RESEARCH
.Bridging the Gap between Geohydrologic Data and Distributed Hydrologic Modeling, International Environmental Modelling and Software Society (iEMSs), 2008
Abstract: This paper outlines and demonstrates a strategy for coupling of integrated hydrologic model and Geographic Information System (GIS) to meet pre/post processing of data and visualization. Physically based fully distributed integrated hydrologic models seek to simulate hydrologic state variables and their interactions in space and time. The process requires interaction with a range of heterogeneous data layers such as topography, soils, hydrogeology, climate, and land use. Clearly, this requires a strategy for defining topology definitions, data gathering and development. Traditionally GIS has been used for data management, analysis and visualization. Integrated use and streamlineed development of sophisticated numerical models and commercial Geographic Information Systems (GISs) poses challenges inherited from proprietary data structures, rigidity in their data-models, non-dynamic data interaction with pluggable software components and platform dependence. Independent hydrologic modeling systems (HMSs), GISs and Decision Support Systems (DSSs) not only increase model setup and analysis time but they also result in data isolation, data integrity problems and broken data flows between models and the tools used to analyze their inputs and results. In this paper we present an open-source, extensible and pluggable architecture, platform independent “tightly-coupled” GIS interface to Penn State Integrated Hydrologic Model (PIHM) called PIHMgis. The tight- coupling between the GIS and the model is achieved by the development of PIHMgis shared-data model to promote minimum data redundancy and optimal retrievability [Kumar et al., 2008]. The procedural framework of PIHMgis is demonstrated through its application to Shaver’s Creek Watershed located in Susquehanna River Basin in Pennsylvania.
An efficient domain decomposition framework for accurate representation of geodata in distributed hydrologic models, International Journal of Geographical Information Science, IJGIS-2008-0069.R1
Abstract: Physically-based, fully-distributed hydrologic models simulate hydrologic state variables in space and time while using information regarding heterogeneity in climate, land use, topography and hydrogeology. Since fine spatio-temporal resolution and increased process dimension will have large data requirements, there is a practical need to strike a balance between descriptive detail and computational load for a particular model application. In this paper we present a flexible domain decomposition strategy for efficient and accurate integration of the physiographic, climatic and hydrographic watershed features. The approach takes advantage of different GIS feature types while generating high-quality unstructured grids with user-specified geometrical and physical constraints. The framework is able to anchor the efficient capture of spatially distributed and temporally varying hydrologic interactions and also ingest the physical prototypes effectively and accurately from a geodatabase. The proposed decomposition framework is a critical step in implementing high quality, multiscale, multiresolution, temporally adaptive and nested grids with least computational burden. We also discuss the algorithms for generating the framework using existing GIS feature objects. The framework is successfully being used in a finite volume based integrated hydrologic model. The framework is generic and can be used in other finite element/volume based hydrologic models.
Towards a Dynamic Digital Observatory: Synthesizing Community Data and Model Development in the Susquehanna River Basin and Chesapeake Bay, Eos Trans. American Geophysical Union, 88(52), Fall Meeting 2007
Abstract: Physically-based fully-distributed hydrologic models simulate hydrologic state variables spatiotemporally using information on forcing (climate) and landscape (topography, land use, hydrogeology) heterogeneities. Incorporating physical data layers in the hydrologic model requires intensive data development. Traditionally, GIS has been used for data management, data analysis and visualization; however, proprietary data structures, platform dependence, isolated data model and non-dynamic data-interaction with pluggable software components of existing GIS frameworks, makes it restrictive to perform sophisticated numerical modeling. In this effort we present a "tightly-coupled" GIS interface to Penn State Integrated Hydrologic Model (PIHM; www.pihm.psu.edu) called PIHMgis which is open source, platform independent and extensible. The tight coupling between GIS and the model is achieved by developing a shared data-model and hydrologic-model data structure. Domain discretization is fundamental to the approach and an unstructured triangular irregular network (e.g. Delaunay triangles) is generated with both geometric and parametric constraints. A local prismatic control volume is formed by vertical projection of the Delaunay triangles forming each layer of the model. Given a set of constraints (e.g. river network support, watershed boundary, altitude zones, ecological regions, hydraulic properties, climate zones, etc), an "optimal" mesh is generated. Time variant forcing for the model is typically derived from time series data available at points that are transferred onto a grid. Therefore, the modeling environment can use the Observations Database model developed by the Hydrologic Information Systems group of the Consortium of Universities for the Advancement of Hydrologic Sciences, Inc. (CUAHSI). As part of a initial testbed series the database has been implemented in support for the Susquehanna and Chesapeake Bay watersheds and is now being populated by national (USGS-NWIS; EPA- STORET), regional (Chesapeake Information Management System, CIMS; National Air Deposition Program, NADP), and local (RTH-Net, Burd Run) datasets. The data can be searched side by side in a one-stop-querying- center, www.hydroseek.org , another application developed as part of the CUAHSI HIS effort. The ultimate goal is to populate the observations database with as many catalogues (i.e. collections of information on what data sources contain) as possible including the build out of the local data sources, i.e. the Susquehanna River Basin Hydrologic Observatory System (SRBHOS) time series server.
Using Integrated Hydrologic Models to Trace the Source and Dynamics of Fresh Water Discharge in a Coastal Watershed, American Geophysical Union, Fall Meeting 2007
Abstract: Quantization of freshwater discharge to estuary is vital to coastal aquatic life and several management practices and ecosystem restoration projects as it is very closely related to the salinity and chemical characteristic of the estuary. Freshwater-discharge directly to estuary is often ignored as it is difficult to measure. However, it is important to identify the sources of freshwater and their dynamics to address many issues such as water quality. In this research, the Penn State Integrated Hydrologic Model (PIHM) is used to investigate the spatial and temporal behavior of freshwater discharge in Rhode River Basin, an experiment watershed operated by Smithsonian Environmental Research Center (SERC). PIHM is a multi-process, multi-scale hydrologic model where the major hydrological processes are fully coupled using the semi-discrete finite volume method. The model is used to partition surface and groundwater discharges to estuary and streams. It is shown that the groundwater discharge directly to estuary is substantial (overall approximately 25% of total discharge) as compared to baseflow to streams. Noticeably, during dry periods they are comparable (groundwater discharge to the estuary is approximately 43% of the total discharge). It is also demonstrated that, the discharge directly to estuary is forced by precipitation events and topography and varies widely across the estuary boundary in magnitude but it is mostly controlled by the tide events. As ongoing research “Age” of water theory is being developed which will allows looking at the circulation of tidal water in the basin and exact estimation of the “age” of freshwater discharge.
PIHMgis: A "Tightly Coupled" GIS Framework for Integrated Hydrologic Modeling, American Geophysical Union, Fall Meeting 2006
Abstract: Physically-based fully-distributed hydrologic models try to simulate hydrologic state variables in space and time while using information regarding heterogeneity in climate, land use, topography and hydrogeology. However incorporating a large number of physical data layers in the hydrologic model requires intensive data development and topology definitions. Proprietary data structures, platform dependence, isolated data model and non-dynamic data-interaction with pluggable software components of existing GIS frameworks, makes it restrictive to perform sophisticated numerical modeling. Here we present a "tightly-coupled" integrated GIS interface to Penn State Integrated Hydrologic Model (PIHM) called PIHMgis which is open source, platform independent and extensible. The tight coupling between GIS and the model is achieved by developing a shared data-model and hydrologic-model data structure
PIHMgis: More details can be found here. Straight to download section here
Coupling hydrologic processes across different spatio-temporal scales: The Juniata River Basin, American Geophysical Union, 2006 Joint Assembly
A spatially-distributed, physically-based, fully-coupled PennState Integrated Hydrologic model (PIHM) which simulates surface, subsurface, streamflow and climatic state variables in space and time is presented. The basic idea of PIHM (Qu, 2004, PhD Thesis) is to first identify the hydrologic relationships and the corresponding physical equations on each discretized unit elements of the model domain. All physical equations are integrated over an elemental control volume to obtain semi-discrete ordinary differential equations (ODEs) which is referred as a model kernel. Model kernel equations from all across the watershed are assembled and solved simultaneously. Channel routing and overland flow has been handled using St. Venant's equation, unsaturated and groundwater flow using Richards' equation, evaporation using Pennman-Monteith equation and snow melt using energy based SNOBAL algorithm. Due to significantly different time scales of the interacting hydrologic processes, the resulting ODE system is stiff and is solved using CVODE (http://www.llnl.gov/casc/sundials/), a state of the art ODE solver. The modeling framework can be universally applied from hillslope to catchment to synoptic scales with different number and approximation of process equations depending on model purpose and computational constraint. Coupling of the hydrologic processes across disparate spatio-temporal scale is demonstrated in Juniata River basin which is located in south-central Pennsylvania, encompassing 3,400 square miles. A computationally efficient large scale model implementation in the study area is facilitated by nested- modeling strategy, adaptive refinement-derefinement modeling strategy and model parallelization.
Automated Detection and Spatio-Temporal Classification of Channel Reaches in Semi-arid Southwestern US using ASTER, American Geophysical Union, 2006 Joint Assembly
Abstract: Groundwater-surfacewater dynamics are strikingly different in ephemeral, intermittent and perennial channels. So it is imperative to identify their spatial distribution over a large basin for reducing uncertainties in aquifer recharges and transmission losses while performing water budget calculations. This study describes an automated method for detection of stream channels and their classification into ephemeral, intermittent and perennial reaches using remote sensing ASTER Level 1B data. The methodology involves calculation of normalized difference water index map of the area of interest followed by bi-level thresholding to obtain watered channel network at the time of image acquisition. The aforesaid process is repeated over a set of multi-temporal images followed by application of change-detection algorithm to obtain a classified map of the channel network in reaches of different time scales.
Evaluation of Traditional and Nontraditional Optimization Techniques for Determining Well Parameters from Step-Drawdown Test Data, Journal of Hydrologic Engineering, Vol. 11, No. 6, November 1, 2006.
Abstract: Adequate knowledge of the hydraulic characteristics of production wells is indispensable for the proper development and management of wells and for the selection of suitable pumps. In this study, the characteristic hydraulic parameters of production wells were determined by the widely used graphical analysis of step-drawdown pumping test data as well as the two traditional gradient-based nonlinear optimization techniques (viz., Levenberg–Marquardt and Gauss–Newton) and the nontraditional optimization technique, genetic algorithm. Three stand-alone and interactive computer programs were developed to optimize the hydraulic parameters of production wells by these numerical techniques. The efficacy and robustness of the developed computer codes were examined using eleven sets of step-drawdown data from diverse hydrogeologic conditions. The results of this study revealed that all the three numerical techniques yielded superior well parameters with lower values of root-mean-square errors (RMSE) for all the eleven data sets compared to the parameters obtained by the graphical method. The values of the exponent (P) ranged from 1.2 to 6.0, which gave minimum RMSE values. Furthermore, it was found that both the Gauss–Newton and Levenberg–Marquardt techniques are sensitive to the initial guess values of well parameters. It is concluded that both the traditional and nontraditional numerical techniques offer an efficient and reliable tool for the determination of well parameters with a greater accuracy and better insight.
Impact of dynamic response characteristics on uncertain streamflow prediction
Abstract: The influence of various dynamic watershed response characteristics as constraints on uncertain model prediction is studied. A new approach to predictions in ungauged basins is presented in this paper where dynamic response characteristics are regionalized to watershed physical characteristics. The approach therefore uses a data driven regionalization method under uncertainty rather than the standard hydrologic model driven regionalization. Uncertainty is propagated into the model predictions from ungauged basins in the form of constraints on acceptable hydrologic model behavior. This study tries to look at the impact of different response characteristics as constraints on streamflow prediction and on different parts of streamflow, driven, non driven flow and peak flows. Initial results identify key response characteristics and their importance in terms of the characteristics function of the watershed. Full Text
Master of Technology Thesis: Development of Software for Furrow Irrigation
Surface irrigation is practiced in most of the irrigated land. The main problems faced in surface irrigation are non-uniformity and low efficiency. Poor design and management are generally responsible for inefficient irrigation, leading to wastage of water, water-logging, salinization and pollution of surface water and groundwater resources. Non-uniform application of water results in water-stressed conditions for crops in some part of the field, while over-irrigation leads to wastage of water through runoff from the end of the field and deep percolation below the root zone.
Among surface irrigation methods, furrow Irrigation is most widely practiced method of irrigation for the row crops like maize, sugarcane, cotton, and vegetables. The practice of furrow irrigation has traditionally been based upon experience gained through experimentation over many seasons.
In furrow irrigation, due to infiltration, there are variations in the flow depth at furrow section with time as well as the flow rate with furrow length. Furthermore, the depth of flow along the furrow length varies gradually. For these reasons, the flow in furrow irrigation is and example of unsteady, non-uniform and gradually varied flow over a porous bed. The basic design parameter such as furrow length, slope and cross-section, flow rate, infiltration characteristics and hydraulic resistance interact with each other during an irrigation event and consequently affect the advance and recession curves during irrigation.
The presence of large number of independent parameters makes the analysis of furrow irrigation quite complex. A comprehensive study of all the parameters involved through field trials is difficult. Mathematical modeling that related system performance to design parameters is an alternative approach. A number of surface irrigation models that include sub models for simulating border or furrow irrigation or both are currently available for simulating irrigation events. Implementing these models as software is a need so that irrigation events can be managed properly. Therefore, to meet this necessity, it was decided to develop a furrow-irrigation-simulation-software having different mathematical models. The specific objectives of the project are listed below:
(1) To develop a user-friendly software for the hydraulic simulation of furrow irrigation.
(2) To test the accuracy of the developed software with field observed data.
(3) To compare the performance of Kinematic-Wave model, Volume-Balance model, Zero Inertia model and Full Hydrodynamic model with respect to observed data.
Bachelor of Technology Thesis: Development of Software for Furrow Irrigation
