Precise astronomical dates and fixed meteorological calendars do not capture the near-surface temporal variability of seasons. This study presents a purely data-driven framework to objectively quantify trends in seasons locally, validated using 45 years (1980–2024) of hourly ERA5 data for a subtropical location in India. By integrating spectral analysis for primary interval detection with unsupervised k-means clustering, the framework reveals dynamic seasonal soft boundaries.
The Sample 0001 AINPP-PB-LATAM-DS dataset provides a compact, self-contained example of the AINPP Precipitation Benchmark for Latin America, designed to support reproducible research in precipitation analysis, nowcasting, and multi-source data integration. The dataset covers the period from 1 to 31 January 2024 at hourly resolution and spans the Latin American domain from 55°S to 33°N and 120°W to 23°W on a regular 0.10° × 0.10° grid.
This study proposes an innovative credibility theory-driven framework for wind turbine power prediction during high-impact weather events. Addressing limitations of conventional approaches in handling complex meteorological disturbances, our methodology integrates multiscale physical constraints with data-driven modeling through three core components: First, a non-linear heavy-tailed error characterization model combining Gaussian Mixture Model with dynamic Copula functions effectively captures asymmetric prediction errors during typhoon-induced gust conditions.
The dataset for a smart agriculture system consists of a weather station that collects (air temperature, humidity, and UV index), and soil (moisture and temperature).
This dataset provides processed Sound Speed Profile (SSP) time-series derived from the GDCSM ARGO global ocean dataset for the period 2004–2024. The original monthly NetCDF files provided by original authors contain temperature, salinity, and pressure measurements on a 0.5° × 0.5° spatial grid and 58 non-uniform depth levels. From these files, sound speed values were already provided in the source dataset and were extracted and organized into a structured machine-learning–ready format.
The dataset used in this study consists of multiple text files containing gridded atmospheric temperature measurements derived from radio occultation observations. Each file name follows the format Lat_a_b_Lon_c_d.txt, where a and b denote the starting and ending latitudes, and c and d denote the starting and ending longitudes defining the spatial coverage of the data segment.
Focusing on global atmospheric and oceanic research frontiers, applying artificial intelligence (AI) to weather and climate prediction has become a hot research topic, especially for improving forecasts of extreme weather events. The El Niño–Southern Oscillation (ENSO), the strongest and most significant interannual climate signal occurring in the tropical Pacific, often triggers floods, droughts, heatwaves, and snow disasters. For example, the extreme cold winter in China at the end of 2020 was closely linked to ENSO.
This Excel file compiles historical public weather data from Morocco, the data spans from 1989 to 2014, depending on the stations. Each station is accompanied by its geographical coordinates (longitude, latitude). collected through the national meteorological network.
This dataset presents a processed, machine-learning-ready compilation of hourly time-series meteorological parameters for the purpose of short-term rooftop solar energy prediction. Sourced from public archives of the NASA POWER Project and NOAA, the raw data was rigorously cleaned, merged, and structured. Key parameters include
a. Solar Irradiance (Wh/m²) –Target variable
b. Temperature (°C)
c. Surface Pressure (kPa)
d. Specific Humidity (g/kg)
e. Relative Humidity (%)
f. Wind Speed (m/s)
g. Wind Direction (degrees)
