How to read NEXRAD Level III products in Python in a Jupyter notebook
Read a Level III product file with Py-ART: pyart.io.read_nexrad_level3(path) returns a Radar object with one field, such as reflectivity or rain rate. In a Scigantic notebook the unidata-nexrad-level3 bucket is mounted at /mnt/archive, and each object key is a file name directly under it.
The code on this page runs in a Scigantic hosted notebook, where the dataset is mounted read-only at /mnt/archive. The same path is in the SCIGANTIC_MOUNT_PATH environment variable and linked from ~/archive. The public source is s3://unidata-nexrad-level3 (AWS us-east-1, public, no account needed).
Open NEXRAD Level III on SciganticOpen the sample notebook: Is it raining right now?
Code
Find the newest scan for a radar and plot it
import os, pyart, matplotlib.pyplot as plt
from datetime import datetime, timedelta, timezone
import boto3
from botocore import UNSIGNED
from botocore.config import Config
s3 = boto3.client("s3", region_name="us-east-1", config=Config(signature_version=UNSIGNED))
site, product = "FTG", "N0B" # Denver, super-res base reflectivity (always decodes)
now = datetime.now(timezone.utc)
for h in range(0, 48):
t = now - timedelta(hours=h)
r = s3.list_objects_v2(Bucket="unidata-nexrad-level3", Prefix=f"{site}_{product}_{t:%Y_%m_%d_%H}", MaxKeys=100)
if r.get("Contents"):
newest = r["Contents"][-1]; break
radar = pyart.io.read_nexrad_level3(f"{os.environ.get('SCIGANTIC_MOUNT_PATH', '/mnt/archive')}/{newest['Key']}")
field = next(iter(radar.fields)); print(newest["Key"], field, radar.fields[field]["units"], radar.nrays, "x", radar.ngates)
fig, ax = plt.subplots(figsize=(8, 8)); pyart.graph.RadarDisplay(radar).plot(field, 0, ax=ax, vmin=-10, vmax=60); ax.set_aspect("equal")
The starter cell from this dataset's schema card. It ran without error on 2026-11-15 in the Scigantic notebook image against the mounted data.
Read one product file
import os
import pyart
radar = pyart.io.read_nexrad_level3(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], 'ABC_DAA_2020_03_31_08_54_05'))
print(radar.fields.keys(), radar.nrays, radar.ngates)
Run on 2026-11-15 in the Scigantic notebook image against the mounted data. It returned: Py-ART Radar (Level III): fields ['radar_estimated_rain_rate'], 360 rays x 920 gates
What the files are
Level III files are the derived products the National Weather Service makes from each radar scan: base reflectivity, base velocity, correlation coefficient, one-hour and storm-total rain accumulation, echo tops and more.
Each file is one product from one sweep of one radar, a small binary file between about 200 bytes and 150 KB. The archive is live, with new scans landing within minutes, and continuous since 2020-03-30.
How the files are laid out
| Path under the mount | What it holds |
|---|---|
| SITE_PRODUCT_YYYY_MM_DD_HH_MM_SS | Every file sits in one flat namespace with no folders. SITE is the radar ID minus its first letter (KMTX becomes MTX). |
| FTG_N0B_2026_09_09_00_47_41 | Example: Denver (KFTG) super-resolution base reflectivity, 141,504 bytes, 720 rays by 1,840 gates, field reflectivity in dBZ. |
| PRODUCT codes | N0B base reflectivity, N0G base velocity, N0C correlation coefficient, DAA one-hour accumulation, DTA storm-total accumulation, EET echo tops, DVL vertically integrated liquid. |
Size, format and license
| Format | NEXRAD Level III (NIDS) binary products, read with Py-ART |
|---|---|
| Size | 206 radar sites, over 100 product codes per site, a new file per product every 2 to 10 minutes since 2020-03-30. The object count is in the billions. |
| License | Custom; see the provider documentation linked from the archive page |
| Provider | Unidata, on AWS Open Data |
| Public source | s3://unidata-nexrad-level3 (AWS us-east-1, public, no account needed) |
| Notebook image | Geospatial (Py-ART, MetPy and cartopy preinstalled) |
Things to know before you read it
- Do not list the mount root. Choose a key by listing the bucket with a prefix that includes the site, product and date, as the first code block does, then open that key under the mount.
- S3 lists keys in ascending order, so a prefix of only site and product returns the oldest scans (March 2020), not the newest. Put the date and hour in the prefix.
- A
DAAfile of about 222 to 274 bytes is a "no precipitation detected" stub.pyart.io.read_nexrad_level3raises an AssertionError on it.
Run it
Start a notebook from the NEXRAD Level III archive page. The data is mounted when the notebook opens, so the code above runs as written and nothing is downloaded first. For a longer worked example, open the sample notebook Is it raining right now?.
Related guides
- How to open NEXRAD Level II radar data in a Jupyter notebook without downloading it
- How to open ECMWF IFS S2S reforecasts stored as Icechunk in a Jupyter notebook
- How to open OME-Zarr volumes in a Jupyter notebook
- How to query 1000 Genomes VCF files on S3 from a Jupyter notebook
Last checked 2026-11-21. All dataset guides.