Changes in Heavy and Extreme Precipitation in Texas 1900-2015

Heavy and extreme precipitation presents challenges to urban planners, wastewater engineers, and emergency managers. The state of Texas depends on regular and predictable precipitation for agriculture, industry, and the needs of the growing population. Changes in the hydrologic cycle due to a warming climate present new challenges like drought and flash flooding that have significant economic and human costs. A study was undertaken to document long-term changes in the distribution of heavy and extreme precipitation in Texas. Data were obtained from the National Climatic Data Center (NCDC) archive for fourteen cities in all parts of the state. Records were included if a continuous record of daily precipitation totals was available beginning in at least 1900. A peak-over-threshold method was employed for the extreme value analysis. In this method, all of the daily precipitation totals above a given threshold were included, in contrast to the traditional method where only the highest value in each analysis period was included. This method is favorable for precipitation data because multiple extreme events often occur in a given season or one-year period. Data exceeding the threshold were then fit to a Generalized Pareto Distribution and the distribution parameters used to create the return period curve. In addition to return period curves, non-zero precipitation data were fit to a Weibull distribution to assess changes in the distribution of daily precipitation totals at a broad range of accumulations. A set of each type of curve was created for the thirty year period at the beginning and the thirty year period at the end of the record. Curves were created for both the yearly data and for each season. Preliminary results indicate that most stations in Texas experienced a change in return period curve between the beginning and end of the twentieth century. Alpine, Brownsville, College Station, El Paso, Houston, and San Marcos all had an increase in return period curve over this period, indicating a higher magnitude of rain event during the latter part of the century. Amarillo, Austin, Blanco, and Crosbyton all showed a decrease in return period curve for the late period. Further exploration of these century-scale changes including attribution and spatial influences will be evaluated.