Latest Release		Downloads
License		People
Build Status		Coverage

---

📍 Fast, Accurate Python library for Raster Operations

⚡ Extensible with Numba

⏩ Scalable with Dask

🎊 Free of GDAL / GEOS Dependencies

🌍 General-Purpose Spatial Processing, Geared Towards GIS Professionals

---

Xarray-Spatial implements common raster analysis functions using Numba and provides an easy-to-install, easy-to-extend codebase for raster analysis.

Installation

# via pip
pip install xarray-spatial

# via conda
conda install -c conda-forge xarray-spatial

Downloading our starter examples and data

Once you have xarray-spatial installed in your environment, you can use one of the following in your terminal (with the environment active) to download our examples and/or sample data into your local directory.

xrspatial examples : Download the examples notebooks and the data used.

xrspatial copy-examples : Download the examples notebooks but not the data. Note: you won't be able to run many of the examples.

xrspatial fetch-data : Download just the data and not the notebooks.

In all the above, the command will download and store the files into your current directory inside a folder named 'xrspatial-examples'.

xarray-spatial grew out of the Datashader project, which provides fast rasterization of vector data (points, lines, polygons, meshes, and rasters) for use with xarray-spatial.

xarray-spatial does not depend on GDAL / GEOS, which makes it fully extensible in Python but does limit the breadth of operations that can be covered. xarray-spatial is meant to include the core raster-analysis functions needed for GIS developers / analysts, implemented independently of the non-Python geo stack.

Our documentation is still under construction, but docs can be found here.

Raster-huh?

Rasters are regularly gridded datasets like GeoTIFFs, JPGs, and PNGs.

In the GIS world, rasters are used for representing continuous phenomena (e.g. elevation, rainfall, distance), either directly as numerical values, or as RGB images created for humans to view. Rasters typically have two spatial dimensions, but may have any number of other dimensions (time, type of measurement, etc.)

Supported Spatial Functions with Supported Inputs

✅ = native backend    🔄 = accepted (CPU fallback)

Classification · Diffusion · Focal · Morphological · Fire · Multispectral · Multivariate · Pathfinding · Proximity · Reproject / Merge · Raster / Vector Conversion · Surface · Hydrology · Flood · Interpolation · Dasymetric · Zonal · Utilities

---

GeoTIFF / COG I/O

Native GeoTIFF and Cloud Optimized GeoTIFF reader/writer. No GDAL required.

Name	Description	NumPy	Dask	CuPy GPU	Dask+CuPy GPU	Cloud
read_geotiff	Read GeoTIFF / COG / VRT	✅️	✅️	✅️	✅️	✅️
write_geotiff	Write DataArray as GeoTIFF / COG	✅️	✅️	✅️	✅️	✅️
write_vrt	Generate VRT mosaic from GeoTIFFs	✅️

read_geotiff and write_geotiff auto-dispatch to the correct backend:

read_geotiff('dem.tif')                              # NumPy
read_geotiff('dem.tif', chunks=512)                  # Dask
read_geotiff('dem.tif', gpu=True)                    # CuPy (nvCOMP + GDS)
read_geotiff('dem.tif', gpu=True, chunks=512)        # Dask + CuPy
read_geotiff('https://example.com/cog.tif')          # HTTP COG
read_geotiff('s3://bucket/dem.tif')                  # Cloud (S3/GCS/Azure)
read_geotiff('mosaic.vrt')                           # VRT mosaic (auto-detected)

write_geotiff(cupy_array, 'out.tif')                 # auto-detects GPU
write_geotiff(data, 'out.tif', gpu=True)             # force GPU compress
write_vrt('mosaic.vrt', ['tile1.tif', 'tile2.tif'])  # generate VRT

Compression codecs: Deflate, LZW (Numba JIT), ZSTD, PackBits, JPEG (Pillow), uncompressed

GPU codecs: Deflate and ZSTD via nvCOMP batch API; LZW via Numba CUDA kernels

Features:

• Tiled, stripped, BigTIFF, multi-band (RGB/RGBA), sub-byte (1/2/4/12-bit)
• Predictors: horizontal differencing (pred=2), floating-point (pred=3)
• GeoKeys: EPSG, WKT/PROJ (via pyproj), citations, units, ellipsoid, vertical CRS
• Metadata: nodata masking, palette colormaps, DPI/resolution, GDALMetadata XML, arbitrary tag preservation
• Cloud storage: S3 (s3://), GCS (gs://), Azure (az://) via fsspec
• GPUDirect Storage: SSD→GPU direct DMA via KvikIO (optional)
• Thread-safe mmap reads, atomic writes, HTTP connection reuse (urllib3)
• Overview generation: mean, nearest, min, max, median, mode, cubic
• Planar config, big-endian byte swap, PixelIsArea/PixelIsPoint

Read performance (real-world files, A6000 GPU):

File	Format	xrspatial CPU	rioxarray	GPU (nvCOMP)
render_demo 187x253	uncompressed	0.2ms	2.4ms	0.7ms
Landsat B4 1310x1093	uncompressed	1.0ms	6.0ms	1.7ms
Copernicus 3600x3600	deflate+fp3	241ms	195ms	872ms
USGS 1as 3612x3612	LZW+fp3	275ms	215ms	747ms
USGS 1m 10012x10012	LZW	1.25s	1.80s	990ms

Read performance (synthetic tiled, GPU shines at scale):

Size	Codec	xrspatial CPU	rioxarray	GPU (nvCOMP)
4096x4096	deflate	265ms	211ms	158ms
4096x4096	zstd	73ms	159ms	58ms
8192x8192	deflate	1.06s	859ms	565ms
8192x8192	zstd	288ms	668ms	171ms

Write performance (synthetic tiled):

Size	Codec	xrspatial CPU	rioxarray	GPU (nvCOMP)
2048x2048	deflate	424ms	110ms	135ms
2048x2048	zstd	49ms	83ms	81ms
4096x4096	deflate	1.68s	447ms	302ms
8192x8192	deflate	6.84s	2.03s	1.11s
8192x8192	zstd	847ms	822ms	1.03s

Consistency: 100% pixel-exact match vs rioxarray on all tested files (Landsat 8, Copernicus DEM, USGS 1-arc-second, USGS 1-meter).

---

Reproject / Merge

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Reproject	Reprojects a raster to a new CRS using an approximate transform and numba JIT resampling	Standard (inverse mapping)	✅️	✅️	✅️	✅️
Merge	Merges multiple rasters into a single mosaic with configurable overlap strategy	Standard (mosaic)	✅️	✅️	🔄	🔄

---

Utilities

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Preview	Downsamples a raster to target pixel dimensions for visualization	Custom	✅️	✅️	✅️	🔄
Rescale	Min-max normalization to a target range (default [0, 1])	Standard	✅️	✅️	✅️	✅️
Standardize	Z-score normalization (subtract mean, divide by std)	Standard	✅️	✅️	✅️	✅️

---

Surface

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Aspect	Computes downslope direction of each cell in degrees	Horn 1981	✅️	✅️	✅️	✅️
Northness	North-south component of aspect: cos(aspect) for linear models	Stage 1976	✅️	✅️	✅️	✅️
Eastness	East-west component of aspect: sin(aspect) for linear models	Stage 1976	✅️	✅️	✅️	✅️
Curvature	Measures rate of slope change (concavity/convexity) at each cell	Zevenbergen & Thorne 1987	✅️	✅️	✅️	✅️
Hillshade	Simulates terrain illumination from a given sun angle and azimuth	GDAL gdaldem	✅️	✅️	✅️	✅️
Roughness	Computes local relief as max minus min elevation in a 3×3 window	GDAL gdaldem	✅️	✅️	✅️	✅️
Sky-View Factor	Measures the fraction of visible sky hemisphere at each cell	Zakek et al. 2011	✅️	✅️	✅️	✅️
Slope	Computes terrain gradient steepness at each cell in degrees	Horn 1981	✅️	✅️	✅️	✅️
Terrain Generation	Generates synthetic terrain from fBm or ridged fractal noise with optional domain warping, Worley blending, and hydraulic erosion	Custom (fBm)	✅️	✅️	✅️	✅️
TPI	Computes Topographic Position Index (center minus mean of neighbors)	Weiss 2001	✅️	✅️	✅️	✅️
TRI	Computes Terrain Ruggedness Index (local elevation variation)	Riley et al. 1999	✅️	✅️	✅️	✅️
Landforms	Classifies terrain into 10 landform types using the Weiss (2001) TPI scheme	Weiss 2001	✅️	✅️	✅️	✅️
Viewshed	Determines visible cells from a given observer point on terrain	GRASS GIS r.viewshed	✅️	✅️	✅️	✅️
Min Observable Height	Finds the minimum observer height needed to see each cell (experimental)	Custom	✅️
Perlin Noise	Generates smooth continuous random noise for procedural textures	Perlin 1985	✅️	✅️	✅️	✅️
Worley Noise	Generates cellular (Voronoi) noise returning distance to the nearest feature point	Worley 1996	✅️	✅️	✅️	✅️
Hydraulic Erosion	Simulates particle-based water erosion to carve valleys and deposit sediment	Custom	✅️	✅️	✅️	✅️
Bump Mapping	Adds randomized bump features to simulate natural terrain variation	Custom	✅️	✅️	✅️	✅️

---

Hydrology

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Flow Direction (D8)	Computes D8 flow direction from each cell toward the steepest downhill neighbor	O'Callaghan & Mark 1984	✅️	✅️	✅️	✅️
Flow Direction (Dinf)	Computes D-infinity flow direction as a continuous angle toward the steepest downslope facet	Tarboton 1997	✅️	✅️	✅️	✅️
Flow Direction (MFD)	Partitions flow to all downslope neighbors with an adaptive exponent (Qin et al. 2007)	Qin et al. 2007	✅️	✅️	✅️	✅️
Flow Accumulation (D8)	Counts upstream cells draining through each cell in a D8 flow direction grid	Jenson & Domingue 1988	✅️	✅️	✅️	✅️
Flow Accumulation (Dinf)	Accumulates upstream area by splitting flow proportionally between two neighbors (Tarboton 1997)	Tarboton 1997	✅️	✅️	✅️	🔄
Flow Accumulation (MFD)	Accumulates upstream area through all MFD flow paths weighted by directional fractions	Qin et al. 2007	✅️	✅️	✅️	🔄
Flow Length (D8)	Computes D8 flow path length from each cell to outlet (downstream) or from divide (upstream)	Standard (D8 tracing)	✅️	✅️	✅️	🔄
Flow Length (Dinf)	Proportion-weighted flow path length using D-inf angle decomposition (downstream or upstream)	Tarboton 1997	✅️	✅️	✅️	🔄
Flow Length (MFD)	Proportion-weighted flow path length using MFD fractions (downstream or upstream)	Qin et al. 2007	✅️	✅️	✅️	🔄
Watershed	Labels each cell with the pour point it drains to via D8 flow direction	Standard (D8 tracing)	✅️	✅️	✅️	✅️
Basins	Delineates drainage basins by labeling each cell with its outlet ID	Standard (D8 tracing)	✅️	✅️	✅️	✅️
Stream Order	Assigns Strahler or Shreve stream order to cells in a drainage network	Strahler 1957, Shreve 1966	✅️	✅️	✅️	✅️
Stream Order (Dinf)	Strahler/Shreve stream ordering on D-infinity flow direction grids	Tarboton 1997	✅️	✅️	✅️	✅️
Stream Order (MFD)	Strahler/Shreve stream ordering on MFD fraction grids	Freeman 1991	✅️	✅️	✅️	✅️
Stream Link	Assigns unique IDs to each stream segment between junctions	Standard	✅️	✅️	✅️	✅️
Stream Link (Dinf)	Stream link segmentation on D-infinity flow direction grids	Tarboton 1997	✅️	✅️	✅️	✅️
Stream Link (MFD)	Stream link segmentation on MFD fraction grids	Freeman 1991	✅️	✅️	✅️	✅️
Snap Pour Point	Snaps pour points to the highest-accumulation cell within a search radius	Custom	✅️	✅️	✅️	✅️
Flow Path	Traces downstream flow paths from start points through a D8 direction grid	Standard (D8 tracing)	✅️	✅️	🔄	🔄
HAND	Computes Height Above Nearest Drainage by tracing D8 flow to the nearest stream cell	Nobre et al. 2011	✅️	✅️	🔄	🔄
TWI	Topographic Wetness Index: ln(specific catchment area / tan(slope))	Beven & Kirkby 1979	✅️	✅️	✅️	🔄

---

Flood

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Flood Depth	Computes water depth above terrain from a HAND raster and water level	Standard (HAND-based)	✅️	✅️	✅️	✅️
Inundation	Produces a binary flood/no-flood mask from a HAND raster and water level	Standard (HAND-based)	✅️	✅️	✅️	✅️
Curve Number Runoff	Estimates runoff depth from rainfall using the SCS/NRCS curve number method	SCS/NRCS	✅️	✅️	✅️	✅️
Travel Time	Estimates overland flow travel time via simplified Manning's equation	Manning 1891	✅️	✅️	✅️	✅️
Vegetation Roughness	Derives Manning's roughness coefficients from NLCD land cover or NDVI	SCS/NRCS	✅️	✅️	✅️	✅️
Vegetation Curve Number	Derives SCS curve numbers from land cover and hydrologic soil group	SCS/NRCS	✅️	✅️	✅️	✅️
Flood Depth (Vegetation)	Manning-based steady-state flow depth incorporating vegetation roughness	Manning 1891	✅️	✅️	✅️	✅️

---

Multispectral

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Atmospherically Resistant Vegetation Index (ARVI)	Vegetation index resistant to atmospheric effects using blue band correction	Kaufman & Tanre 1992	✅️	✅️	✅️	✅️
Burn Area Index (BAI)	Spectral distance to charcoal reflectance point for burn scar detection	Chuvieco et al. 2002	✅️	✅️	✅️	✅️
Enhanced Built-Up and Bareness Index (EBBI)	Highlights built-up areas and barren land from thermal and SWIR bands	As-syakur et al. 2012	✅️	✅️	✅️	✅️
Enhanced Vegetation Index (EVI)	Enhanced vegetation index reducing soil and atmospheric noise	Huete et al. 2002	✅️	✅️	✅️	✅️
Green Chlorophyll Index (GCI)	Estimates leaf chlorophyll content from green and NIR reflectance	Gitelson et al. 2003	✅️	✅️	✅️	✅️
Modified Soil Adjusted Vegetation Index (MSAVI2)	Self-adjusting soil line vegetation index, no L parameter needed	Qi et al. 1994	✅️	✅️	✅️	✅️
Normalized Burn Ratio (NBR)	Measures burn severity using NIR and SWIR band difference	USGS Landsat	✅️	✅️	✅️	✅️
Normalized Burn Ratio 2 (NBR2)	Refines burn severity mapping using two SWIR bands	USGS Landsat	✅️	✅️	✅️	✅️
Normalized Difference Built-up Index (NDBI)	Picks out built-up and urban areas from SWIR and NIR bands	Zha et al. 2003	✅️	✅️	✅️	✅️
Normalized Difference Moisture Index (NDMI)	Detects vegetation moisture stress from NIR and SWIR reflectance	USGS Landsat	✅️	✅️	✅️	✅️
Normalized Difference Snow Index (NDSI)	Separates snow and ice from clouds using green and SWIR bands	Hall et al. 1995	✅️	✅️	✅️	✅️
Normalized Difference Water Index (NDWI)	Maps open water bodies using green and NIR band difference	McFeeters 1996	✅️	✅️	✅️	✅️
Modified Normalized Difference Water Index (MNDWI)	Detects water in urban areas using green and SWIR bands	Xu 2006	✅️	✅️	✅️	✅️
Normalized Difference Vegetation Index (NDVI)	Quantifies vegetation density from red and NIR band difference	Rouse et al. 1973	✅️	✅️	✅️	✅️
Optimized Soil Adjusted Vegetation Index (OSAVI)	SAVI with fixed L=0.16, tuned for sparse vegetation	Rondeaux et al. 1996	✅️	✅️	✅️	✅️
Soil Adjusted Vegetation Index (SAVI)	Vegetation index with soil brightness correction factor	Huete 1988	✅️	✅️	✅️	✅️
Structure Insensitive Pigment Index (SIPI)	Estimates carotenoid-to-chlorophyll ratio for plant stress detection	Penuelas et al. 1995	✅️	✅️	✅️	✅️
True Color	Composites red, green, and blue bands into a natural color image	Standard	✅️	✅️	✅️	✅️

---

Classification

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Box Plot	Classifies values into bins based on box plot quartile boundaries	PySAL mapclassify	✅️	✅	✅	🔄
Equal Interval	Divides the value range into equal-width bins	PySAL mapclassify	✅️	✅	✅	✅
Head/Tail Breaks	Classifies heavy-tailed distributions using recursive mean splitting	PySAL mapclassify	✅️	✅	🔄	🔄
Maximum Breaks	Finds natural groupings by maximizing differences between sorted values	PySAL mapclassify	✅️	✅	🔄	🔄
Natural Breaks	Optimizes class boundaries to minimize within-class variance (Jenks)	Jenks 1967, PySAL	✅️	✅	🔄	🔄
Percentiles	Assigns classes based on user-defined percentile breakpoints	PySAL mapclassify	✅️	✅	✅	🔄
Quantile	Distributes values into classes with equal observation counts	PySAL mapclassify	✅️	✅	✅	🔄
Reclassify	Remaps pixel values to new classes using a user-defined lookup	PySAL mapclassify	✅️	✅	✅	✅
Std Mean	Classifies values by standard deviation intervals from the mean	PySAL mapclassify	✅️	✅	✅	✅

---

Focal

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Apply	Applies a custom function over a sliding neighborhood window	Standard	✅️	✅️	✅️	✅️
Hotspots	Identifies statistically significant spatial clusters using Getis-Ord Gi*	Getis & Ord 1992	✅️	✅️	✅️	✅️
Emerging Hotspots	Classifies time-series hot/cold spot trends using Gi* and Mann-Kendall	Getis & Ord 1992, Mann 1945	✅️	✅️	✅️	✅️
Mean	Computes the mean value within a sliding neighborhood window	Standard	✅️	✅️	✅️	✅️
Focal Statistics	Computes summary statistics over a sliding neighborhood window	Standard	✅️	✅️	✅️	✅️
Bilateral	Feature-preserving smoothing via bilateral filtering	Tomasi & Manduchi 1998	✅️	✅️	✅️	✅️
GLCM Texture	Computes Haralick GLCM texture features over a sliding window	Haralick et al. 1973	✅️	✅️	🔄	🔄
Sobel X	Horizontal gradient via Sobel operator (detects vertical edges)	Sobel & Feldman 1968	✅️	✅️	✅️	✅️
Sobel Y	Vertical gradient via Sobel operator (detects horizontal edges)	Sobel & Feldman 1968	✅️	✅️	✅️	✅️
Laplacian	Omnidirectional second-derivative edge detector	Standard	✅️	✅️	✅️	✅️
Prewitt X	Horizontal gradient via Prewitt operator (detects vertical edges)	Prewitt 1970	✅️	✅️	✅️	✅️
Prewitt Y	Vertical gradient via Prewitt operator (detects horizontal edges)	Prewitt 1970	✅️	✅️	✅️	✅️

---

Proximity

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Allocation	Assigns each cell to the identity of the nearest source feature	Standard (Dijkstra)	✅️	✅	✅️	✅️
Balanced Allocation	Partitions a cost surface into territories of roughly equal cost-weighted area	Custom	✅️	✅	✅️	✅️
Cost Distance	Computes minimum accumulated cost to the nearest source through a friction surface	Standard (Dijkstra)	✅️	✅	✅️	✅️
Least-Cost Corridor	Identifies zones of low cumulative cost between two source locations	Standard (Dijkstra)	✅️	✅	✅️	✅️
Direction	Computes the direction from each cell to the nearest source feature	Standard	✅️	✅	✅️	✅️
Proximity	Computes the distance from each cell to the nearest source feature	Standard	✅️	✅	✅️	✅️
Surface Distance	Computes distance along the 3D terrain surface to the nearest source	Standard (Dijkstra)	✅️	✅	✅️	✅️
Surface Allocation	Assigns each cell to the nearest source by terrain surface distance	Standard (Dijkstra)	✅️	✅	✅️	✅️
Surface Direction	Computes compass direction to the nearest source by terrain surface distance	Standard (Dijkstra)	✅️	✅	✅️	✅️

---

Zonal

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Apply	Applies a custom function to each zone in a classified raster	Standard	✅️	✅️	✅️	✅️
Crop	Extracts the bounding rectangle of a specific zone	Standard	✅️	✅️	✅️	✅️
Regions	Identifies connected regions of non-zero cells	Standard (CCL)	✅️	✅️	✅️	✅️
Trim	Removes nodata border rows and columns from a raster	Standard	✅️	✅️	✅️	✅️
Zonal Statistics	Computes summary statistics for a value raster within each zone	Standard	✅️	✅️	✅️	🔄
Zonal Cross Tabulate	Cross-tabulates agreement between two categorical rasters	Standard	✅️	✅️	🔄	🔄

---

Interpolation

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
IDW	Inverse Distance Weighting from scattered points to a raster grid	Standard (IDW)	✅️	✅️	✅️	✅️
Kriging	Ordinary Kriging with automatic variogram fitting (spherical, exponential, gaussian)	Standard (ordinary kriging)	✅️	✅️	✅️	✅️
Spline	Thin Plate Spline interpolation with optional smoothing	Standard (TPS)	✅️	✅️	✅️	✅️

---

Morphological

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Erode	Morphological erosion (local minimum over structuring element)	Standard (morphology)	✅️	✅️	✅️	✅️
Dilate	Morphological dilation (local maximum over structuring element)	Standard (morphology)	✅️	✅️	✅️	✅️
Opening	Erosion then dilation (removes small bright features)	Standard (morphology)	✅️	✅️	✅️	✅️
Closing	Dilation then erosion (fills small dark gaps)	Standard (morphology)	✅️	✅️	✅️	✅️
Gradient	Dilation minus erosion (edge detection)	Standard (morphology)	✅️	✅️	✅️	✅️
White Top-hat	Original minus opening (isolate bright features)	Standard (morphology)	✅️	✅️	✅️	✅️
Black Top-hat	Closing minus original (isolate dark features)	Standard (morphology)	✅️	✅️	✅️	✅️

---

Fire

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
dNBR	Differenced Normalized Burn Ratio (pre minus post NBR)	USGS	✅️	✅️	✅️	✅️
RdNBR	Relative dNBR normalized by pre-fire vegetation density	USGS	✅️	✅️	✅️	✅️
Burn Severity Class	USGS 7-class burn severity from dNBR thresholds	USGS	✅️	✅️	✅️	✅️
Fireline Intensity	Byram's fireline intensity from fuel load and spread rate (kW/m)	Byram 1959	✅️	✅️	✅️	✅️
Flame Length	Flame length derived from fireline intensity (m)	Byram 1959	✅️	✅️	✅️	✅️
Rate of Spread	Simplified Rothermel spread rate with Anderson 13 fuel models (m/min)	Rothermel 1972, Anderson 1982	✅️	✅️	✅️	✅️
KBDI	Keetch-Byram Drought Index single time-step update (0-800 mm)	Keetch & Byram 1968	✅️	✅️	✅️	✅️

---

Raster / Vector Conversion

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Polygonize	Converts contiguous regions of equal value into vector polygons	Standard (CCL)	✅️	✅️	✅️	🔄
Contours	Extracts elevation contour lines (isolines) from a raster surface	Standard (marching squares)	✅️	✅️	🔄	🔄
Rasterize	Rasterizes vector geometries (polygons, lines, points) from a GeoDataFrame	Standard (scanline, Bresenham)	✅️		✅️

---

Multivariate

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Mahalanobis Distance	Measures statistical distance from a multi-band reference distribution, accounting for band correlations	Mahalanobis 1936	✅️	✅️	✅️	✅️

---

Pathfinding

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
A* Pathfinding	Finds the least-cost path between two cells on a cost surface	Hart et al. 1968	✅️	✅	🔄	🔄
Multi-Stop Search	Routes through N waypoints in sequence, with optional TSP reordering	Custom	✅️	✅	🔄	🔄

---

Diffusion

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Diffuse	Runs explicit forward-Euler diffusion on a 2D scalar field	Standard (heat equation)	✅️	✅️	✅️	✅️

---

Dasymetric

Name	Description	Source	NumPy xr.DataArray	Dask xr.DataArray	CuPy GPU xr.DataArray	Dask GPU xr.DataArray
Disaggregate	Redistributes zonal totals to pixels using an ancillary weight surface	Mennis 2003	✅️	✅️	✅️	✅️
Pycnophylactic	Tobler's pycnophylactic interpolation preserving zone totals via Laplacian smoothing	Tobler 1979	✅️

---

Usage

Quick Start

Importing xrspatial registers an .xrs accessor on DataArrays and Datasets, giving you tab-completable access to every spatial operation:

import xrspatial
from xrspatial.geotiff import read_geotiff

# Read a GeoTIFF (no GDAL required)
elevation = read_geotiff('dem.tif')

# Surface analysis — call operations directly on the DataArray
slope = elevation.xrs.slope()
hillshaded = elevation.xrs.hillshade(azimuth=315, angle_altitude=45)
aspect = elevation.xrs.aspect()

# Classification
classes = elevation.xrs.equal_interval(k=5)
breaks = elevation.xrs.natural_breaks(k=10)

# Proximity
distance = elevation.xrs.proximity(target_values=[1])

# Multispectral — call on the NIR band, pass other bands as arguments
nir = xr.DataArray(np.random.rand(100, 100), dims=['y', 'x'])
red = xr.DataArray(np.random.rand(100, 100), dims=['y', 'x'])
blue = xr.DataArray(np.random.rand(100, 100), dims=['y', 'x'])

vegetation = nir.xrs.ndvi(red)
enhanced_vi = nir.xrs.evi(red, blue)

Dataset Support

The .xrs accessor works on Datasets too. Single-input functions apply the operation to each data variable. Multi-input functions (multispectral indices) accept string kwargs that map band aliases to variable names:

ds = xr.Dataset({'band_4': red, 'band_5': nir})

# Single-input: slope computed for each variable
slope_ds = ds.xrs.slope()

# Multi-input: map variable names to band parameters
ndvi_result = ds.xrs.ndvi(nir='band_5', red='band_4')

Function Import Style

All operations are also available as standalone functions if you prefer explicit imports:

from xrspatial import hillshade, slope, ndvi

hillshaded = hillshade(elevation)
slope_result = slope(elevation)
vegetation = ndvi(nir, red)

Check out the user guide here.

---

Dependencies

Core: numpy, numba, scipy, xarray, matplotlib, zstandard

Optional:

• pyproj — WKT/PROJ CRS resolution
• cupy — GPU acceleration
• dask — out-of-core processing
• libnvcomp — GPU batch decompression (deflate, ZSTD)
• kvikio — GPUDirect Storage (SSD → GPU)
• fsspec + s3fs/gcsfs/adlfs — cloud storage

Notes on GDAL

xarray-spatial does not depend on GDAL. The built-in GeoTIFF/COG reader and writer (xrspatial.geotiff) handles raster I/O natively using only numpy, numba, and the standard library. This means:

• Zero GDAL installation hassle. pip install xarray-spatial gets you everything needed to read and write GeoTIFFs, COGs, and VRT files.
• Pure Python, fully extensible. All codec, header parsing, and metadata code is readable Python/Numba, not wrapped C/C++.
• GPU-accelerated reads. With optional nvCOMP, compressed tiles decompress directly on the GPU via CUDA -- something GDAL cannot do.

The native reader is pixel-exact against rasterio/GDAL across Landsat 8, Copernicus DEM, USGS 1-arc-second, and USGS 1-meter DEMs. For uncompressed files it reads 5-7x faster than rioxarray; for compressed COGs it is comparable or faster with GPU acceleration.

Citation

Cite this code:

xarray-contrib/xarray-spatial, https://github.com/xarray-contrib/xarray-spatial, ©2020-2026.