Experimental Dataset of Starlink Performance Across Different Weather Conditions in Southeast Texas Coastal Areas

This dataset comprises experimental measurements collected using a Starlink Flat High-Performance Gen-3 terminal across three representative deployment environments in Southeast Texas: open flatland, semi-urban, and coastal near-shore locations in the vicinity of Sabine Lake. Data were collected over multiple field sessions conducted throughout the summer season, yielding over 900 one-minute samples after data cleaning and quality control.Each record includes downlink and uplink throughput (Mbps), round-trip latency (ms), jitter (ms), packet loss ratio (%), ping success rate (%), and terminal power consumption (W). Network performance metrics were primarily obtained through the Starlink application telemetry interface and, where applicable, were cross-validated using independent measurement tools such as Cloudflare Speed Test and Wireshark-based traffic analysis.Environmental context was captured at the same temporal resolution and includes ambient temperature (°C), relative humidity (%), wind speed (mph), rainfall intensity (mm/hr), and cloud cover (%). These variables were sourced from location-specific online weather services accessed via both laptop and mobile platforms. All environmental measurements were time-aligned with the corresponding Starlink telemetry to support joint analysis of atmospheric effects on low Earth orbit (LEO) satellite link performance.Raw measurements were standardized to a uniform one-minute sampling interval. Short data gaps caused by intermittent reporting delays were addressed using forward and backward interpolation, while longer outages or corrupted records were excluded to preserve temporal continuity. After preprocessing, the dataset consists of continuous segments suitable for supervised learning and time-series forecasting.The resulting dataset provides a curated, analysis-ready empirical resource for studying Starlink maritime and near-shore connectivity behavior under realistic operating conditions. It is well-suited for machine learning-based throughput prediction, environment-aware performance modeling, link reliability analysis, and broader research in satellite communications and maritime networking.