Browsing by Subject "Data Mining"

Now showing 1 - 1 of 1

Texas Groundwater Conservation District policy : content mining and statistical analysis
(2019-05-22) Kase, Sydney Jean; Olmstead, Sheila M.
Groundwater is an increasingly significant and precious commodity within the state of Texas. The only statewide regulatory vehicle for governance and management of the groundwater resources are the Groundwater Conservation Districts (GCDs). A comprehensive statewide planning process was established by two senate bills in 1997 and 2001 which set forth the required actions for districts to manage and conserve the groundwater resources within the State of Texas. The bills require that all water conservation districts (including groundwater conservation districts, underwater conservation districts and subsidence districts) develop a management plan and update it at regular intervals. The management plans include a full accounting of the district’s water demands and the water supplies, the resultant water need (shortage or surplus) within each district as well as the rules of the district. Each district’s management plans are also required to establish a set of goals that the district will use to manage its water resources in order to meet its reported shortage or maintain a surplus water budget. GCDs are mandated to produce management plans during their initiation, as well as periodic updates over time. In order to understand if the current management plan structure is working, I used content mining to turn the management plans into a dataset and then ran a series of statistical models to describe impacts. This research outlines a method of quantitative analysis to understand the relationship between groundwater management plans and groundwater resources that utilizes current and historic GCD management plans, and a water supply need metric developed by the Texas Water Development Board (TWDB). Statistical classification techniques were employed to evaluate the association between the management plans and the water supply class of each GCD. The statistical learning methods returned between 75% and 90% correct classifications depending on the model. The most impactful predictors when determining class were found to be shortage, recharge and groundwater when classifying as a surplus and precipitation, demand and aquifer when classifying a shortage.