Implementation of a feature-constraint mesh generation algorithm within a GIS Thomas J. Heinzer a , M. Diane Williams a , Emin C. Dogrul b,n , Tariq N. Kadir b , Charles F. Brush b , Francis I. Chung b a Michael Thomas Group, Inc., 541 45th St., Sacramento, CA 95819, USA b State of California, Department of Water Resources, Bay-Delta Office, Modeling Support Branch, Room 252-A, Sacramento, CA 95814, USA article info Article history: Received 23 March 2012 Accepted 1 June 2012 Available online 13 June 2012 Keywords: Mesh generation GIS Hydrologic modeling Constrained Delaunay triangulation Discretization Finite element abstract Unstructured mesh generation has long been a labor-intensive and time-consuming aspect of hydrologic modeling. This is largely due to the iterative nature of the process and the limited integration between the specialized mesh generation software and Geographic Information Systems (GIS). To facilitate mesh generation for integrated hydrologic models, a triangular finite-element mesh generator was directly embedded into a GIS platform. Additionally, a set of tools were developed to efficiently address some of the common issues that arise in a linked GIS-mesh generator software. These issues are (i) alignment of common geographic features that appear in multiple GIS features, (ii) redefining vertices with uniform spacing along GIS features, (iii) refinement of the mesh around selected GIS features, and (iv) efficient visualization of the generated mesh in the GIS platform. A GIS-platform extension integrating the mesh generator and the newly developed tools allows access to the full functionality of the GIS software for the analysis and reconditioning of geospatial datasets for mesh generation purposes. The linked GIS-mesh generator along with the newly developed tools facilitates efficient mesh generation and exploration of multiple meshes in an iterative manner without leaving the GIS platform. The features of the new software are highlighted using a realistic case study. & 2012 Elsevier Ltd. All rights reserved. 1. Introduction Integrated hydrologic models are commonly employed in water resources management and planning studies. These models allow their users to assess impacts to surface and subsurface water resources. These impacts may be related to urban and/or agricultural development in a basin or a predicted future state of climatic and land use conditions. Most integrated hydrologic models require a structured or unstructured computational mesh to be superimposed over the study domain where the mass and momentum conservation equations are discretized and solved. For models that operate on a structured mesh (e.g. finite difference models such as MODFLOW) (Harbaugh, 2005), it is straightforward to generate the mesh using a general-purpose spreadsheet application. Models that require unstructured meshes (e.g. finite element models such as MicroFEM) (Hemker, 2011), HydroGeoSphere (Blessent et al., 2009; Therrien et al., 2010), and Integrated Water Flow Model (IWFM) (Dogrul, 2011) can follow irregular boundaries to more closely represent flow lines, sub-boundary-level mass balances or detailed flow characteristics around important features such as pumping wells and streams. However, generating an unstructured mesh for integrated hydrologic models is generally difficult for problems of any size. The process involves analysis of complex geospatial data including physical (e.g. hydrography data, watershed boundaries, groundwater basin bound- aries) and operational (e.g. water district maps) water boundaries, digital elevation models (DEMs), and maps of land-use, soils and geology by using a Geographic Information Systems (GIS) software. The unstructured mesh generation is further complicated because it requires iterations between generating the mesh and superimposing it on the maps of underlying geospatial data until a mesh that accurately represents important boundaries with a desired resolution is achieved. Therefore, a tight integration between the specialized mesh generation software and the GIS is necessary to efficiently incorporate and process complex geospatial data sources in the iterative mesh generation process. There are many commercial or public-domain mesh genera- tors available (Owen, 1998; Gorman et al., 2008). Some of these mesh generators are standalone software and require the iterative process mentioned above to be performed manually, and some of them are linked to GIS. For instance, Argus ONE (Argus, 2011) allows ArcGIS (ESRI, 2011) shapefiles to be imported into their software platform to aid the users in developing their models. Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/cageo Computers & Geosciences 0098-3004/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.cageo.2012.06.004 n Corresponding author. Tel.: þ1 916 654 7018; fax: þ1 916 653 9574. E-mail addresses: tom@michaelthomasgis.com (T.J. Heinzer), diane@michaelthomasgis.com (M.D. Williams), dogrul@water.ca.gov (E.C. Dogrul), kadir@water.ca.gov (T.N. Kadir), cbrush@water.ca.gov (C.F. Brush), chung@water.ca.gov (F.I. Chung). Computers & Geosciences 49 (2012) 46–52