Carcará: sistema para integração de dados meteorológicos e hidrológicos no semiárido brasileiro

Abstract

The Brazilian Semi-Arid region requires integrated access to meteorological and hydrological data to support environmental analyses and agricultural decision-making; however, this information is fragmented across multiple sources, with heterogeneous formats and low standardization. This work proposes Carcará, a system aimed at the automatic integration of data from sources such as INMET, ANA, and ERA5, covering data collection, standardization, storage, and availability. The solution is structured as an Extract, Transform, Load (ETL) pipeline composed of ingestion, messaging, a data lake, transformation, and loading into a data warehouse, orchestrated by a workflow manager. The system prioritizes scalability, traceability, and governance, enabling the consolidation of large data volumes with updates as they become available in the source repositories. As a result, an integrated dataset with 17,472 hourly records (192 timestamps) was generated, covering 44 municipalities, 8 states, and 91 reservoirs, with temporal alignment via a D−6 window (ERA5 lag); within the analyzed period, ERA5 achieved full coverage, ANA provided level/volume data for 58/91 reservoirs (63.7% of records), and INMET was the main completeness bottleneck (data for 20/44 municipalities and at least one meteorological field filled in 39.6% of records), validating end-to-end execution under real operational constraints; finally, the demonstrated applications include analyses of evaporative demand (RH vs. VPD), comparison of RH by state, drivers of PEV (SSRD, VPD, t2m, and WS), and the relationship between PEV and observed humidity, among others.