Tropical Livestock Units from GLW4

Compute Tropical Livestock Units (TLU) by combining GLW4 species density layers with standard conversion factors, streaming directly from the Hub's Zarr store.

Livestock density is published one species at a time — cattle, buffalo, sheep, goats — but many analyses want a single grazing-pressure number. The Tropical Livestock Unit (TLU) provides that: each animal counts as a fraction of a 250 kg reference animal, so species can be weighted and summed into one comparable layer.

This notebook builds a TLU/km² map from Gridded Livestock of the World v4, streaming just the data it needs — nothing is downloaded ahead of time.

python

import xarray as xr

Open the data

GLW4 lives on object storage as a Zarr cube, so xarray opens it over HTTPS and only ever reads the chunks a computation touches. Inspecting the dataset shows one density variable per species, in head/km².

python

ds = xr.open_zarr(
    "https://digital-atlas.s3.amazonaws.com/cdh/data/glw4-2020/glw4-2020.zarr"
)

python

ds

<xarray.Dataset> Size: 224MB
Dimensions:      (y: 2160, x: 4320)
Coordinates:
  * y            (y) float64 17kB 89.96 89.88 89.79 ... -89.79 -89.87 -89.96
  * x            (x) float64 35kB -180.0 -179.9 -179.8 ... 179.8 179.9 180.0
Data variables:
    buffalo      (y, x) float32 37MB dask.array<chunksize=(1080, 1080), meta=np.ndarray>
    cattle       (y, x) float32 37MB dask.array<chunksize=(1080, 1080), meta=np.ndarray>
    chicken      (y, x) float32 37MB dask.array<chunksize=(1080, 1080), meta=np.ndarray>
    goat         (y, x) float32 37MB dask.array<chunksize=(1080, 1080), meta=np.ndarray>
    pig          (y, x) float32 37MB dask.array<chunksize=(1080, 1080), meta=np.ndarray>
    sheep        (y, x) float32 37MB dask.array<chunksize=(1080, 1080), meta=np.ndarray>
    spatial_ref  int64 8B ...
Attributes:
    title:    GLW4 2020 livestock density
    source:   Gridded Livestock of the World v4 (GLW4), 2020, dasymetric

Weight species into TLU

The FAO weights express each species as a fraction of a 250 kg tropical reference animal — a head of cattle counts as 0.7 TLU, a sheep or goat as 0.1. Multiplying each species layer by its weight and summing collapses the cube into a single TLU/km² surface.

python

# Tropical Livestock Unit weights
# head/km² × weight = TLU/km².
TLU_RUMINANT = {
    "cattle": 0.7,
    "buffalo": 0.7,
    "sheep": 0.1,
    "goat": 0.1,
}

python

layers = [
    w * ds[sp].fillna(0) for sp, w in TLU_RUMINANT.items()
]
da = (
    xr.concat(layers, dim="species")
    .sum("species")
    .rename("TLU")
)
da.attrs["units"] = "TLU/km2"

python

da

<xarray.DataArray 'TLU' (y: 2160, x: 4320)> Size: 37MB
dask.array<sum-aggregate, shape=(2160, 4320), dtype=float32, chunksize=(1080, 1080), chunktype=numpy.ndarray>
Coordinates:
  * y        (y) float64 17kB 89.96 89.88 89.79 89.71 ... -89.79 -89.87 -89.96
  * x        (x) float64 35kB -180.0 -179.9 -179.8 -179.7 ... 179.8 179.9 180.0
Attributes:
    long_name:               Cattle density
    units:                   TLU/km2
    source_url:              https://storage.googleapis.com/fao-gismgr-glw4-2...
    spatial:dimensions:      ['y', 'x']
    spatial:bbox:            [-180.0, -89.99999999999994, 179.99999999999983,...
    spatial:transform_type:  affine
    spatial:transform:       [0.0833333333333333, 0.0, -180.0, 0.0, -0.083333...
    spatial:shape:           [2160, 4320]
    spatial:registration:    pixel
    proj:code:               EPSG:4326
    zarr_conventions:        [{'uuid': '689b58e2-cf7b-45e0-9fff-9cfc0883d6b4'...

xarray.DataArray

'TLU'

y: 2160
x: 4320

dask.array<chunksize=(1080, 1080), meta=np.ndarray>

	Array	Chunk
Bytes	35.60 MiB	4.45 MiB
Shape	(2160, 4320)	(1080, 1080)
Dask graph	8 chunks in 34 graph layers
Data type	float32 numpy.ndarray

Coordinates: (2)

(y)

float64

89.96 89.88 89.79 ... -89.87 -89.96

array([ 89.958333,  89.875   ,  89.791667, ..., -89.791667, -89.875   ,
       -89.958333], shape=(2160,))

(x)

float64

-180.0 -179.9 ... 179.9 180.0

array([-179.958333, -179.875   , -179.791667, ...,  179.791667,  179.875   ,
        179.958333], shape=(4320,))

Attributes: (11)
long_name :
Cattle density
units :
TLU/km2
source_url :
https://storage.googleapis.com/fao-gismgr-glw4-2020-data/DATA/GLW4-2020/MAPSET/D-DA/GLW4-2020.D-DA.CTL.tif
spatial:dimensions :
['y', 'x']
spatial:bbox :
[-180.0, -89.99999999999994, 179.99999999999983, 90.0]
spatial:transform_type :
affine
spatial:transform :
[0.0833333333333333, 0.0, -180.0, 0.0, -0.0833333333333333, 90.0]
spatial:shape :
[2160, 4320]
spatial:registration :
pixel
proj:code :
EPSG:4326
zarr_conventions :
[{'uuid': '689b58e2-cf7b-45e0-9fff-9cfc0883d6b4', 'schema_url': 'https://raw.githubusercontent.com/zarr-conventions/spatial/refs/tags/v1/schema.json', 'spec_url': 'https://github.com/zarr-conventions/spatial/blob/v1/README.md', 'name': 'spatial:', 'description': 'Spatial coordinate information'}, {'uuid': 'f17cb550-5864-4468-aeb7-f3180cfb622f', 'schema_url': 'https://raw.githubusercontent.com/zarr-experimental/geo-proj/refs/tags/v1/schema.json', 'spec_url': 'https://github.com/zarr-experimental/geo-proj/blob/v1/README.md', 'name': 'proj:', 'description': 'Coordinate reference system information for geospatial data'}]

Subset and plot a region

Everything so far is lazy — no pixels have moved yet. Slicing before plotting means only the chunks inside the window are downloaded: here a roughly 8°×10° box over Kenya and northern Tanzania. .plot() gives a quick look; the same array is ready for zonal statistics or export.

python

regional_tlu = da.sel(
    y=slice(5.0, -5.2), x=slice(34.0, 42.0)
)

python

regional_tlu.plot()

Cell output

Where to go next

Change the slice to your own region — the lazy reads mean any window costs only its own chunks. Per-species metadata, licensing, and the citation live on the GLW4 record page, and the weights dictionary is the place to add camels or other species if your context needs them.