Francisco Olivera

Francisco Olivera, Ph.D., P.E.
Associate Professor

Texas A&M University
Zachry Department of Civil Engineering
3136 TAMU
College Station, Texas 77843-3136

Tel.: (979) 845-1404 - FAX: (979) 862-1542
Web page: http://ceprofs.tamu.edu/folivera
e-mail: folivera@civil.tamu.edu

*** You need a subscription to the journal to have online access to some of the articles. ***

In this paper, we present a novel multi-modal optimization algorithm for finding local optima in model validation problems. We build from Particle Swarm Optimization(PSO) by using deterministic sampling to generate new particles during the optimization process, by implementing proximity-based speciation coupled with speciation of isolated particles, and by including “turbulence regions” around already found solutions to prevent unnecessary function evaluations. We compare the performance of the new algorithm (Isolated Speciation-based PSO or ISPSO) with the performance of Species-based PSO and NichePSO on a variety of standard test problems. ISPSO outperforms these algorithms in terms of computational cost, consistency, and scalability.

The spatial variability of the data used in models includes the spatial discretization of the system into subsystems, the data resolution, and the spatial distribution of hydrologic features and parameters. In this study, we investigate the effect of the spatial distribution of land use, soil type, and precipitation on the simulated flows at the outlet of “small watersheds” (i.e., watersheds with times of concentration shorter than the model computational time step). The Soil and Water Assessment Tool (SWAT) model was used to estimate runoff and hydrographs. Different representations of the spatial data resulted in comparable model performances and even the use of uniform land use and soil type maps, instead of spatially distributed, was not noticeable. It was found that, although spatially distributed data help understand the characteristics of the watershed and provide valuable information to distributed hydrologic models, when the watershed is small, realistic representations of the spatial data do not necessarily improve the model performance. The results obtained from this study provide insights on the relevance of taking into account the spatial distribution of land use, soil type, and precipitation when modeling small watersheds.

A method to identify and track rainfall structures using one-hour accumulated NEXt generation RADar (NEXRAD) rainfall data is presented and used to analyze the dynamics of storm features over an area in Texas. Storm features are identified from a Gaussian mixture model using the expectation maximization algorithm. The method assigns NEXRAD pixels to storm features while simultaneously producing a smooth fitted function to the rainfall intensity distribution. Once the storm features are identified, they are tracked using inverse cost functions and using the fact that continuous features overlap each other from frame to frame in the accumulated data. The inverse cost functions also account for storm feature merging, splitting, birth, and death. Application of this storm identification and tracking algorithm for Brazos County (1,500 square km) in southeastern Texas distinguishes several characteristics of the storm feature dynamics. From September through April, storm features are predominantly of a frontal nature, with storm features following geostrophic flow along low pressure fronts moving in from the north. In summer (May-August), storm features are convective in nature following random track directions. Both types of storm features have durations of one to three hours in Brazos County due to the county’s relatively small size compared to the measured average storm speed of 40 km/hr and due to the fact that most storms only intersect the county over part of their area.

The Griewank function is commonly used to test the ability of different solution procedures to find local optima. It is important to know the exact number of minima of the function to support its use as a test function. However, to the best of our knowledge, no attempts have been made to analytically derive the number of minima. Because of the complex nature of the function surface, a numerical method is developed to restrict domain spaces to hyperrectangles satisfying certain conditions. Within these domain spaces, an analytical method to count the number of minima is derived and proposed as a recursive functional form. The numbers of minima for two search spaces are provided as a reference.

This paper presents key issues associated with uncertainty in flood inundation mapping. Currently, flood inundation extent is represented as a deterministic map without consideration to the inherent uncertainties associated with various uncertain variables (precipitation, stream flow, topographic representation, modeling parameters and techniques, and geospatial operations) that are used to produce it. Therefore, it is unknown how the uncertainties associated with topographic representation, flow prediction, hydraulic model and inundation mapping techniques are transferred to the flood inundation map. In addition, the propagation of these individual uncertainties and how they affect the overall uncertainty in the final flood inundation map is not well understood. By using a sample dataset for Strouds Creek in North Carolina, we highlight key uncertainties associated with flood inundation mapping. In addition, we articulate the idea of probabilistic flood inundation map, and present an integrated framework approach that will connect data, models and uncertainty analysis techniques in producing probabilistic flood inundation mapping. The proposed framework will address both the propagation of uncertainty or errors from model inputs and parameters up to the final output, and also the assessment of the relative importance of input uncertainties on the output uncertainty.

A method to disaggregate daily rainfall into hourly precipitation is evaluated across Texas. Based on Socolofsky et al. (2001), the method chooses representative storm intensity patterns from measured hourly databases to generate the synthetic data using a single parameter for the smallest expected one-hour event. The model is applied across Texas using historic hourly precipitation data; performance is evaluated by the model’s ability to reproduce hourly rainfall statistics. Based on a cluster analysis to determine which precipitation databases should be used for disaggregation, no trends in space or among gauge characteristics (e.g. period of record, precipitation statistics) were identified. As a result, a Texas state database containing all the measured hourly data in Texas is proposed for use by disaggregation. The state database is verified for a selection of gauges and performed as well at a given station as using that station’s measured rainfall for the disaggregation. The method is further applied to estimate intensity–duration curves which show that the method matches the majority of storm intensities needed to track soil moisture and diverges by less than 17% for the extreme runoff-generating events.

Precipitation areal reduction factors (ARFs) for the 685,000 km2 of Texas were calculated using NEXRAD rainfall estimates. The study was based on 18,531 storms of different durations that took place in different seasons and regions of Texas. The rainfall field was considered anisotropic, and the storms were assumed of elliptical shape. It was found that, in addition to the storm duration and area, other factors such as the season, region and precipitation depth have a statistically significant effect on the ARFs. Elongated ellipses and orientation angles somewhat parallel to the Texas gulf coast were found more frequent in winter, when warm and cold fronts produce frontal storms, than in summer. The effect of the precipitation depth on the ARFs was found to be stronger in summer than in winter. Even though part of the ARF variability could be explained by seasonality, regionality and precipitation depth, the uniqueness of each storm event appears to be an important cause of it. Lower ARF values were observed compared to previous studies.

The primary objective of this study is the documentation of stormwater quality of vegetated roadsides of two Texas highways (State Highway 6 in College Station and Loop 360 in Austin. Both had high average daily traffic (ADT). Three sites each in Austin and College Station were monitored using passive "first flush" stormwater samplers for 16 months. Results from this study indicate that a significant removal of sediment and heavy metals occur over the width of vegetated roadsides. In contrast, vegetated roadsides seem ineffective in reducing nutrient (nitrogen and phosphorus) concentrations. The results also show that vegetation density has a direct effect on the performance of vegetated roadsides. When roadsides are densely covered with grasses above 90%, significant sediment removal is expected often within the first four meters of the edge of pavement. A stepwise regression analysis identifies the antecedent dry period (ADP) as the most significant predictor to pollutant concentration. The pollutant event mean concentration (EMC) was found to decrease with increasing ADP for all pollutants at the College Station sites, but not the Austin ones.

The capacity of a watershed to urbanize without changing its hydrologic response and the relationship between that response and the spatial configuration of the developed areas was studied. The study was conducted in the Whiteoak Bayou watershed (223 km2), located northwest of Houston, Texas, over an analysis period from 1949 to 2000. Annual development data were derived from parcel data collected by the Harris County Appraisal District. Using these data, measures of the spatial configuration of the watershed urban areas were calculated for each year. Based on regression models, it was determined that the annual runoff depths and annual peak flows depended on the annual precipitation depth, the developed area and the maximum 12-hour precipitation depth on the day and day before the peak flow took place. It was found that, since the early 1970s, when the watershed reached a 10% impervious area, annual runoff depths and peak flows have increased by 146% and 159%, respectively. However, urbanization is responsible for only 77% and 32% of the increase, respectively, while precipitation changes are responsible for the remaining 39% and 96%, respectively. Likewise, an analysis of the development data showed that, starting in the early 1970s, urbanization in the watershed consisted more of connecting already developed areas than of creating new ones, which increases the watershed’s conveyance capacity and explains the change in its response. Before generalizing conclusions, though, further research on other urban watersheds with different urbanization models appears to be necessary.

This paper presents ArcGIS-SWAT, a geodata model and GIS interface for the Soil and Water Assessment Tool (SWAT). The ArcGIS-SWAT data model is a system of geodatabases that store SWAT geographic, numeric and text input data and results in an organized fashion. Thus, it is proposed that a single and comprehensive geodatabase be used as the repository of a SWAT simulation. The ArcGIS-SWAT interface uses programming objects that conform to the Component Object Model (COM) design standard, which facilitate the use of functionality of other Windows-based applications within ArcGIS-SWAT. In particular, the use of MS Excel and MATLAB functionality for data analysis and visualization of results is demonstrated. Likewise, it is proposed to conduct hydrologic model integration through the sharing of information with a not-model-specific hub data model where information common to different models can be stored and from where it can be retrieved. As an example, it is demonstrated how the Hydrologic Modeling System (HMS) – a computer application for flood analysis – can use information originally developed by ArcGIS-SWAT for SWAT. The application of ArcGIS-SWAT to the Seco Creek watershed in Texas is presented.

Traditionally, stream and sub-watershed characterization in GIS has been done using a DEM-based terrain analysis approach; however, there is a large amount of existing vector hydrographic data difficult to accurately reproduce using DEMs. WaterNet is a GIS/hydrologic application for the integration and analysis of stream and sub-watershed networks in vector format. Even with vector data, hydrologic inconsistencies between streams and sub-watersheds do exist, and are revealed in the form of streams crossing drainage divides and sub-watersheds with more than one outlet. WaterNet rectifies these inconsistencies and couples the two datasets. Most algorithms involving traces of dendritic networks employ a form of tree traversal which requires topologic information to be organized into specialized data structures. On the contrary, WaterNet develops topologic relationships from GIS attributes table, which, in combination with sorting and querying algorithms, make the calculation process efficient and easy to implement. With the topologic relationships of the streams and sub-watersheds, WaterNet can perform traces to calculate cumulative network parameters, such as flow lengths and drainage areas. WaterNet was applied to the catchment of the Texas Gulf coast for a total of 100 cataloging units (411,603 km2) and 60,145 stream lines (183,228 km).

A model for estimating the watershed response is presented and used for assessing how the watershed size and spatial variability of the hydrodynamic parameters (i.e., wave celerity and dispersion coefficient) affects the comparative importance of advective processes with respect to hydrodynamic dispersive processes. A parameter W was defined to quantify this comparative importance. This parameter represents the fraction of the watershed response variance that is explained by advection. A series of simulations were performed for basins of different sizes and different spatial distributions of their hydrologic parameters. It was found that, for spatially uniform hydrodynamic parameters, the effect of hydrodynamic dispersion decreases compared to that of advection as the watershed size increases, and vice versa; and that, for non-uniform hydrodynamic parameters, the spatial distribution of the parameter values over the watershed, in conjunction with the watershed size, determines which process – advection or hydrodynamic dispersion – prevails.

This paper aims at comparing the output of two methodologies in the delimitation and characterization of a catchment located at the Fazenda Experimental Vale do Curu in Pentecoste County - Brazil (3°48'49.1"S; 39°20'17.8"W). Results from the extension CRWR-PrePro under ARCVIEW environment were compared with the traditional (planimeter and curvimeter) procedure. The database was extracted from a topographic map scaled to 1:5,000 and elevation contours spaced every 5 m. The topographic map was sampled to produce a 50 x 50 m grid of elevation by interpolating between contours and generating a Digital Elevation Map (DEM). Using the extension CRWR-PrePro under ArcView GIS 3.2 two catchments above 300 ha were identified. One of the catchments was selected for further analysis. The watershead was characterized using both CRWR-PrePro and the planimeter/curvimeter procedure, and the following parameters were computed: drainage divide length, catchmentís area, number of stream, total stream length, drainage density, main stream length, mean main stream slope, catchmentís length, mean catchmentís slope, form ratio and elongation ratio. Results shown that there CRWR-PrePro underestimation the parameters catchmentís area (5.4%), mean main stream slope (8.7%), mean catchmentís slope, (5.4%), form ratio (16.7%) e elongation ratio (8.1%) and overestimation the parameter drainage divide length (24.6%), total stream length (17.4%), drainage density (24.1%), main stream length (8.4%) e catchmentís area (4.8%). The number of streams was the same to the two methods, leading to the conclusion that both methodologies produce closer results with the advantage of standardization, easier and faster analysis by using the CRWR-PrePro extension upon availability of a Digital Elevation Map.

The Network Tracing Method (NTM) has been developed to determine gridded coarse river networks for modeling large hydrologic systems. For a coarse-resolution grid, the NTM determines the downstream cell of each cell, and the distance along the actual meandering flow paths between them. As opposed to previously developed methods, the NTM uses fine-resolution vector river networks as the source of information of the flow patterns, rather than digital elevation models. The main advantage of using vector river networks as input is that they capture the hydrologic terrain features better than topographic data do, particularly in areas of low topographic relief. The NTM was applied to South America with a grid resolution of 1°×1°, and to the globe with a resolution of 2.815°×2.8125°. Overall, the method captured the flow patterns well. Generated digital river networks and drainage divides showed minor disagreement with those obtained from existing maps, and most of them were consistent with the resolution of the coarse river network. The majority of estimated basin areas were also close to documented values. River lengths calculated with the NTM, however, were consistently underpredicted.

Including global river networks in the land component of global climate models (GCMs) is necessary in order to provide a more complete representation of the hydrologic cycle. The process of creating these networks is called river-network upscaling, and consists of lowering the resolution of already available fine networks to make them compatible with GCMs. Fine resolution river networks have a level of detail appropriate for analysis on the watershed scale, but are too intensive for global hydrologic studies. A river-network upscaling algorithm, which processes fine-resolution digital elevation models to determine the flow directions that best describe the flow patterns in a coarser user-defined scale, is presented. The objectives of this study were to develop an algorithm that advances the previous work in the field by being applicable at a global scale, allowing for the upscaling to be performed in a projected environment, and generating evenly distributed flow directions.

The hydro network is the backbone of Arc Hydro, created from edges and junctions. The topological connection of its HydroEdges and HydroJunctions in a geometric network enables tracing of water movement upstream and downstream through streams, rivers, and water bodies. Relationships built from the HydroJunctions connect drainage areas and point features such as stream-gauging stations to the hydro network. Locations in the hydro network are defined by a river-addressing scheme that defines where points are located on lines within drainage areas, allowing measurement of flow distance between any two points on a flow path.

Water and land interact with one another: the shape of the land surface directs the drainage of water through the landscape, while the erosive power of water slowly reshapes the land surface. Streams, rivers, and water bodies lie in the valleys and hollows of the land surface, and drainage from the ridges and higher land areas flows downhill into these water systems. Digital elevation models (DEMs) are used to analyze the drainage patterns of the land-surface terrain, and drainage areas are delineated from outlets chosen either manually or automatically according to physical rules. Raster analysis using fine-resolution DEMs is practical only over limited areas, but these results may be combined with vector networks to carry out regional hydrologic studies. Drainage areas can be traced upstream and downstream, either through their attachment to the hydro network or by using area-to-area navigation, thereby identifying the region of hydrologic influence upstream and downstream of a catchment or watershed.

Computer models play a pivotal role in hydraulic analysis by aiding in the determination of water surface profiles associated with different flow conditions. Many hydraulic models contain a wealth of detailed cross-section data that were developed specifically for the modeling effort, typically from land surveys. Unfortunately, these high-resolution data are generally stored in hydraulic model coordinates, a format that does not maintain the geographic coordinates of the cross-sections. As a result, cross-section data typically must be digitized in order to assign them geographic coordinates. Automating this process would result in significant savings of both time and resources. The research presented in this paper offers an automated geographic information system (GIS) based approach for the development of a terrain model from output of the HEC-RAS hydraulic model. As input, the approach requires a completed HEC-RAS model, a digital elevation model (DEM) of the study area, and a GIS representation of the stream thalweg. The process begins with the conversion of terrain data from hydraulic model coordinates to geographic coordinates. A terrain model is subsequently synthesized by merging the HEC-RAS data, which describe the stream channel geometry, and comparatively lower resolution DEM data. The resulting surface model provides a good representation of the general landscape and contains additional detail within the stream channel. An example application to Waller Creek in Austin, Texas is presented.

A methodology is presented for extracting topographic, topologic and hydrologic information from digital spatial data of a hydrologic system, for hydrologic modeling with HMS. Methods for stream and watershed delineation from digital elevation models are presented. After vectorizing the raster-based stream segments and watersheds, navigation fields are added to their tables to support topologic analysis based entirely on the information stored in the tables of the spatial datasets. Algorithms for calculation of stream segment and watershed hydrologic parameters are also discussed. These parameters are curve number, area and lag-time for the watersheds, and routing model and travel-time for the stream segments. Once the hydrologic parameters and the topology are defined, a basin model that summarizes all the information and that can be used as input for HMS is created. Using this methodology, the determination of the spatial parameters for HMS is an automatic process that accelerates the setting up of a hydrologic model and leads to reproducible results.

In this paper, the development and global application of a new approach to large-scale river routing is described. It differs from previous methods by the extent to which the information content of high-resolution global digital elevation models is exploited in a computationally-efficient framework. The model transports runoff directly from its source of generation in a land model cell to its sink on a continental margin or in an internally draining basin (and hence is referred to as source-to-sink routing) rather than from land cell to land cell (which we call cell-to-cell routing). It advances the development of earlier source-to-sink models by allowing for spatially-distributed flow velocities, attenuation coefficients and loss parameters. The method presented here has been developed for use in climate system models, with a specific goal of generating hydrographs at continental margins for input into an ocean model. However, the source-to-sink approach is flexible and can be applied at any space-time scale, and in a number of other types of large-scale hydrological and Earth system models. Hydrographs for some of the world's major river basins resulting from a global application, as well as hydrographs for the Nile River from a more detailed application, are discussed.

CRWR-PrePro is a system of ArcView scripts and associated controls, developed to extract topographic, topologic and hydrologic information from digital spatial data of a hydrologic system, and to prepare ASCII files for the basin and precipitation components of HEC-HMS. These files, when opened by HEC-HMS, automatically create a topologically correct schematic network of sub-basins and reaches attributed with hydrologic parameters, and a protocol to relate gage to sub-basin precipitation time series. Starting from the DEM, CRWR-PrePro delineates the sub-basins and the reach network, calculates parameters for each hydrologic element, determines their interconnectivity, and prepares an input file for HEC-HMS that includes the computed hydrologic parameters. CRWR-PrePro also generates an input file for the precipitation component of HEC-HMS. Two methods to interpolate precipitation records are supported: one to calculate average precipitation at the sub-basins based on Thiessen polygons, and another to calculate the routing parameters of precipitation cells for hydrograph determination. Using CRWR-PrePro, the determination of the spatial parameters for HEC-HMS is a simple and automatic process that accelerates the setting up of a hydrologic model and leads to reproducible results.

Recent advances in Geographic Information Systems (GIS) technology offer hydrologic engineers, watershed managers, and data collection agencies unprecedented capabilities for storing and manipulating large quantities of detailed, spatially-distributed, watershed data. However, there is a shortfall of widely accepted techniques to take full advantage of hydrologic data in GIS format for actual watershed analysis. The ASCE Task Committee (TC) on GIS Modules and Distributed Models of the Watershed was established to identify existing data sources and techniques for watershed analysis within a GIS framework. The TC also provided a forum for the presentation of recent advances in this field at national ASCE conferences. An international mail survey was conducted to identify recent developments in GIS modules and distributed hydrologic models. This report presents state-of-the-art integrated GIS hydrologic analysis software and techniques, along with an overview and discussion of GIS data types and map projections. Data commonly required for hydrologic analysis using GIS techniques and available sources are listed; limitations of available data are discussed; and the integration of watershed hydrological analysis software and GIS techniques are presented.

A method is proposed for routing spatially distributed excess precipitation over a watershed to produce runoff at its outlet. The land surface is represented by a (raster) digital elevation model from which the stream network is derived. A routing response function is defined for each digital elevation model cell, so that water movement from cell to cell can be convolved to give a response function along a flow-path, and responses from all cells can be summed to give the outlet hydrograph. An example application of analysis of runoff on Waller Creek in Austin, Texas, is presented.

A significant part of the cost of most highway projects is attributable to drainage facilities such as storm drains, highway culverts, bridges, and water quality and quantity control structures. In this paper, we present a geographic information system (GIS) for hydrologic data development for design of drainage facilities, developed to reduce the analysis time and improve its accuracy by integrating spatial data describing the watershed, with hydrologic theory. A grid-based GIS to estimate potential extreme peak discharges, and watershed parameters, peak discharges for different frequencies, isochrone lines and runoff curve numbers is presented, using data from the State of Texas as an example application.

A unit hydrograph model is proposed in which the watershed is decomposed into subareas which are individual cells or zones of neighboring cells. The unit hydrograph is found for each subarea and the response at the outlet to excess rainfall on each subarea is summed to produce the watershed runoff hydrograph. The cell to cell flow path to the watershed outlet is determined from a digital elevation model. A constant flow velocity is assigned to each cell and the time lag between subarea input and response at the watershed outlet is found by integrating the flow time along the path from the subarea to the outlet. The response function for a subarea is modeled as a lagged linear reservoir in which the flow time is equal to the sum of a time of translation and an average residence time in the reservoir. It is shown that the assumption of a spatially-varying, but time-invariant, velocity field underlying this model produces a linear system model for all subareas whose outputs can be summed in the manner indicated. An example application is presented for the 8.70-Km2 Severn watershed at Plynlimon in Wales using a 50-meter digital elevation model in which the cell velocity is calculated by modifying an average velocity according to the terrain slope and the drainage area of each cell. The resulting model reasonably reproduces the observed unit hydrograph.