Results
Our investigation returned over 120 named processes; see the supplementary material for a complete list. Processes could be satisfactorily arranged into a hierarchical structure, with an initial division into surface, subsurface and channel processes, and a maximum of five underlying layers. Some processes had ambiguity in their position, for example transient saturation at the bedrock interface could fall into soil or groundwater processes categories. Processes often had many alternative names, for example runoff generation by displacement of groundwater is also referred to as pipeflow, pistonflow, old water mobilization, pre-event water mobilization, translatory flow, pressure wave and hydraulic displacement.
Each process was assigned a unique identifier, similar to a hashtag. The identifiers use dots to signify hierarchical level, with each level given a text tag of at most six characters. Thus, change in grain size during snowpack aging is classified as Surface - Snow and Cold Region Processes – Snow Storage – Snowpack Aging – Change in grain size, with the identifier Surf.Cold.Snow.Age.Grain . This provides a short reference string for any process. If a different hierarchical structure is used in future, it would be relatively simple to remap the identifiers.
Some names could be disputed as representing stores rather than processes, and might be classed as a physical entity in an ontological framework. However, water storage has long been regarded as part of watershed function, and was therefore included in the taxonomy. Functional classes are listed in Table 1, and give more detail than previous classifications, for example differentiating between ‘store’, ‘filling of store’ and ‘release from store’. For example, snowmelt is designated as ‘release from store’ as only the origin store is specified. We created an interactive version of the taxonomy that can be accessed at http://mcmillanhydrology.org/ProcessTaxonomy/ProcessTaxonomyDiagram.html . A view of the interactive taxonomy with overland flow processes expanded is given in Figure 1.
When hydrologists described watershed function, a variety of information types were provided. Information could relate to the existence or magnitude (of stores, fluxes), spatial variation (location), temporal variation (seasonal variation or wetness conditions for processes to occur or stores to fill), or response time. These types could be used as an additional descriptor of hydrological information when stored in a database or similar.