Tropentag, September 10 - 12, 2025, Bonn "Reconciling land system changes with planetary health"

Tempqc: An interactive ai-powered application for quality control of climate temperature data

¹West African Science Service Center on Climate Change and Adapted Land Use (WASCAL), Data Management, Burkina Faso
²West African Science Service Center on Climate Change and Adapted Land Use (WASCAL), Burkina Faso
³West African Science Service Center on Climate Change and Adapted Land Use (WASCAL), Data Mamagement

Abstract

The quality of climate data is a critical concern for environmental analysis, climate modelling, and evidence-based decision-making. Within the WASCAL network, multiple meteorological stations located across diverse regions collect high-frequency measurements (every 10 minutes) of key climatic variables, including temperature. However, sensor failures and acts of vandalism often lead to missing values, outliers, or structural breaks, which compromise the integrity of the time series.
To address these issues, we designed a rigorous quality control protocol named the Climate Data Quality Control Protocol. This protocol was initially implemented for temperature, given its central role in numerous domains such as agriculture, hydrology, and public health. As part of this initiative, we developed TempQC, an interactive web application built with Streamlit (Python), to provide a systematic and reproducible assessment of temperature data quality.
The application translates into multiple modules, the methodology developed for the Data Quality Control Protocol which consists of the following steps: (i) detection of missing and duplicate values; (ii) filtering based on user-defined extreme thresholds; (iii) temporal consistency checks between successive observations; (iv) spatial consistency analysis using neighbouring stations; (v) statistical outlier detection; (vi) homogeneity testing to identify structural breaks; (vii) gap-filling using machine learning algorithms; and (viii) automated generation of a summary report documenting all quality control procedures applied.
This approach aims to strengthen the reliability and transparency of climate datasets used in both research and operational contexts by automating the quality control process and reducing human prone-errors. Future developments will include support for multi-variable datasets and the integration of advanced anomaly detection techniques.

Keywords: Anomalies, climate data, homogeneity, machine learning, quality control, Streamlit, temperature

Contact Address: Marcel Jocelyn Wendemi Michaelange TOE, West African Science Service Center on Climate Change and Adapted Land Use (WASCAL), Data Management, 00226 Ouagadougou, Burkina Faso, e-mail: toe.mwascal.org