# Site02 soil and erosion analysis — local review edition

Open [Soil & Erosion](Site02_Soil_Erosion.html), or start at the [cover](index.html). This edition is local only and has not been published. The previous atlas is preserved separately.

## Evidence and scale

The official NSW Soil Landscapes of Central and Eastern NSW service identifies three landscape units inside the project rectangle: Winnunga (52.2702 ha; 76.6%), Paddy's River variant c (11.5980 ha; 17.0%), and Queanbeyan (4.3379 ha; 6.4%). Their union covers the 68.2060 ha rectangle without gaps. These GIS areas include the river, and the rectangle is a research reference rather than a cadastral boundary. The Canberra source mapping is 1:100,000, so it cannot establish individual soil beds, urban fill, pavement or local bank material.

The original reports by B. R. Jenkins (2000), *Soil Landscapes of the Canberra 1:100 000 Sheet*, are retained in `Documents/Soil`. Report `8727wn.pdf` describes Winnunga; `8727pdc.pdf` describes the broader Paddy's River landscape and explicitly identifies variant c; `8727qn.pdf` describes Queanbeyan. The website summaries concern regional landscape associations. They are not findings from soil sampling at Site02. Report page 1 contains the summaries used; the full reports contain soil materials and limitations.

Source links, retrieved 23 September 2026:

- [Soil landscapes metadata](https://datasets.seed.nsw.gov.au/dataset/published-soil-landscapes-of-central-and-eastern-nsw37d37)
- [Soil landscape feature service](https://mapprod1.environment.nsw.gov.au/arcgis/rest/services/Soil/SoilLandscapes_CentralAndEasternNSW/MapServer/0)
- [Winnunga original report](https://espaderesources.environment.nsw.gov.au/resources/SPADE/reports/8727wn.pdf)
- [Paddy's River / variant c original report](https://espaderesources.environment.nsw.gov.au/resources/SPADE/reports/8727pdc.pdf)
- [Queanbeyan original report](https://espaderesources.environment.nsw.gov.au/resources/SPADE/reports/8727qn.pdf)
- [Modelled Hillslope Erosion over NSW](https://datasets.seed.nsw.gov.au/dataset/modelled-hillslope-erosion-over-new-south-wales)
- [Official erosion raster service](https://mapprod3.environment.nsw.gov.au/arcgis/rest/services/Soil/NSW_Annual_Hillslope_Erosion_Modelling/MapServer)

NSW source soil information is attributed to the State of NSW / DCCEEW. eSPADE describes its soil information licence as CC BY 4.0; retain attribution and individual document copyright notices. See [eSPADE](https://espade.environment.nsw.gov.au/). The service feature version/date and package metadata are retained; a compilation release date does not establish that a soil survey was conducted in that year.

## Regional model sampling

The erosion service was queried at 396 regular grid centres, spaced 100 m in EPSG:28355. Coordinates range from E702250 to E704350 and N6085050 to N6086750. Each identify request used `tolerance=0`, projected coordinates, a 2100×1800 image display and the stated context extent. Source layers: 49 (K), 50 (LS), 22 (2023 covered erosion), and 26 (2023 bare erosion). All samples returned covered-model values; original responses and coordinates are in `Metadata/Soil/soil_raster_samples.json`.

These are raster-service samples, not a native-resolution raster download. The 100 m squares in the interface are display supports around sample centres. They must not be treated as original raster pixels, newly mapped soil polygons or uniformly eroding patches. No interpolation is applied between samples. The point GeoJSON is the authoritative downloaded form; a separate display-grid layer explicitly describes its presentation-only role.

K is modelled soil erodibility in metric RUSLE units (t·ha·h/(ha·MJ·mm)); LS is dimensionless. Covered and bare erosion values are regional model results in t/ha/yr, not measured site erosion. The service currently exposes annual layers through 2023; this edition intentionally uses that model year, not a claim about current 2026 conditions.

## Scenario calculation

For a valid sample with positive bare-soil model value:

`q2023 = Acovered2023 / Abare2023`

`qscenario = min(1, q2023 × cover multiplier)`

`Ascenario = selected erosivity multiplier × Abare2023 × qscenario`

Bare-soil reference sets qscenario=1. Protection uses a cover-effect multiplier of 0.5; disturbance uses 5; the 2023 model condition uses 1. Those alternative cover multipliers are authored assumptions, not measured vegetation-cover percentages or confirmed restoration effects. Erosivity is varied between 0.5 and 2 times the source model condition. This is not a rainfall-depth forecast. For a valid zero/zero pair the output is zero; missing or inconsistent inputs remain unknown.

The source model already includes its topographic LS factor. The separately derived local slope is not multiplied into the regional model again. The covered/bare ratio is interpreted as a combined model cover effect, without claiming a measured C factor.

Severity bins are fixed, author-selected display bands: very low <0.5, low 0.5–2, moderate 2–5, high 5–10, very high ≥10 t/ha/yr. They are not statutory or engineering thresholds. Playback sweeps the assumed erosivity input and reverses at the endpoints. It does not animate measured historical erosion, a storm hydrograph, future years, sediment transport or river migration. Results are hypothetical annualised potentials with the source model units. No site-wide tonnes/year total, loss of soil depth or affected area is inferred from this sampling grid.

Mapped-water sample centres are excluded from soil scenario summaries and exported scenario values. Summary counts use valid non-water sample centres inside the redline, not area weights. Urban hard surfaces and buildings are not reliably resolved in the regional model; the modelled values cannot establish actual erodible soil at each point.

## Terrain and river-edge inspection

Complete 5×5 groups of the source ACT 2015 1 m DEM are averaged to 5 m. Centred elevation differences produce slope in degrees; missing ground and missing required neighbours stay NoData. Median and 90th-percentile slope inside complete site cells are reported in `Metadata/soil_data_qa.json`; these include mapped river terrain, not just river banks. The input vertical datum remains unverified. Slope calculations assume the recorded horizontal and vertical units are metres; no absolute AHD elevations are asserted.

River-edge inspection candidates are sample centres outside mapped water, at most 50 m from the mapped water boundary, with at least one valid 5 m slope cell ≥10° within a 105×105 m sampling window. These author-selected thresholds flag terrain to inspect. They do not diagnose erosion or instability. There is no hydraulic shear stress, cohesive strength, root reinforcement, bathymetry, bank undercutting, bridge scour or sediment-transport model.

The separate schematic distinguishes hillslope wash from bank erosion. It is not to scale and is not a surveyed Site02 section. Regional reports discuss gully erosion, but RUSLE output is not a gully or riverbank erosion estimate.

## Soil questions remaining open

No site-specific spatial data for pH, texture fractions, permeability, infiltration, salinity, sodicity, organic carbon, contamination or geotechnical bearing capacity have been acquired. Qualitative soil descriptions are not lab results. Recommended next investigations are a ground-surface survey, soil augering/material logging, texture and dispersion testing, documentation of fill and drainage outlets, and bank-condition/root surveys. Those recommendations are proposed investigations, not completed studies.

## Reuse

- `GIS/GeoJSON/Soil/Soil_Landscapes_Study.geojson`: source polygons clipped to the display context, with provenance and site-intersection area.
- `GIS/GeoJSON/Soil/Official_Erosion_Raster_Samples_100m.geojson`: source-derived point values and terrain screening fields.
- `GIS/GeoJSON/Soil/Sample_Display_Grid_100m.geojson`: display supports only.
- `GIS/GeoJSON/Soil/River_Edge_Inspection_Candidates.geojson`: conceptual screening locations.
- `GIS/Soil_Slope_5m_degrees.asc` and `.prj`: 5 m slope grid, EPSG:28355. NoData=-9999.
- `Tables/Soil_Erosion_Samples.csv` and `Soil_Landscape_Areas.csv`.
- Export the selected scenario directly from the soil page; the CSV records its input multipliers and evidence status.

The previous hydrology/ecology and mobility datasets remain intact. New soil layers are delivered as GIS files and are not inserted into the copied ArcGIS project. The copied APRX retains its original references to the previous English package. The website and supplied relative file links work together in this extracted review folder.


## Georeferenced imagery overlay

The soil page now opens with a local NSW aerial/satellite imagery basemap. All four evidence layers — soil landscapes, erosion severity, K and terrain slope — support analysis-layer opacity from 0% to 100% (default 45%). A plain basemap remains available. Redline and location labels remain opaque. Set opacity to 0% for an imagery-only location check; choose 100% and a plain background to compare unblended category colours with the legend.

Use Site view or Context view, the zoom buttons, or Centre on sample. Drag the map to pan; keyboard focus on the map enables + / − and arrow-key navigation. The four optional place labels come from the existing atlas OpenStreetMap inventory, transformed from that map's local origin into the soil map coordinates. They are mapped facilities, not people-count observations.

Source: [NSW Spatial Services NSW_Imagery service](https://maps.six.nsw.gov.au/arcgis/rest/services/public/NSW_Imagery/MapServer). Retrieved 23 September 2026. Attribution returned by the service: © Department of Customer Service 2020. Service metadata also credits © AAM 2011, 2012; © Jacobs Group Ausimage 2002, 2006, 2009, 2014; © LANDSAT 2014. Their inclusion in the service description does not establish which supplier produced each local pixel. The original service metadata is retained.

The imagery service export requested both the input and output CRS as GDA94 / MGA zone 55, EPSG:28355, with bbox E 702200–704400 m and N 6085000–6086800 m. The returned extent exactly matches the soil map's 2200 × 1800 m domain. The SVG image is placed at local (0,0), width 2200 m, height 1800 m. Its 3300 × 2700 output pixels correspond to 0.6667 m output spacing; this is an export sampling choice, not a claim of native resolution or positional accuracy.

The [official imagery-date footprint service](https://maps.six.nsw.gov.au/arcgis/rest/services/public/NSW_Imagery_Dates/MapServer/0) returns one intersecting footprint: CANBERRA, SKM_Regional, block start date 21 February 2014, reported resolution 10 cm. The date/resolution describe the footprint metadata; they do not independently verify the provenance, acquisition day or effective resolution of every pixel in the exported mosaic. The basemap is older than the 2023 soil model. Current infrastructure, vegetation, land use and river conditions may differ. Retrieval date must not be treated as capture date.

This basemap aids orientation only. The soil-landscape source remains 1:100,000; erosion values remain point samples every 100 m. Detailed roofs or roads in the image cannot support more precise soil boundaries, measured erosion rates, bank change or contamination conclusions. Extent and CRS matching do not certify survey-level registration.

Files:

- `Maps/Soil/NSW_Imagery_Soil_Context.jpg`, `.jgw`, `.prj`: image, pixel-centre world file and CRS. Keep all three together when adding the JPEG in GIS.
- `Maps/Soil/Soil_Imagery_Overlay_Review.png`: standalone soil/erosion comparison at 45% overlay opacity, inspected as a rendered map; it is not a browser screenshot.
- `Metadata/Soil/soil_imagery_metadata.json`, `Soil_Imagery_Export_Response.json`, `Imagery_Date_Footprints.json`: provenance and original source responses.
- `Metadata/Soil/soil_imagery_context.json`: display placement and source feature identifiers for the location labels.
- `Metadata/Soil/soil_imagery_qa.json`: extent, pixel-spacing and world-file checks.

The image is bundled with the website; the map makes no runtime imagery-service request. If the local file is missing, an explicit status message appears and the soil analysis remains available on a plain map. This update is included in the online research edition.