Operational SAR Flood Mapping as a Full-Stack Systems Problem: An AI-Enabled Perspective

This repository contains the evaluation scripts, workflow notes, and derived summary outputs used to support the manuscript:

Operational SAR Flood Mapping as a Full-Stack Systems Problem: An AI-Enabled Perspective

The materials document the comparative evidence package used to evaluate the SIENA flood-mapping workflow across four evaluation settings:

1. West Coast atmospheric-river case (December 2025)

   Comparison of SIENA against earlier operational candidate algorithms using NOAA VIIRS Flood Inundation Maps (VIIRS-FIM) as a common coarse reference.

   This workflow includes: per-granule binary flood metrics; false-alarm behavior analysis; process-time extraction and visualization; manuscript-ready summary outputs.

2. Kentucky flood case (April 2025)

   Comparison of SIENA against NOAA Emergency Response Imagery (ERI) airborne RGB mosaics for event-specific spatial consistency analysis.

   This workflow includes: tile-based SIENA vs ERI comparison; coverage-aware confusion statistics; area-fraction analysis; curated representative figure selection; manuscript-ready summary outputs.

3. Central Valley, California multi-date case (winter 2025–2026)

   Three-way comparison among SIENA, DSWX-HLS, and NOAA VIIRS-FIM over a fixed regional domain.

   This workflow includes: pairwise inter-product metrics; pooled and per-date summaries; flood-subset summaries; wet-fraction scatter analysis; curated comparison figures; manuscript-ready summary outputs.

4. Event-based SIENA / DSWX-S1 / VIIRS-FIM case (December 2025–May 2026)

   Event-based comparison among SIENA, DSWX-S1, and NOAA VIIRS-FIM across three representative flood episodes: the Snohomish River event near Seattle, the Jackson County, Indiana event, and the Alaska Yukon River event.

   This workflow includes: pooled per-event inter-product metrics; flood-subset summaries; event-window tracking; comparison summaries across multiple flood settings; manuscript-ready summary outputs.

Repository contents

The archive is organized into four main task folders:

SIENA_VIIRS_6Panels_metric/ — West Coast VIIRS-based evaluation and process-time analysis

SIENA_NOAA_eri_metric/ — Kentucky SIENA vs NOAA ERI evaluation

SIENA_DSWX-HLS_VIIRs_CA/ — Central Valley SIENA / DSWX-HLS / VIIRS three-way comparison

SIENA_VIIRS_DSWX-S1_events/ — Event-based SIENA / DSWX-S1 / VIIRS comparison across three representative flood episodes

Within these folders, the repository includes:

Python analysis scripts

shell wrappers for running the workflows

WORKFLOW.md files documenting methods, assumptions, thresholds, and outputs

plotting utilities

figure-curation utilities

scripts for generating Word-format summary documents

derived CSV summaries and figure products where included in the packaged archive

Purpose

These materials are intended to provide a transparent, manuscript-linked record of the comparative evaluation workflows used in the SIENA case study. The repository is not presented as a polished standalone software release for general operations. Instead, it serves as a research-support archive for the quantitative analyses, figure generation, and summary reporting referenced in the manuscript and supplementary materials.

Important notes

The workflows rely on external flood products and imagery sources, including NOAA VIIRS-FIM, NOAA ERI, DSWX-HLS, and DSWX-S1.

Some scripts assume access to local project directory structures, cached ancillary datasets, and previously generated intermediate products.

Thresholds, valid-domain rules, permanent-water masking conventions, and aggregation logic are documented in the included WORKFLOW.md files.

The repository emphasizes comparative evaluation and manuscript reproducibility support, rather than one-click reproduction in a fully containerized environment.

This archive contains the evaluation workflows, analysis scripts, plotting utilities, and derived summaries supporting the manuscript comparisons. It does not constitute a full release of the operational SIENA production system or all agent configurations.

Related manuscript use

This archive supports the manuscript’s comparative evidence package by documenting and preserving the workflows used to assess operational behavior, false-alarm discipline, inter-product consistency, event-specific spatial agreement, flood-setting-dependent comparison behavior, and runtime characteristics of the SIENA rebuild.

本仓库包含支撑以下学术论文的评估脚本、工作流程说明及衍生汇总成果：
《作为全栈系统问题的业务化合成孔径雷达（Synthetic Aperture Radar, SAR）洪水制图：一种赋能人工智能的视角》

本材料记录了用于在四类评估场景中验证SIENA洪水制图工作流程的对比验证数据集：

1. 西海岸大气河事件（2025年12月）
以NOAA可见光红外成像辐射仪（Visible Infrared Imaging Radiometer Suite, VIIRS）洪水淹没图（VIIRS-FIM）作为通用粗分辨率参考基准，将SIENA与早期业务候选算法进行对比。
本工作流程包含：逐粒元二进制洪水指标计算、虚警行为分析、处理时长提取与可视化、可直接用于论文的汇总成果。

2. 肯塔基州洪水事件（2025年4月）
以NOAA应急响应影像（Emergency Response Imagery, ERI）机载RGB镶嵌图为基准，开展SIENA的事件级空间一致性分析对比。
本工作流程包含：基于瓦片的SIENA与ERI对比、覆盖度感知的混淆统计分析、面积占比分析、精选代表性图表遴选、可直接用于论文的汇总成果。

3. 加利福尼亚中央谷多时段事件（2025年冬季—2026年冬季）
在固定区域范围内开展SIENA、DSWX-HLS与NOAA VIIRS-FIM的三方对比。
本工作流程包含：产品间两两对比指标、汇总及逐日期统计结果、洪水子集统计结果、水域占比散点分析、精选对比图表、可直接用于论文的汇总成果。

4. 基于事件的SIENA/DSWX-S1/VIIRS-FIM对比（2025年12月—2026年5月）
在三个典型洪水事件中开展SIENA、DSWX-S1与NOAA VIIRS-FIM的事件级对比：西雅图附近的斯诺霍米什河事件、印第安纳州杰克逊县事件以及阿拉斯加州育空河事件。
本工作流程包含：逐事件汇总的产品间对比指标、洪水子集统计结果、事件窗口追踪、多洪水场景下的对比汇总结果、可直接用于论文的汇总成果。

## 仓库内容说明
本归档文件分为四个主要任务文件夹：
- SIENA_VIIRS_6Panels_metric/ — 基于VIIRS的西海岸评估与处理时长分析
- SIENA_NOAA_eri_metric/ — 肯塔基州SIENA与NOAA ERI对比评估
- SIENA_DSWX-HLS_VIIRs_CA/ — 中央谷SIENA、DSWX-HLS与VIIRS三方对比
- SIENA_VIIRS_DSWX-S1_events/ — 基于三个典型洪水事件的SIENA/DSWX-S1/VIIRS对比评估

在上述文件夹内，仓库包含以下内容：
Python分析脚本、用于运行工作流程的Shell封装脚本、记录研究方法、假设条件、阈值参数与输出结果的WORKFLOW.md文档、绘图工具集、图表遴选工具集、用于生成Word格式汇总文档的脚本、打包归档中包含的衍生逗号分隔值（Comma-Separated Values, CSV）汇总文件与图表成果。

## 用途说明
本材料旨在为SIENA案例研究中使用的对比评估工作流程提供透明可追溯且与论文关联的记录。本仓库并非面向通用业务场景的成熟独立软件发布包，而是作为支撑学术论文及补充材料中提及的定量分析、图表生成与汇总报告的研究支持归档文件。

## 重要说明
本工作流程依赖外部洪水产品与影像数据源，包括NOAA VIIRS-FIM、NOAA ERI、DSWX-HLS及DSWX-S1。
部分脚本要求访问本地项目目录结构、缓存辅助数据集以及预先生成的中间产物。
阈值、有效区域规则、永久水体掩膜约定与聚合逻辑已在附带的WORKFLOW.md文档中说明。
本仓库侧重对比评估与论文复现支持，而非提供可一键复现的全容器化运行环境。
本归档包含支撑论文对比分析的评估工作流程、分析脚本、绘图工具集与衍生汇总成果，并非完整发布的业务化SIENA生产系统或全部智能体配置。

## 论文支撑用途
本归档通过记录并保存用于评估SIENA重构版本的业务化运行行为、虚警管控能力、产品间一致性、事件级空间匹配度、不同洪水场景下的对比表现以及运行时特征的工作流程，为论文的对比验证数据集提供支撑。