent, Hierarchical, High-resolution Geography Database), publicly released by the United States National Oceanic and Atmospheric Administration (NOAA) on June 15, 2017.GSHHG combines three openly available coastal datasets: World Vector Shorelines (WVS), CIA World Data Bank II (WDBII), and the Atlas of the Cryosphere (AC).This dataset provides comprehensive global coastline information encompassing landmasses and islands." The outcomes of water body classification for the experimental region are depicted in figure 3 (a).

Water body elevation editing
Water body elevation editing involves the application of distinct algorithms to edit elevations within different water body categories, tailored to their unique characteristics and production requirements.The sequential process unfolds as follows: (1) Geoid undulation heights are computed for each ocean pixel, followed by corresponding adjustments to elevation values in the InSAR-DSM.
(2) Lake pixels within the same connected component are assigned the lowest elevation value among the surrounding lake shore pixels.
(3) Multiple riverbank pixels proximate to each river pixel are identified, and the lowest elevation value among them is assigned to the river pixel.
(4) Several rounds of smoothing filters are applied to the assigned river pixels.
(5) At the boundaries separating various water body categories, a smooth transition process is implemented using the inverse distance method.
Figure 3 (b) illustrates a comparison between the InSAR-DSM before and after water body elevation editing, revealing that voids within water bodies have been filled, and elevation anomalies and noise have been significantly reduced, yielding a substantially smoother water body elevation editing outcome.
Figure 4 presents a comparative analysis of water body elevation profiles before and after elevation editing.It is discernible that, prior to editing, elevations within marine, lake, and river areas exhibited notable fluctuations, ranging from several meters to as high as 51 meters, diverging from the typical elevation attributes of natural water bodies.
Subsequently, following the elevation editing process, elevation fluctuations within these areas were confined within a 1-meter range, and, in certain cases, reduced to as little as 0.1 meters.This adjustment corresponds harmoniously with the inherent elevation characteristics of water bodies, featuring minimal elevation variability in oceans, static elevation levels in lakes, and gradual elevation changes in rivers." Table 2 presents changes in pertinent metrics of water bodies within the InSAR-DSM before and after the water body elevation editing process.Notably, this process fills all voids within the InSAR-DSM water bodies, reduces elevation anomaly pixels by 68.5%, and diminishes elevation standard deviation (reflecting elevation variability) by 47.31%.These findings underscore the effectiveness of the water body elevation editing method in mitigating voids, noise, and elevation anomalies within InSAR-DSM water bodies.

Conclusion
This paper has introduced an automated elevation editing method for water bodies based on InSAR-DSM, tailored to the characteristics of InSAR-DSM water bodies and DEM production standards.It effectively addresses issues of noise, voids, and elevation anomalies prevalent within the original InSAR-DSM water bodies.Beyond providing a theoretical foundation for global DEM production, this method contributes to the realization of the "Digital China" and "Digital Earth" strategies.

Figure 1 .
Figure 1.Experimental region, (a) displays the extent of the experimental area, (b) displays the corresponding InSAR-DSM.

Figure 2 .
Figure 2. Overall workflow of the automated elevation editing method for water bodies based on InSAR-DSM.

Figure 3 .
Figure 3. (a) represents the water body classification results, (b) illustrates the outcomes of water body elevation editing.

Figure 4 .
Figure 4.The elevation profiles of aquatic features were compared pre-and postelevation editing.The left panel displays DSM-based elevation renderings for marine, lake, and river environments before and after editing, denoting profile lines in yellow.The right panel depicts corresponding elevation profiles, with the highest elevation differential denoted.

Table 1 .
Fundamental Parameters of Experimental Data.

Table 2 .
Comparison of Water Body Metrics in the InSAR-DSM Before and After Water Body Elevation Editing