Local and distant archives of observed weather data present unique opportunities for scientists to obtain long time series of the historical hydrology and climate for many regions of the world. Unfortunately, most of these observational records are still to-date available only on paper, and thus require digitization and transcribing to machine-readable formats to facilitate analysis of hydroclimatic trends. Here we discuss the data rescue efforts for hydroclimatic data recorded at 36 climate stations in the Democratic Republic of Congo from the early 1950’s to-date. We describe the procedures we follow to digitize over 10,000 paper records of daily precipitation and temperature within archives both in the Democratic Republic of Congo and Belgium, and subsequently the steps to transcribe this data set using different methods including machine learning. Furthermore, we undertake quality control and quality assessment of the transcribed data. The resultant time series, comprised of millions of observations from the archived data, will resolve the challenges of limited available hydroclimatic data within the Congo basin and expedite research on the hydroclimate in the basin.
No takes yet. Share an insight, caveat, or question.
Muheki et al. (2024) studied this question.