Working with Zarr¶

Intro¶

titiler.xarray is a submodule designed specifically for working with multidimensional dataset. With version 0.25.0, we've introduced a default application with only support for Zarr dataset.

In [ ]:

# setup
import httpx
import json
from IPython.display import Image

# Developmentseed Demo endpoint. Please be kind. Ref: https://github.com/developmentseed/titiler/discussions/1223
# titiler_endpoint = "https://xarray.titiler.xyz"

# Or launch your own local instance with:
# uv run --group server uvicorn titiler.xarray.main:app --host 127.0.0.1 --port 8080 --reload
titiler_endpoint = "http://127.0.0.1:8080"

zarr_url = "https://nasa-power.s3.us-west-2.amazonaws.com/syn1deg/temporal/power_syn1deg_monthly_temporal_lst.zarr"

# setup import httpx import json from IPython.display import Image # Developmentseed Demo endpoint. Please be kind. Ref: https://github.com/developmentseed/titiler/discussions/1223 # titiler_endpoint = "https://xarray.titiler.xyz" # Or launch your own local instance with: # uv run --group server uvicorn titiler.xarray.main:app --host 127.0.0.1 --port 8080 --reload titiler_endpoint = "http://127.0.0.1:8080" zarr_url = "https://nasa-power.s3.us-west-2.amazonaws.com/syn1deg/temporal/power_syn1deg_monthly_temporal_lst.zarr"

Dataset Metadata¶

The /dataset/dict endpoint returns general metadata about the Zarr Dataset

Endpoint: /dataset/dict

QueryParams:

• url: Zarr store URL

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/dataset/dict",
    params={
        "url": zarr_url,
    },
).json()

print(json.dumps(r, indent=2))

r = httpx.get( f"{titiler_endpoint}/dataset/dict", params={ "url": zarr_url, }, ).json() print(json.dumps(r, indent=2))

List of available variables¶

Endpoint: /dataset/keys

QueryParams:

• url: Zarr store URL

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/dataset/keys",
    params={
        "url": zarr_url,
    },
).json()

print(json.dumps(r, indent=2))

r = httpx.get( f"{titiler_endpoint}/dataset/keys", params={ "url": zarr_url, }, ).json() print(json.dumps(r, indent=2))

Variable Info¶

We can use /info endpoint to get more Geo information about a specific variable.

QueryParams:

• url: Zarr store URL
• variable: Variable's name (e.g AIRMASS, found in /dataset/keys response)

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/info",
    params={"url": zarr_url, "variable": "AIRMASS"},
).json()

print(json.dumps(r, indent=2))

r = httpx.get( f"{titiler_endpoint}/info", params={"url": zarr_url, "variable": "AIRMASS"}, ).json() print(json.dumps(r, indent=2))

Or as a GeoJSON feature

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/info.geojson",
    params={"url": zarr_url, "variable": "AIRMASS"},
).json()

print(json.dumps(r, indent=2))

r = httpx.get( f"{titiler_endpoint}/info.geojson", params={"url": zarr_url, "variable": "AIRMASS"}, ).json() print(json.dumps(r, indent=2))

Knowledge¶

Looking at the info response we can see that the AIRMASS variable has 348 (count) bands, each one corresponding to as specific TIME (day).

We can also see that the data is stored as float32 which mean that we will have to apply linear rescaling in order to get output image as PNG/JPEG.

The min/max values are also indicated with valid_max=31.73 and valid_min=1.0.

Dimension Reduction¶

We cannot visualize all the bands at once, so we need to perform dimension reduction to go from array in shape (348, 360, 180) to a 1b (1, 360, 180) or 3b (3, 360, 180) image.

To do it, we have two methods whitin titiler.xarray:

• using bidx=: same as for COG we can select a band index
• using sel={dimension}=value: which will be using xarray .sel method

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/bbox/-180,-90,180,90.png",
    params=(
        ("url", zarr_url),
        ("variable", "AIRMASS"),
        # Select 1 specific band
        ("bidx", 50),
        ("rescale", "1,20"),
        ("colormap_name", "viridis"),
    ),
    timeout=10,
)

Image(r.content)

r = httpx.get( f"{titiler_endpoint}/bbox/-180,-90,180,90.png", params=( ("url", zarr_url), ("variable", "AIRMASS"), # Select 1 specific band ("bidx", 50), ("rescale", "1,20"), ("colormap_name", "viridis"), ), timeout=10, ) Image(r.content)

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/bbox/-180,-90,180,90.png",
    params=(
        ("url", zarr_url),
        ("variable", "AIRMASS"),
        # Select 1 specific time slices
        ("sel", "time=2003-06-30"),
        ("rescale", "1,20"),
        ("colormap_name", "viridis"),
    ),
    timeout=10,
)

Image(r.content)

r = httpx.get( f"{titiler_endpoint}/bbox/-180,-90,180,90.png", params=( ("url", zarr_url), ("variable", "AIRMASS"), # Select 1 specific time slices ("sel", "time=2003-06-30"), ("rescale", "1,20"), ("colormap_name", "viridis"), ), timeout=10, ) Image(r.content)

In [ ]:

r = httpx.get(
    f"{titiler_endpoint}/bbox/-180,-90,180,90.png",
    params=(
        ("url", zarr_url),
        ("variable", "AIRMASS"),
        # Select 3 specific time slices to create a 3 band image
        ("sel", "time=2003-06-30"),
        ("sel", "time=2004-06-30"),
        ("sel", "time=2005-06-30"),
        ("rescale", "1,10"),
    ),
    timeout=10,
)

Image(r.content)

r = httpx.get( f"{titiler_endpoint}/bbox/-180,-90,180,90.png", params=( ("url", zarr_url), ("variable", "AIRMASS"), # Select 3 specific time slices to create a 3 band image ("sel", "time=2003-06-30"), ("sel", "time=2004-06-30"), ("sel", "time=2005-06-30"), ("rescale", "1,10"), ), timeout=10, ) Image(r.content)