Reading and preprocessing radar data

First, we need to import ADAM in order to use its functionality:

import adam

ADAM contains a very easy to use function that will automatically download the NEXRAD data for a specified time period and perform preprocessing on the radar data in order to make it suitable for inference by ADAM’s fine-tuned ResNet50 models. In order to perform this preprocessing for a given time period at the Romeoville radar (KLOT) in the Chicago metro area, simply do the following code:

rad_scan1 = adam.io.preprocess_radar_image('KLOT', '2025-07-15T18:00:00')

Lakebreeze inference

The next step is to develop the binary lakebreeze mask from the preprocessed radar data. This is also done with one line of code in ADAM:

rad_scan1 = adam.model.infer_lake_breeze(
    rad_scan1, model_name='lakebreeze_model_fcn_resnet50_no_augmentation')

Visualizing your result

To visualize your lakebreeze image on a base reflectivity plot, simply

adam.vis.visualize_lake_breeze(rad_scan1, vmin=0, vmax=30, cmap='ChaseSpectral')

(Source code, png, hires.png, pdf)