数据分析
Identify and monitor aquaculture ponds from multi-temporal remote sensing. Detects coastal and inland ponds, computes area statistics, tracks expansion/abandonment, and analyzes conversion with wetlands/cropland. Use when mapping aquaculture distribution, monitoring pond dynamics, assessing environmental impacts, or generating pond inventories.
Identify and monitor aquaculture ponds from multi-temporal remote sensing. Detects coastal and inland ponds, computes area statistics, tracks expansion/abandonment, and analyzes conversion with wetlands/cropland. Use when mapping aquaculture distribution, monitoring pond dynamics, assessing environmental impacts, or generating pond inventories.
Identifies coastal and inland aquaculture ponds from multi-temporal satellite imagery. Separates aquaculture ponds from natural water bodies and paddy fields using shape, texture, water frequency, and adjacency features. Outputs pond locations, aquaculture zones, change detection, and area statistics.
Use when the user wants to:
# Basic analysis with default parameters (rules method, single year)
python scripts/aquaculture_pond_mapping.py --output-dir ./apm-output
# Multi-year analysis with custom minimum area
python scripts/aquaculture_pond_mapping.py \
--years 2020 2021 2022 2023 \
--min-area 1000 \
--output-dir ./apm-output
# ML-based classification (requires training data)
python scripts/aquaculture_pond_mapping.py \
--method ml \
--training-data ./training_samples.geojson \
--output-dir ./apm-output
# With custom input NDWI stack
python scripts/aquaculture_pond_mapping.py \
--input-ndwi ./ndwi_stack.tif \
--years 2020 2021 2022 \
--output-dir ./apm-output
| Parameter | Default | Description |
|---|---|---|
--input-ndwi | None | Input NDWI time series GeoTIFF (multi-band, optional) |
--method | rules | Classification method: rules, ml |
--years | [2023] | Years for analysis (space-separated) |
--min-area | 500.0 | Minimum pond area in m² (default: 500) |
--output-dir | ./apm-output | Output directory |
| File | Description |
|---|---|
ponds.geojson | Point features for detected aquaculture ponds with confidence |
aquaculture_zones.geojson | Point features for aquaculture cluster zones |
change.geojson | Point features showing new/abandoned/stable ponds |
area_by_admin.csv | Area statistics per classification class |
accuracy.json | Classification summary and confidence metrics |
request.json | Analysis request metadata |
dataset-manifest.json | Input data and parameters manifest |
output-manifest.json | Output file inventory and area statistics |
qa.json | Quality assurance checks |
| Code | Name | Description |
|---|---|---|
| 1 | aquaculture_pond | Managed aquaculture pond (rectangular, embanked, clustered) |
| 2 | natural_water | Natural water body (irregular shape, permanent, no embankments) |
| 3 | paddy_field | Rice paddy (small, rectangular, seasonal flooding) |
| 0 | non_water | Non-water or below minimum area |
| Code | Name | Description |
|---|---|---|
| 1 | new_pond | Pond present in latest period but not baseline |
| -1 | abandoned_pond | Pond present in baseline but not latest |
| 0 | stable_pond | Pond present in both periods |
| Code | Meaning |
|---|---|
| 0 | Success |
| 2 | Argument error |
| 3 | Dependency missing |
| 6 | Data validation failure |
| 7 | Processing failure |
本 skill 可自动从 Microsoft Planetary Computer 下载数据 (无需 API key):
python aquaculture_pond_mapping.py --bbox 116,39,117,40 --date-range 2024-06-01,2024-06-30 --output-dir
--bbox W,S,E,N: WGS-84 边界框 (西, 南, 东, 北)--date-range START,END: 日期范围 (YYYY-MM-DD,YYYY-MM-DD)--aoi-file : 替代 --bbox 的 GeoJSON 多边形--cache-dir : 缓存目录 (默认 ~/.geoskill_cache)当用户只给 --bbox + --date-range (没有 --image) 时，skill 自动下载数据。
当用户给 --image 时，走原文件路径 (向后兼容)。
Monitor wetland extent, inundation frequency, and land cover transitions to identify degradation, recovery, and human encroachment. Use when the user wants to track wetland changes over time, assess wetland health, or detect conversion of wetlands to other land covers.
Identify major crop types from multi-temporal optical/SAR imagery using phenological features. Produces pixel/field-level classification, area statistics, and confidence maps. Use when mapping crop distributions, estimating planted areas, or generating agricultural intelligence from remote sensing data.
Detect and classify cropland changes from before/after NDVI rasters. Identifies suspected construction, water, forest, and bare soil changes and generates compliance investigation materials. Use when the user wants to check farmland for unauthorized changes, monitor cropland conversion, or generate compliance reports.
Monitor construction progress from multi-temporal satellite imagery. Classifies project stages (clearing, earthwork, foundation, structure, finishing, completed) using spectral indices and detects stagnation. Use when tracking infrastructure projects, auditing construction timelines, or generating progress reports from remote sensing data.
Multi-temporal shoreline change rate analysis. Generates transects, computes Endpoint Rate (EPR) and Linear Regression Rate (LRR), and identifies erosion hotspots. Use when monitoring coastline retreat/accretion or identifying erosion-prone segments.
Multi-year cropland abandonment detection using NDVI time series. Identifies suspected abandoned cropland based on consecutive years without cultivation signals. Use when screening for abandoned fields, generating field verification lists, or monitoring cultivation continuity.