Data and code for: Diagnosing renewable curtailment in Brazil with public data: Measurement shifts, mechanism-specific observability, and controlled-flow programming signals

Abstract

This repository contains the processed research data, monthly measurement partitions, final modeling datasets, frozen analytical outputs, documentation, and computational pipeline supporting the manuscript “Diagnosing renewable curtailment in Brazil with public data: Measurement shifts, mechanism-specific observability, and controlled-flow programming signals”. The study uses public data from the Brazilian National System Operator (ONS) to assess renewable curtailment observability, measurement changes, mechanism-specific event reconstruction, Revision-08 structural diagnostics, explainable machine-learning results, and controlled-flow programming signals. The deposit includes the harmonized curtailment dataset, monthly wind and solar measurement partitions, final hourly modeling datasets, v1.0.4 measurement/event/econometric outputs, v1.0.6 explainable-AI outputs, v1.1/v1.12 controlled-flow rolling-origin results, source manifests, authorship and CRediT documentation, checksums, and the computational pipeline required for reproduction of the reported analyses. The frozen analytical outputs are preserved unchanged for provenance and traceability. Some internal files use legacy labels such as “transmission_aware”; in the manuscript these are described more precisely as controlled-flow-programming augmented models.