Data-Driven Decision Support in Environmental Management: Hybrid GNN-PINN Modeling of Subsurface Soil Temperature

This dataset comprises a comprehensive collection of daily meteorological records obtained from the Turkish State Meteorological Service (MGM) via the MEVBIS platform. The data spans a five-year period from January 1, 2020, to December 31, 2024, and includes recordings from 15 distinct regional stations representing diverse climatic and geographic zones across Turkey (including arid central regions, coastal Mediterranean zones, continental highlands, and humid Black Sea areas).The primary focus of this dataset is to support research in environmental decision support systems (DSS), precision agriculture, and the development of advanced machine learning models, specifically Physics-Informed Neural Networks (PINNs) and Graph Neural Networks (GNNs).

The dataset contains the following key components:
Subsurface Data: Daily mean, minimum, and maximum soil temperatures recorded at five distinct depths: 5 cm, 10 cm, 20 cm, 50 cm, and 100 cm.
Surface Data: Daily mean 2-meter air temperature, serving as the primary boundary condition/input feature for modeling.
Geospatial Metadata: Station-specific coordinates including latitude, longitude, and altitude to facilitate spatial dependency modeling in graph-based architectures.

Data Processing:The raw data has undergone rigorous preprocessing to ensure suitability for computational modeling:
Temporal Filtering: Corrupted entries and sensor errors were removed. Sequences with consecutive missing values exceeding three days were excluded.
Normalization: Temperature values are min-max scaled to the range [0, 1] to enhance numerical stability during model training.
Structuring: The data is structured to support depth encoding z as a continuous input variable.
Potential Applications:This dataset is ideal for researchers working on:
Soil temperature forecasting and thermal diffusion modeling.Training and benchmarking hybrid PINN-GNN architectures. Agricultural optimization scenarios, such as irrigation planning and crop risk modeling.Analysis of climate generalization under extreme seasonal conditions (Summer vs. Winter).
Files Description (Variable Dictionary):
Station_ID: Unique identifier for the meteorological station.
Date: Observation date (Daily resolution).
t_day: Day index (cumulative count since Jan 1, 2020).
T_air: Daily mean 2-meter air temperature (°C).
T_soil_5cm / 10cm / 20cm / 50cm / 100cm: Daily mean soil temperatures at respective depths (°C).
Latitude / Longitude: Geographic coordinates of the station.Altitude: Elevation of the station (meters).