High-quality observational data are crucial to establish reliable datasets for studying historical changes, the current state, and the future evolution of Earth's climate. This session provides a forum to discuss best practices for quality control, harmonization, and homogenization of observational data for climate services and applications. Key topics include data assimilation into reanalyses (e.g., ECMWF's forthcoming ERA-6), as well as integration into physical retrieval algorithms and AI/ML techniques (e.g., within ESA's Climate Change Initiative or EUMETSAT Satellite Application Facilities). We also invite contributions analyzing the impact of these data on long-term Essential Climate Variable (ECV) datasets.

Integrating observational data into climate services demands strict quality and consistency standards, including calibration, spatial/temporal coverage, and homogeneity. Reanalyses perform best with well-calibrated data that preserve each sensor's unique characteristics (harmonized data), whereas long-term ECV datasets across atmospheric, terrestrial, and oceanic domains rely on data adjusted to a common reference (homogenized data).

We invite contributions on climate datasets from in-situ observation networks, satellite remote sensing, and reanalyses. The aim is to bring together data developers and users to demonstrate the downstream impact of these observational data on climate applications. Specifically, the session seeks presentations on:
- Data rescue and recovery of historical climate datasets.
- Harmonization, homogenization, and validation of observational climate datasets.
- Novel prospects (AI/ML) to exploit observations and their emerging requirements.
- Impact evaluation of observational datasets on climate services, trend analysis, and attribution studies of weather and climate extremes.