A fundamental challenge of modeling the vulnerability of coupled human-natural systems (e.g., energy, water, land) is that they are complex systems: their behaviors are not easily predicted due to non-linearities and feedbacks. Simulating these complexities requires coupling detailed, process-based models across multiple sectors and scales. For example, the dynamics of energy-water-land systems in a particular watershed could be addressed with fine-scale hydrologic and water management models coupled to agent-based models of local land use and irrigation and nested within regional-scale water management and electric grid operations models. This type of modeling is computationally-intensive and time-consuming and raises the following questions: How transferable are the insights from this research? What types of conclusions can be drawn regarding modeling needs for coupled systems in other locations? How might scientists choose their research testbeds in order to deliver the greatest scientific value with limited research resources? This paper develops and demonstrates a classification scheme for coupled human-natural systems across the U.S. based on the hypothesis that teleconnections (i.e., distal relationships) are an emergent property of these complex systems. For example, a simple system would be a city with a private, local watershed for its drinking water and a municipal utility operating a microgrid providing for its energy needs – zero teleconnections along energy-water dimensions. A highly complex system would be a city with multiple, distant watersheds supplying its drinking water, multiple urban, agricultural and power generation demands on those watersheds, and an electric utility operating in a market region dependent on water availability for remote hydropower and thermoelectric generation. Using high-resolution geospatial data derived from publicly-available sources, including urban drinking water source catchment areas, land use and land cover, locations of irrigation demands, and locations of water-dependent power generation, we identify and analyze over 100 interconnected energy-water-land networks in the U.S. and classify them according to the total number, diversity, intensity, and scales of their teleconnections.