In this project, we develop methods for using satellite imagery to nowcast Permian Basin gas flaring and test whether those signals can estimate oil production prior to delayed state reports. We evaluate two approaches on a 2012–2025 dataset combining VIIRS Nightfire radiant heat with operator records. The first approach is an ARIMAX model using real-time satellite data to predict monthly flared gas volumes weeks before official releases. This method lifts directional accuracy from 56.5% to 68.5% over a 92-month backtest, outperforming baselines. The second approach attempts to predict oil production directly. We find this fails after 2019 because new pipelines dropped flaring to 2% of total gas while gas-oil ratios doubled, amplifying estimation errors. Ultimately, we demonstrate that satellite radiant heat reliably tracks flaring but cannot infer oil production. Our approach serves as an early, objective tool for analysts and regulators to monitor regional flaring.