Science & sources
The data, software, equations, and screening methods behind LUNA. Primary datasets are listed by source-file size; supporting references and tools follow by role.
01 / Source data
Largest source files first. Sizes describe acquired source files, not the smaller crops or derived samples that may be bundled with the app.
Regional elevation raster used for curved-surface horizon and terrain visibility calculations. Acquired source: 189,158,392 bytes (189.2 MB).
Native-resolution Shackleton rim surface used for the local horizon override and landing-slope/roughness screening. Acquired source: 40,980,806 bytes (41.0 MB); the app also retains a small crop.
Product notes: NASA GSFC PGDA product 78 · README and frame/units
Global 16 pixels-per-degree topography used for lunar globe context and terrain mesh. Acquired raster: 33,177,600 bytes (33.2 MB), 0.5 m height scale, 1,737,400 m reference radius.
PDS label: LDEM_16 label
Planetary translation kernel used with SPICE for Sun, Earth, and body geometry. Acquired file: 32,726,016 bytes (32.7 MB).
Long-term Earth-orientation kernel used for station-frame geometry. Acquired file: 31,318,016 bytes (31.3 MB); long-range predicted orientation has lower accuracy.
NASA FDF/GSFC ACTUAL spacecraft trajectory for the November 2022 eclipse example; coverage is 2022-09-15 to 2022-12-15. Acquired file: 7,658,496 bytes (7.66 MB).
Coverage label: SPK label
Dated Earth-orientation prediction kernel used within its coverage. Acquired file: 5,160,960 bytes (5.16 MB).
Lunar orientation kernel used with DE421 and the MOON_ME frame. Acquired file: 1,770,496 bytes (1.77 MB).
Planetary constants used by the SPICE geometry setup. Acquired file: 131,226 bytes.
Station frame definition paired with the ITRF93 coordinates. Acquired file: 69,400 bytes.
ITRF93 station coordinates for DSS-14, DSS-43, and DSS-63. Acquired file: 26,624 bytes.
DE421 lunar frame definition used to interpret orientation data. Acquired file: 21,437 bytes.
NAIF time kernel for UTC-to-ET conversion. Acquired file: 5,257 bytes.
Network-served map context, loaded by requested map area rather than as one fixed-size dataset: LRO WAC global mosaic, LOLA color shade, LOLA shade, and LOLA color slope layers.
Layers: LRO_WAC_Mosaic_Global_303ppd_v02, LRO_LOLA_ClrShade_Global_256ppd_v06, LRO_LOLA_Shade_Global_128ppd_v04, LRO_LOLA_ClrSlope_Global_16ppd.
API requests and responses are retained as reference fixtures for solar and Earth geometry. The interactive mission-analysis calculations use local SPICE kernels.
Reviewed but not ingested as numerical inputs: PGDA product 90 (20 m DEM listed at 2.6 GB), the alternative 5 m south-pole mosaic, and long-term polar illumination averages. The 2.6 GB product is not downloaded or analyzed by this project.
02 / Context and verification
Primary sources and NASA-authored materials used for examples, map/visual attribution, and model context.
03 / Implementation
Libraries and APIs used to build, run, or author the project. Lockfiles record resolved versions and transitive packages.
04 / Quantitative models
Equations reflect the implemented screening models. Inputs are configurable where noted; numerical sampling and physical assumptions limit interpretation.
Swipe to browse, or use the arrows and keyboard ← → keys.
At latitude φ and longitude λ, define local up, east, and north. The observer sits at reference lunar radius plus terrain elevation and mast height.
v is the observer-to-body vector, d its distance, and rbody the body radius. Geometry uses SPICE light-time plus stellar aberration (LT+S).
The horizon at each bearing is the maximum sampled elevation. Radial spacing is 5 m to 2 km, 20 m to 5 km, then 80 m to 250 km. Missing terrain stays unknown.
For separation d and apparent radii r1, r2, the overlap gives the blocked solar fraction. Terrain visibility uses 512 deterministic disk samples; the joint nominal fraction is a product approximation.
A, efficiency η, derating δ, panel orientation, and direct-sun fraction come from the study configuration. Tracking arrays use unit incidence.
Capacity C bounds the stored energy. Activity start and end times split intervals exactly; generation is held constant between samples.
This simplified free-space budget uses editable hardware, rate, losses, and margin assumptions. A positive margin does not guarantee geometric access or DSN availability.
The site screen fits a plane to native DEM pixels. Eight-neighbor A* routing excludes unknown, over-slope, and prohibited cells, then accounts for uphill effort and turns.
Vacuum, impulsive hop under constant lunar gravity. This is a screening estimate, not a trajectory or a model of a specific vehicle.
The user-defined reference orbit uses altitude, initial phase, inclination, and RAAN. It is distinct from the NASA LRO definitive-trajectory example.
05 / How results are produced
LUNA is a research-planning and education aid. Results screen candidate conditions; they are not landing clearance, guaranteed communications, or flight-qualified predictions.
NASA LOLA raster cells are sampled in their lunar frame; NASA NAIF kernels provide ephemerides, orientation, time conversion, and station locations. The study export records inputs, source manifest, kernel metadata, and code hashes.
UTC is converted to SPICE ET. Local Sun, Earth, LRO, and station geometry is evaluated in supported frames with LT+S where used. Surface visibility accounts for the local terrain horizon; contact windows are found from sampled sign changes and refined by bisection.
Terrain-height perturbations create deterministic sensitivity bounds, not statistical confidence intervals. Sunlight samples the apparent solar disk; the joint eclipse/terrain nominal fraction is a product approximation. Unknown terrain remains marked unknown.
Solar production and overlapping activity loads feed a bounded battery model. DSN visibility requires both lunar antenna elevation and the terrestrial station mask; RF margin is evaluated separately. Candidate activity starts are sample-aligned and recomputed against conservative state of charge.
Site04 footprint cells are fit to a plane for slope and residual roughness screening. Rover routing uses an eight-neighbor A* grid, slope and polygon constraints, uphill-effort cost, and turn penalties. Finite grid resolution can miss local hazards.
The globe display is a textured sphere, not a terrain-displaced mesh. Horizon results depend on finite DEM coverage and sampling; the far-field +15 km limit is an explicit assumed height. The Sun is uniform, Earth spherical, and neither atmospheric refraction nor limb darkening is modeled. No DSN reservation, weather/interference model, detailed thermal/battery-management model, rock-scale hazard map, landing dynamics, or mission hardware validation is included.