Structured product record

TerrainLogic technical reference.

A concise, extractable record of current capabilities, tested hardware, validation statistics, and the boundaries of every claim.

Open llms.txt ↗Download structured data

01 / Capabilities

One local-first field model.
Connected planning workflows.

TerrainLogic links field selection, public data, management, terrain, daily model state, engineering context, and reporting without requiring a cloud account.

Automated site characterization

Location, county, soils, climate context, elevation, slope, and flow context from a selected boundary.

Erosion and management

Daily RUSLE2-aligned factors, annual A versus T, exact operations, scenario comparison, and retained evidence.

Nutrient accounting

Planning-level N and P additions, uptake, export, residue and root returns, fixation, transformations, losses, and state.

Terrain screening

Potential concentrated-flow corridors with contributing area, slope, length, severity, and DEM context.

Preliminary engineering

Grassed-waterway and terrace planning, calculation checks, survey workflows, inspections, and documented outputs.

Stateful fields and reports

Managed history, carried daily state, local portfolio tools, custom fields, and auditable PDF records.

Boundary of use

TerrainLogic supports planning and evaluation. Professional judgment, field verification, applicable criteria, hydrology, survey, permits, utilities review, and certification remain with qualified professionals and responsible agencies.

02 / System requirements

Designed for an ordinary Windows field computer.

Tested low-end baseline

Successful complete workflows

Processor: Intel Core i3-10110Y @ 1.00 GHz
Memory: 8 GB RAM
Graphics: Intel UHD integrated
Storage: 128 GB SSD
System: Windows 11, 64-bit

Minimum supported

Normal workflows

Windows 11, 64-bit
Intel Core i3-class or comparable AMD CPU
8 GB RAM
SSD with at least 5 GB free
Current Microsoft Edge WebView runtime
Broadband for initial public-data retrieval
No dedicated GPU required

Recommended

Faster and larger work

Intel Core i5-class or comparable AMD CPU
16 GB RAM
SSD storage
Windows 11, 64-bit

Cached site data can support later offline work.

Observed low-end memory: approximately 215–250 MB for routine analyses, briefly near 600 MB for an intensive terrace DEM scan, with memory released after processing.

03 / Validation

Evidence classified by what it actually proves.

Implementation verification, reference-model parity, and independent observations are reported separately.

Erosion benchmark749completed RUSLE2 parity cases

Strict agreement80.1%within 20% of RUSLE2

Planning tolerance94.4%within 20% or 0.25 ton/ac/yr

Engineering suite50 / 50equation and invariant checks passed

Erosion parity

Pearson r 0.9993 · R² 0.9986

National mean bias error is +0.475 ton/ac/year. This measures agreement with RUSLE2, not independent measured-plot accuracy.

Grain nitrogen

112 independent paired rows

PBIAS +6.60%, RMSE 19.53 kg N/ha, Spearman 0.850, and NSE 0.790.

Soil-derived phosphorus

15 independent records

PBIAS -12.61%, correlation 0.860, and NSE 0.580. Applied-P validation remains incomplete where required drivers are missing.

Surface urea NH₃

18 complete site-years

PBIAS +3.77%, MAE 6.01 kg N/ha, RMSE 7.14 kg N/ha, and NSE 0.485. This is an input-limited candidate.

Read the full public science and validation record ↗

04 / Machine-readable resources

Stable files for indexing and evaluation.

Each JSON document uses a versioned top-level schema and links back to its canonical human-readable source.

Text