ABSTRACT:The focus of this research was to determine the extent of spatio-temporal changes in precipitation patterns using methods that can be regionalized and applied to various water resources such as lakes, streams, reservoirs or other water bodies. Many water resources throughout the world have demonstrated changes in historic water levels. As precipitation is the major input to these water bodies, it is important to investigate any changes in rainfall in both time and space. To investigate these patterns regionally, appropriate distributions, either gamma or generalized extreme value (GEV), were fitted to variables at a number of rainfall gauges within 40 km of a particular lake using maximum likelihood estimation. The variables included total annual rainfall, rainfall days per year, maximum annual event rainfall and maximum annual interevent days. Once the distribution parameters were estimated, an average, representative, distribution for each variable with 99% confidence limits was developed to determine if the fits for all gauges and variables were contained by these limits. To examine the temporal variation, distribution parameters were allowed to vary with time and then compared to constant parameter fits via likelihood ratio tests to determine if the varying parameter model significantly improved the fit. The spatial distribution of rainfall variables was found to be quite homogenous within the given confidence limits with some exceptions at high and low percentiles. Furthermore, the temporal distribution of rainfall variables was found to be stationary with only one gauge showing a significant trend. The average distribution for each precipitation variable developed can be applied at water resources within the region without rainfall gauges in proximity due to the spatial homogeneity and temporal stationarity identified. The methodology can be adapted to water resource management and planning in other regions.

Understanding changes induced by watershed urbanization is integral to developing an effective long-term management strategy. In this research, the authors study statistical changes of lake water surface levels in two urbanizing watersheds by evaluating serial change in time series parameters, autocorrelation, and variance as well as by developing a regression model to estimate weekly lake level fluctuations. The authors fit a seasonal integrated autoregressive moving average model to lake levels over subperiods of the data record to identify trends in parameter values. The authors fit the regression model with rainfall, lake stage, and temperature components for pre-urbanized and urbanized time periods to identify changes in baseflow. The lakes were located in Pasco County, Florida, USA and have not been significantly influenced by changes in rainfall patterns, pumping, surface water extraction or physical modification. Furthermore, the lakes exhibit consistent watershed urbanization and have sufficiently long and complete records. Based upon the research, the authors reach the following general conclusions about lakes in urbanizing watersheds: (1) the statistical structure of lake level time series is systematically altered, (2) in the absence of other forcing mechanisms, autocorrelation and baseflow decrease, (3) the presence of wetlands adjacent to lakes can offset the reduction in baseflow. These conclusions can be applied globally to similar regions that consist of lakes undergoing urbanization in flat, humid, shallow water table environments with wetlands. Furthermore, the methodology utilized can be applied at lakes in both similar and dissimilar environments to those studied in this research.

One of the most important tools in water management is the accurate forecast of long-term and short-term extreme values for flood and drought conditions. Traditional methods of trend detection are not suited for hydrologic systems while traditional methods of predicting extreme frequencies may be highly inaccurate in lakes. Traditional frequency estimates assume independence from trend or initial stage. However, due to autocorrelation of lake levels, initial stage can greatly influence the severity of an event. This research utilizes the generalized extreme value (GEV) distribution with time and starting stage covariates to more accurately identify trend direction and magnitude and provide improved predictions of flood and drought stages. Traditional methods of predicting flood or drought stages significantly overpredict or underpredict stages depending on the initial stage. Prediction differences can exceed one meter, a substantial amount in regions with flat topography; these differences could result in significant alterations in evacuation plans or other management decisions such as how much lake water to release in preparation for an approaching hurricane, appropriate lake levels to maintain, minimum structure floor elevations and more accurate forecasting of future water supply or impacts to tourism. The methods utilized in this research can be applied globally.