2 Methodology and Materials

2.1 Data sources

The literature data adopted for this research were acquired from two common and influential scientific databases, i.e. the Science Citation Index Expanded (SCI-E) and the Social Science Citation Index (SSCI) of WOS (Wu, 2015). These following terms were used to retrieve related publications: TS = (“uncertain*” AND “water” AND (“modeling” OR “simulat*”) AND (“basin” OR “watershed”) AND (“manage*” OR “allocation”)) (“TS” represents an article subject and “*” represents a fuzzy search). Under these conditions, a total of 2,020 documents with full bibliographic records were retrieved and downloaded as related research from 1991 to 2018.

2.2 Statistical methods

CiteSpace is a popular and widespread tool in the bibliometric study, visualizing frontier knowledge, and constructing a network in the domain of science (Chen 2004). In the view maps generated by CiteSpace, nodes represent items (each item has a keyword, author, journal, and country information), and links represent a co-citation structure. Additionally, every node illustrates three types of colours and different thicknesses to represent its centrality value (Chen, 2006; Xie et al., 2015)). For example, the red ring on a node demonstrates a burst discovered while the purple rim indicates high betweenness centrality (≥0.1), which represents the significance of the whole network (Freeman, 1979; Liu et al., 2015; Chen, 2010).
The following landscape views were acquired from publications in 1991-2018, and the 50 most cited journals were used to build a knowledge network originally. Afterwards, each network was generated and enclosed 2,020 references. The time horizon from 1991 to 2018 was divided into three periods (i.e., 1991-1999, 2000-2009, and 2010-2018). Five types involving author, institution, country, keyword, and cited reference were fixed according to the research need. A few parameters were set as defaults. Then, the collaboration network and co-occurrence network were analyzed based on the frequency, burst, and centrality to identify research characteristics and trends in uncertainty watershed modeling and management.