Despite decades of fieldwork showing that flowing headwaters expand and contract in response to rainfall and runoff (influencing downstream water quantity, quality, and ecosystem health across temporal scales), we still do not explicitly integrate this phenomenon into watershed models and budgets. One reason is that we lack sufficient measurements of headwater temporal variability (both flowing length and width) because of the time-consuming nature of manual stream surveys. For example, at present there are only around 20 watersheds in the United States with these types of surveys and even then, most have no repeat surveys over time or across the full flow regime. To address this, we will test a novel monitoring framework using Arthur Brook, a first-order headwater stream network in the Harvard Forest, that will characterize stream network expansion and contraction in real-time. We will establish trail cameras in the dynamically flowing tributaries that image the stream every 15’, pairing these data to downstream gages, and leveraging computer vision algorithms to quantify when tributaries are flowing (and how much inundation) across flow conditions and events. We will also perform occasional manual surveys, measuring flow width every 5m. This will allow us to validate the approach while also building a dataset to quantify stream hydrological connectivity to the watershed. Finally, we will pair the monitoring framework with ongoing monitoring of river chemistry in Arthur Brook, as well as a handful of our miniDOT dissolved oxygen sensors installed in the catchment to measure how headwater variability affects downstream water quality. In conjunction with similar experiments in a small Connecticut watershed, our goal is to lay the groundwork for multi-watershed remote monitoring of headwater dynamics. This, in turn, will contribute to improved mechanistic understanding of stream wetting and drying across temporal scales.