SWOT-based flood descriptors - South Sudan, 2023-2025

Field name 

Description

Use Case Name 

Flooding and health care service disruption in South Sudan.

Dataset Name

SWOT-based Flood Descriptors

Dataset Description 

This dataset contains SWOT-derived flood statistic computed at specific health facilities (HF) across an area of interest spreading 189.856 km2 within South Sudan. Flood statistics were calculated using SWOT raster frames for the 2023-2025 period after a quality filtering step to remove pixels with a) degraded and bad quality measurements for water surface elevation, b) active flags in fields of "value_bad", "geolocation_qual_degraded", and "classification_qual_degraded", c) layover systematic errors above 95th percentile, d) a cross-track distance outside of 10-60 km in both sides of the swath, e) a dark water fraction greater than 0.2, and f) a water fraction lower than 0.15. Flood descriptors were calculated at square buffers around the 279 facilities, with a buffer size of 5.4x5.4km for PHCU type and 3.6x3.6 km for PHCC. Each flood descriptor includes an additional field with an estimation of the uncertainty propagated from the wse_uncert field.

Each file represents a single SWOT frame and contains one record for each health facility located within its footprint. Output files retain the original names of the corresponding SWOT frames.

Temporal Domain 

2023-2025

Spatial Domain 

South Sudan (bounding box ranging from longitude 28.5942ºE to 4.8367ºE and latitude 31.8383ºN to 9.6283ºN, in EPSG:4326).

Key Variables 

Flood fraction: Number of pixels flooded respect the total one within the buffer (0-1). Mean depth. Water surface elevation average among the identified flooded pixels. Median depth. Water surface elevation median value among the identified flooded pixels. Maximum Depth: Maximum water surface elevation value, defined as the 95th percentile to exclude extreme erroneous values and outliers. Distance: The linear distance (meters) from the health facility (patch center point) to the pixel with the value of the 95th percentile.

All units for flood-depth-related descriptors are meters relative to EGM2008 geoid. Each flood descriptor includes an additional field with an estimation of the uncertainty propagated from the wse_uncert field using empirical formulas and bootstrap resampling.

Data Format 

GeoJSON files.

Source Data 

Surface Water and Ocean Topography (SWOT) mission frames of L2_HR_Raster product, along with Health Facilities of HSF Master Facility List (World Health Organization) that are typically unaffected by flooding—based on VIIRS-detected flood events.

Limitations/ Assumptions 

Main limitations of this approach arise from the propagation of noise measurements throughout the workflow, the irregular SWOT revisit time, and the intrinsic characteristics of SWOT observations. Radar imaging artefacts, such as layover effects and complex scattering over heterogeneous water surfaces, can introduce errors in water surface elevation measurements. One of the main intrinsic limitations of SWOT data is the reduced data quality in the nadir (inner swath) region due to high amount of interferometric errors. Although quality filtering based on uncertainty metrics and ancillary flags were applied to mitigate these effects, residual outliers and biases may persist and propagate through the analysis, potentially affecting the estimated temporal variations.

### 字段名称 | 描述
用例名称 | 南苏丹洪水与医疗服务中断事件
数据集名称 | 基于SWOT的洪水描述符
数据集描述 | 本数据集包含基于表面水与海洋地形（Surface Water and Ocean Topography, SWOT）得到的洪水统计数据，统计范围覆盖南苏丹境内189.856 km²的关注区域内的所有医疗设施（Health Facility, HF）。洪水统计基于2023-2025年的SWOT栅格帧计算得到，计算前已完成质量过滤步骤，以剔除以下像素：a) 水面高程（Water Surface Elevation, WSE）测量质量降级或不合格的像素；b) 包含"value_bad"、"geolocation_qual_degraded"、"classification_qual_degraded"字段激活标记的像素；c) 超出95百分位数的叠掩系统误差像素；d) 跨测绘带两侧的跨轨距离不在10-60 km范围内的像素；e) 暗水体占比大于0.2的像素；f) 水体占比低于0.15的像素。洪水描述符以279个医疗设施周边的方形缓冲区为计算单元，其中初级卫生保健单元（Primary Health Care Unit, PHCU）的缓冲区大小为5.4×5.4 km，初级卫生保健中心（Primary Health Care Center, PHCC）的缓冲区大小为3.6×3.6 km。每个洪水描述符均附带一个额外字段，用于估算从wse_uncert字段传递而来的不确定性。每个文件对应单个SWOT栅格帧，其覆盖范围内的每个医疗设施对应一条记录。输出文件保留对应SWOT栅格帧的原始文件名。
时间范围 | 2023-2025年
空间范围 | 南苏丹（采用EPSG:4326坐标参考系， bounding box为东经28.5942°至4.8367°，北纬31.8383°至9.6283°）
关键变量 | 1. 水体占比：缓冲区内被淹没像素数与总像素数的比值（取值范围0-1）；2. 平均水深：已识别淹没区域的平均水深；3. 水面高程均值：已识别淹没像素的平均水面高程；4. 水深中位数：已识别淹没像素的水面高程中位数；5. 最大水深：以95百分位数界定的最大水面高程值，用于排除极端错误值与异常值；6. 距离：医疗设施（斑块中心点）至95百分位数对应像素的直线距离（单位：米）。所有与洪水深度相关的描述符单位均为相对于EGM2008大地水准面的米数。每个洪水描述符均附带一个额外字段，通过经验公式与bootstrap重采样法估算从wse_uncert字段传递而来的不确定性。
数据格式 | GeoJSON文件
源数据 | 表面水与海洋地形（Surface Water and Ocean Topography, SWOT）卫星任务的L2_HR_Raster产品数据，以及世界卫生组织（World Health Organization, WHO）HSF主设施列表中的医疗设施数据——这类设施基于可见红外成像辐射计套件（Visible Infrared Imaging Radiometer Suite, VIIRS）探测到的洪水事件，通常被判定为未受洪水影响。
局限性与假设 | 本方法的主要局限性源于工作流中测量噪声的传递、SWOT不规则的重访周期，以及SWOT观测的固有特性。雷达成像伪影（如叠掩效应与非均质水体表面的复杂散射）可能会引入水面高程测量误差。SWOT数据的主要固有局限性之一是星下点（测绘带内侧）区域因存在大量干涉误差而导致数据质量下降。尽管已基于不确定性指标与辅助标记完成质量过滤以缓解这些影响，但残余异常值与偏差仍可能持续存在并在分析过程中传递，进而可能对估算的时间变化产生影响。