Tracks of mid-tropospheric systems from 1998 to 2018 (June–September) were identified using the 600 hPa geopotential height minima.

Dataset Description This dataset contains the raw output of an automatic cyclone detection and tracking algorithm (see: Classification of mid-tropospheric cyclones over the Arabian Sea and western India, QJRMS, 2021). The algorithm was applied over the region 0–25°N, 60°–130°E at 600 hPa, primarily targeting mid-level systems. Although the method is based on detecting 600 hPa geopotential minima for mid-level systems, it also effectively captures a large fraction of tropical cyclones, monsoon lows, and monsoon depressions, since these warm-core systems exhibit significant intensity in both the middle and lower troposphere. Thus, this dataset is suitable for analyzing statistics of LPSs, cyclones, and depressions in addition to mid-tropospheric systems. The tracks are highly accurate and can represent very complex cyclone motions. A few sample tracks of systems MTC, LPS and cyclones, as identified by the cyclone tracking algorithm, can be found here : Examples Only tracks persisting for at least 72 hours (≥ 12 points at 6-hourly resolution) are retained. Dataset Structure The dataset is stored as following variables: TIME_RECORD – timestamps of track points LAT_RECORD, LON_RECORD – cyclone center positions DIFF_RECORD – mid-level (600 hPa) minus lower-level (950 hPa) vorticity (MTC vs LPS phase indicator) VOM_RECORD – mid-level vorticity Q_RECORD – mid-level humidity PMAX_RECORD – pressure of maximum vorticity TRACK_LENGTH – total duration of the system INDX_RECORD – system index reference Important Notes This file contains all detected tracks, including many very short-lived or weak “dry” systems. Users should apply appropriate filtering before using the data for scientific analysis. Example filtering strategies: Long-lived systems → TRACK_LENGTH ≥ 72 (≥ 3 days at 6-hourly resolution). Moist cyclones → filter on Q_RECORD (g/kg) above a chosen humidity threshold. Mid-Tropospheric Cyclone (MTC) phase → DIFF_RECORD > 0-1. Low-Tropospheric (or LPS/Cyclone) phase → DIFF_RECORD ≤ 0-2 (with some relaxation, as many LPSs have weak low-level vorticity). Some systems may transition between MTCs and LPSs during their lifecycle (see: Kushwaha, P., J. Sukhatme, and R. Nanjundiah, 2021: A Global Tropical Survey of Midtropospheric Cyclones. MWR, 149, 2737–2753, https://doi.org/10.1175/MWR-D-20-0222.1). Each system track is stored in an array of fixed length 300, though most tracks end earlier. Missing values appear beyond the maximum track length, as specified by the track_length variable. Topographic cyclones This subset on mid level ssytems only considers genesis south of 25°N and is therefore unaffected by topographic systems that may appear on the 600 hPa surface due to the Himalayas or the Tibetan Plateau in the ERA5 data. However, when including systems north of 25°N, topographic systems should be removed during the post-processing stage, as ERA5 does not mask terrain by default. Consequently, the tracking algorithm also identifies such systems if the topography extends above 600 geopotential meters. Systems intersecting high topography should be excluded using surface geopotential filtering. However, if the surface geopotential height within the analysis domain lies below the 600 hPa geopotential height, this step may not be necessary. 📊 Validation The accuracy of the tracking algorithm has been demonstrated by showing a random track example, which successfully reproduces CYCLONE, LPS and MTC tracks with high accuracy (see: GitHub example). Please note that the provided Python scripts are intended to help users understand the data and its structure. These scripts have not been thoroughly reviewed or debugged, so users are advised to check them carefully and, once familiar with the data structure, consider writing their own. Ensure that you update the data path, specify the correct file name, and set the output directory for plots to avoid errors. 📧 Contact For queries, please contact: [e-mail]