Scope and limitations#

Version 0.3.0 contains plate_tectonics, seismology, atmosphere, climate, oceanography, geophysics, volcanology, hydrology and glaciology. Its data are deterministic synthetic examples, not bundled observational products. It provides educational baselines with documented assumptions, not an integrated Earth-system model.

  • Plate kinematics use rigid spherical geometry. There is no plate catalogue, reconstruction inversion or full mantle-convection model.

  • Seismology has homogeneous and single-layer elastic helpers, a ray parameter and a vertical travel-time integral, and a single oscillator. There is no ray tracer, 3D wave propagation, rupture inversion or earthquake forecasting.

  • Atmospheric ascent is dry and idealized. There is no moisture-adjusted parcel solver or weather assimilation. The barotropic vorticity model is a periodic beta-plane teaching model, not numerical weather prediction.

  • Climate uses grey radiation and linear one-box response. The carbon and sea-level helpers are deliberately simplified budget examples.

  • Ocean density is locally linear or the simplified Roquet et al. (2015) fit, not the full TEOS-10 polynomial. Currents omit full primitive-equation dynamics and site-specific boundary conditions; the gyre is Stommel’s linear rectangular-basin solution and the thermohaline model has two boxes.

  • Geophysics uses first-order, spherical or planar reductions without terrain corrections, zonal (axisymmetric) harmonics only, a geocentric axial dipole for paleomagnetic directions, two-dimensional vertically magnetized blocks and the Love numbers of a uniform elastic sphere. There is no gravity or magnetic inversion, full IGRF evaluation or geoid computation from data, and the Rikitake dynamo is a chaotic analogue, not a model of the core.

  • Volcanology dating assumes closed systems with an unweighted isochron fit and uncalibrated radiocarbon ages. Geochemistry uses constant fractionation and partition coefficients. Magma is a crystal- and bubble-free Newtonian melt in laminar dike flow, the Mogi source is a point in an elastic half-space, and the eruption column is a windless top-hat plume without a gas-thrust region, particle fallout or moisture. There is no thermodynamic melting model, conduit-flow solver, tephra dispersal or hazard assessment.

  • Hydrology uses homogeneous, isotropic media and lumped or one-dimensional forms: Dupuit flow without recharge, Theis wells in an infinite confined aquifer, steady Richards profiles in a Gardner soil, Green-Ampt and Horton infiltration in a uniform soil, steady uniform channel flow, a linear time-invariant unit hydrograph, Muskingum routing and a big-leaf Penman-Monteith canopy in neutral stability. There is no transient Richards solver, groundwater model, kinematic or dynamic wave routing, snowmelt, catchment calibration or flood-frequency analysis.

  • Glaciology uses isothermal Glen ice with one rate factor, shallow-ice and plastic profiles, a power-law sliding law, a flowline glacier of uniform width with no sliding or calving, local isostasy with a single viscous half-space relaxation, Stefan growth without snow or ocean heat, and steady one-dimensional ice-core flow. There is no thermomechanical coupling, higher-order or full-Stokes flow, ice-shelf buttressing, firn densification model, positive-degree-day melt or three-dimensional ice sheet.

Physical constants and sample coefficients are documented references, not a service that automatically tracks revised measurements. Raw floats do not carry unit metadata. Algebraic broadcasting can allocate large arrays; callers manage memory. The shared tolerance wrapper uses scalar tolerances. Python 3.12 is the locally verified runtime; the CI matrix is configured for 3.10-3.13.