Thesis (M. S.)--University of Hawaii at Manoa, 1995. / Includes bibliographical references (leaves 95-98). / UHM: Has both book and microform. / U.S. Geological Survey; project no.03; 14-08-0001-G2015

Thesis (M.S.)--Auburn University, 2007. / Abstract. Vita. Includes bibliographic references (ℓ. 111-121)

Thesis (M.S.)--University of Texas at Austin, 1999. / Title from title screen. Includes bibliographical references. Also available in paper.

Thesis (M.S.)--Oregon State University, 1994. / Typescript (photocopy). Includes bibliographical references (leaves 103-119). Also available on the World Wide Web.

Thesis (M.S.)--California Polytechnic State University, 2010. / Title from PDF title page; viewed on June 10, 2010. Major professor: Lynn E. Moody, Ph.D. "Presented to the faculty of California Polytechnic State University, San Luis Obispo." "In partial fulfillment of the requirements for the degree [of] Master of Science in Agriculture, with a Specialization in: Soil Science." "April 2010." Includes bibliographical references (p. 48-51).

Philosophiae Doctor - PhD / The most significant human impacts on the hydrological system are due to land-use change. The conversion of land to agricultural, mining, industrial, or residential uses significantly alters the hydrological characteristics of the land surface and modifies pathways and rates of water flow. If this occurs over large or critical areas of a catchment, it can have significant short and long-term impacts, on the quality of water. While there are methods available to quantify the pollutants in surface water, methods of linking non-point source pollution to water quality at catchment scale are lacking. Therefore, the research presented in this thesis investigated modelling techniques to estimate the effect of land-cover type on water quality. The main goal of the study was to contribute towards improving the understanding of how different landcovers in an urbanizing catchment affect surface water quality. The aim of the research presented in this thesis was to explain how the quality of surface runoff varies on different land-cover types and to provide guidelines for minimizing water pollution that may be occurring in the Kuils-Eerste River catchment. The research objectives were; (1) to establish types and spatial distribution of land-cover types within the Kuils-Eerste River catchment, (2) to establish water quality characteristics of surface runoff from specific land-cover types at the experimental plot level, (3) to establish the contribution of each land-cover type to pollutant loads at the catchment scale. Land-cover characteristics and water quality were investigated using GIS and Remote Sensing tools. The application of these tools resulted in the development of a landcover map with 36 land classifications covering the whole catchment. Land-cover in the catchment is predominantly agricultural with vineyards and grassland covering the northern section of the catchment. Vineyards occupy over 35% of the total area followed by fynbos (indigenous vegetation) (12.5 %), open hard rock area (5.8 %), riparian forest (5.2 %), mountain forest (5 %), dense scrub (4.4 %), and improved grassland (3.6 %). The residential area covers about 14 %. Roads cover 3.4 % of the total area. Surface runoff is responsible for the transportation of large quantities of pollutants that affect the quality of water in the Kuils-Eerste River catchment. The different land-cover types and the distribution and concentration levels of the pollutants are not uniform. Experimental work was conducted at plot scale to understand whether landcover types differed in their contributions to the concentration of water quality attributes emerging from them. Four plots each with a length of 10 m to 12 m and 5 m width were set up. Plot I was set up on open grassland, Plot II represented the neyards, Plot III covered the mountain forests, and Plot IV represented the fynbos landcover. Soil samples analyzed from the experimental plots fell in the category of sandy soil (Sa) with the top layer of Plot IV (fynbos) having loamy sand (LmSa). The soil particle sizes range between fine sand (59.1 % and 78.9 %) to coarse sand (between 7 % and 22 %). The content of clay and silt was between 0.2 % and 2.4 %. Medium sand was between 10.7 % and 17.6 %. In terms of vertical distribution of the particle sizes, a general decrease with respect to the size of particles was noted from the top layer (15 cm) to the bottom layer (30 cm) for all categories of the particle sizes. There was variation in particle size with depth and location within the experimental plots. Two primary methods of collecting water samples were used; grab sampling and composite sampling. The quality of water as represented by the samples collected during storm events during the rainfall season of 2006 and 2007 was used to establish water quality characteristics for the different land-cover types. The concentration of total average suspended solids was highest in the following land-cover types, cemeteries (5.06 mg L-1), arterial roads/main roads (3.94 mg L-1), low density residential informal squatter camps (3.21 mg L-1) and medium density residential informal townships (3.21 mg L-1). Chloride concentrations were high on the following land-cover types, recreation grass/ golf course (2.61 mg L-1), open area/barren land (1.59 mg L- 1), and improved grassland/vegetation crop (1.57 mg L-1). The event mean concentration (EMC) values for NO3-N were high on commercial mercantile (6 mg L-1) and water channel (5 mg L-1). The total phosphorus concentration mean values recorded high values on improved grassland/vegetation crop (3.78 mg L-1), medium density residential informal townships (3mgL-1) and low density residential informal squatter camps (3 mg L-1). Surface runoff may also contribute soil particles into rivers during rainfall events, particularly from areas of disturbed soil, for example areas where market gardening is taking place. The study found that different land cover types contributed differently to nonpoint source pollution. GIS model was used to estimate the diffuse pollution of five pollutants (chloride, phosphorus, TSS, nitrogen and NO3-N) in response to land cover variation using water quality data. The GIS model linked land cover information to diffuse nutrient signatures in response to surface runoff using the Curve Number method and EMC data were developed. Two models (RINSPE and N-SPECT) were used to estimate nonpoint source pollution using various GIS databases. The outputs from the GIS-based model were compared with recommended water quality standards. It was found that the RINSPE model gave accurate results in cases where NPS pollution dominate the total pollutant inputs over a given land cover type. However, the N-SPECT model simulations were too uncertain in cases where there were large numbers of land cover types with diverse NPS pollution load. All land-cover types with concentration values above the recommended national water quality standard were considered as areas that needed measures to mitigate the adverse effects of nonpoint pollution. The expansion of urban areas and agricultural land has a direct effect on land cover types within the catchment. The land cover changes have adverse effect which has a potential to contribute to pollution.

The Agricultural Nonpoint Source model (AGNPS) is a single-event grid-based model used for simulating runoff, sediment and nutrients from agricultural areas. This study involved using geographic information system (GIS) spatial data and functionality to improve the spatial and temporal assignment of parameter values for the AGNPS 5.0 model and incorporated methods for representing urban fringe land uses and their nonpoint source (NPS) pollution contributions in model inputs. Auxiliary procedures for modeling with AGNPS were developed both for enhancing input into the model and for enhancing modeled output. On an event basis, one procedure automated the creation of complex-formatted AGNPS 5.0 model input files using GIS as a spatial data manager. One pair of alternative procedures were developed to automate the assignment of parameter values on an event basis. One procedure used typical average annual parameter values, and the second assigned parameter values using adaptations of existing time-dependent relationships. On a monthly basis, a sequencing procedure was created to perform multiple runs with the model for a list of storms while updating parameters for each event and aggregating monthly modeled spatial output. Another pair of alternative procedures were developed to facilitate the simulation of monthly output from AGNPS modeled events. The first of these aggregated event output for all storms in each month, while the second supplemented the aggregated output with baseflow and septic system loads. The study area was the 6,500 ha urbanizing Bull Run watershed in northern Virginia, which was modeled as 14,621 cells. Databases were assembled and 109 selected storm events within a 16-year period were modeled using the above procedures. Event data were added together, where necessary, to correspond with observed data from composite-sampled intervals. Output from the two event parameterization procedures were compared with monitored loads calculated for 89 composite periods, while output from the two monthly simulation procedures were compared with monthly monitored data for 23 complete months. The monitored-modeled comparisons were considered inconclusive. Evidence strongly suggested that the rainfall records from a rain gauge outside the watershed did not correspond well with monitored runoff. The average runoff produced with the AGNPS model from the 109 selected storms amounted to 40.7% of rainfall, consistent with the calculated long-term average of 38% for the Bull Run watershed. A nonpoint source pollution index was developed to utilize monthly modeled total nitrogen, total phosphorus, and suspended sediment. Individual rating curves were developed to separately transform loads and concentrations of each pollutant into sub-index values. The maximum sub-index from each parameter was added together and averaged for the index. The index was calculated at the watershed outlet from monitored data, and in a spatially-distributed fashion along all streams from simulated output. / Ph. D.

Quantification of total phosphorus (TP) loads entering a lake or reservoir is important because phosphorus is most often the limiting nutrient in terms of algae growth, thus phosphorus can control the extent of eutrophication. Four methods for assessing the annual tributary phosphorus loads to two different Virginia reservoirs were analyzed, three methods that use tributary monitoring program data and one that uses land-use and rainfall data. In this project, one tributary has been extensively monitored for many years and served as a control on which the other methods were tested. The key difference between this research and previous studies is the inclusion of flow-weighted composite storm sampling instead of simple grab sample analyses of storm flow. Three of the methods employed flow stratification, and the impact of the base flow separation point was examined. It was found that the Regression Method developed in this research was the least sensitive to the base flow separation point, which is a valuable attribute because a wrong choice will not significantly affect the estimate. The Monte Carlo Method was found to underestimate the TP loads. The amount of rainfall impacted the accuracy of the methods, with more error occurring in a year with lower precipitation. / Master of Science

Modeling is a very important tool for developing nonpoint source (NPS) pollution control plans. Current NPS models were developed for temperate conditions and, thus, do not appropriately represent tropical conditions. The objective of this research was to develop or adapt a nonpoint source pollution model to simulate transformations and losses of nitrogen (N) and phosphorus (P) in leachate and runoff from crop fields under humid tropical conditions. An extensive literature synthesis identified appropriate relationships for representing hydrologic and NPS processes in the tropics, as well as soil and climate conditions that differ from temperate conditions and impact NPS pollution. The GLEAMS model was selected for adaptation. Changes to the model included calculation of potential evapotranspiration (ET); changes in initial and default values of N and P pools, C:N ratio of soil organic matter, and soil P sorption; changes in simulation of transformations between N and P pools, along with the effect of temperature; and inclusion of a nitrate retardation factor (Ncrit) and pH in the calculation of N transformation and movement. The adapted model, called TROPGLEAMS, was evaluated through model verification, application, and sensitivity analysis. Model verification comprised a mass balance of nutrients and analysis of the variation of variable values in time. Model validation included application of the GLEAMS and TROPGLEAMS to three sites in Brazil: a set of lysimeters planted with sugarcane in Piracicaba, SP; a set of plots planted with sugarcane in Piracicaba; and a set of plots planted in a wheat-soybean rotation in Lages, SC. Model sensitivity to temperature, Ncrit and pH were evaluated in the sensitivity analysis. Model evaluation indicated that TROPGLEAMS is more accurate than GLEAMS in simulating fate and transport of nutrients under tropical conditions. Prediction of actual ET, effect of tillage on losses of N and P in runoff, and N and P kinetics was improved with TROPGLEAMS compared to GLEAMS. However, based on data from the Lages study, TROPGLEAMS did not simulate losses of nutrients in runoff well. Improvements in the model, especially related to losses in runoff, and application of TROPGLEAMS to different areas of the humid tropics are recommended. / Ph. D.

The objective of this study is to evaluate the costs of reducing pesticide, nitrogen, phosphorus, and sediment losses of a representative risk-neutral and risk-averse peanut-cotton farmer in Southeast Virginia. Five currently popular rotations and eight alternative conservation rotations are evaluated for the representative farm. The Erosion-Productivity Impact Calculator (EPIC) model is used to simulate pesticide, nitrogen, phosphorus, and soil loss from each rotation using actual rainfall and temperature data from the study area. A Target-MOTAD mathematical programming model, REPVAFARM, is developed and solved with GAMS. The objective of the farmer is to maximize expected net return, while meeting a target income with certain allowable expected shortfall from the income target. The farmer is also constrained by land, labor, peanut quota, and levels of pesticide, nitrogen, phosphorus, and soil losses. Major findings of this study are: reducing pesticide, nitrogen, phosphorus, and soil losses imposes costs to the farmer regardless of his risk attitude, with costs ranking from high to low in the order of reducing all pollutant losses, reducing nitrogen losses, reducing phosphorus losses, reducing soil losses, and reducing pesticide losses. Costs of reducing pollutant losses are higher for more risk-averse farmers than for less risk-averse and risk-neutral farmers implying that risk-aversion is an obstacle to the adoption of alternative conservation practices. Reducing pesticide losses has little impact on other pollutants. Reducing pesticide and nitrogen losses simultaneously achieves similar reductions in soil loss and phosphorus loss. / Master of Science