Crate kriging_rs 

Geostatistical kriging library with ordinary and binomial kriging, variogram fitting, and optional WASM and GPU support.

This crate provides spatial interpolation via ordinary kriging and prevalence-surface estimation via binomial kriging. It includes empirical variogram computation, parametric model fitting, and Haversine-based geographic coordinates. Build with wasm for browser bindings or gpu for GPU-accelerated batch prediction.

Quick example

use kriging_rs::{GeoCoord, GeoDataset, OrdinaryKrigingModel, VariogramModel, VariogramType};

let coords = vec![
    GeoCoord::try_new(0.0, 0.0)?,
    GeoCoord::try_new(0.0, 1.0)?,
    GeoCoord::try_new(1.0, 0.0)?,
];
let values = vec![1.0, 2.0, 1.5];
let dataset = GeoDataset::new(coords, values)?;
let variogram = VariogramModel::new(0.01, 2.0, 300.0, VariogramType::Exponential)?;
let model = OrdinaryKrigingModel::new(dataset, variogram)?;
let prediction = model.predict(GeoCoord::try_new(0.3, 0.3)?)?;

Module organization

• kriging — Ordinary kriging (OrdinaryKrigingModel, Prediction) and binomial kriging (BinomialKrigingModel, BinomialObservation, etc.) for spatial interpolation and prevalence surfaces.
• variogram — Empirical variogram (compute_empirical_variogram), fitting (fit_variogram), and parametric models (VariogramModel, VariogramType).
• distance — GeoCoord and Haversine distance.
• geo_dataset — Coordinate–value datasets (GeoDataset).
• error — KrigingError.
• wasm (optional, wasm feature) — Browser-facing WASM bindings.
• gpu (optional, gpu feature) — GPU backend and batch covariance helpers.

Features

Default build has no WASM or GPU. Enable with:

• wasm — WASM bindings and browser support.
• gpu — WebGPU-based batch prediction (native and web).
• gpu-blocking — Blocking GPU helpers on native (includes gpu).

Re-exports

pub use utils::Probability;

pub use utils::clamp_probability;

pub use utils::logistic;

pub use utils::logit;

pub use utils::logit_clamped;

pub use distance::GeoCoord;

pub use error::KrigingError;

pub use geo_dataset::GeoDataset;

pub use kriging::binomial::BinomialKrigingModel;

pub use kriging::binomial::BinomialObservation;

pub use kriging::binomial::BinomialPrediction;

pub use kriging::binomial::BinomialPrior;

pub use kriging::ordinary::OrdinaryKrigingModel;

pub use kriging::ordinary::Prediction;

pub use variogram::fitting::FitResult;

pub use variogram::fitting::fit_variogram;

pub use variogram::models::VariogramModel;

pub use variogram::models::VariogramType;

pub use variogram::PositiveReal;

pub use variogram::VariogramConfig;

pub use variogram::compute_empirical_variogram;

Modules

distance

error

geo_dataset

Coordinate–value datasets for kriging (pairs of locations and observed values).

kriging

Kriging models for spatial interpolation and prevalence surfaces.

matrix

utils

variogram

Variogram computation, fitting, and parametric models.

Type Aliases

Real

Floating-point type used for coordinates, values, and variogram parameters; currently f32.