WDs are identified using the feature-tracking algorithm described in Hunt et al (2018). Relative vorticity is averaged across the 450-300 hPa layer, and then spectrally truncated to T42 to remove high-frequency noise that hinders tracking. For each region of positive vorticity, the centroid is located and labelled as a candidate WD. These centroids are connected between timesteps using a nearest-neighbour algorithm, biased to take into account the steering winds of the subtropical jet. Systems that do not on average travel eastward, last fewer than 48 hours, or do not pass through the box [20-42.5°N, 60-80°E] are rejected. Applied to ERA5, this gives over seventy years of track data (1950-2022). The method followed here is identical to Nischal et al (2022), except the northern edge of the catching box is extended from 36.5°N to 42.5°N, to ensure that all WDs that potentially impact North India are included. Column titles are: timestep: a counter indicating the number of 3-hourly timesteps that have passed since 1950-01-01 00:00 track_id: a unique identifier linking points into tracks time: string describing the date and time lon: longitude lat: latitude vort: vorticity measured at the centre of the WD averaged over the 450-300 hPa layer. Can be used for intensity filtering. eccentricity: eccentricity of the region of positive vorticity. Can be used to understand local dynamics. Hunt, K. M. R., Turner, A. G., & Shaffrey, L. C. (2018). The evolution, seasonality and impacts of western disturbances. Quarterly Journal of the Royal Meteorological Society, 144(710), 278-290. Nischal, Attada, R., & Hunt, K. M. (2022). Evaluating winter precipitation over the western Himalayas in a high-resolution Indian regional reanalysis using multisource climate datasets. Journal of Applied Meteorology and Climatology, 61(11), 1613-1633.