The utilization of tree-ring data to predict hydrologic properties of semi-arid watersheds near Tucson, Arizona.

Abstract

Four Douglas-fir chronologies were collected from the surrounding mountains of the Tucson Basin. These chronologies were checked first for their statistical characteristics such as standard deviation, mean sensitivity, and autocorrelation. Then, response functions were executed to discern the relationships of trees to the monthly average temperature and monthly total precipitation. The trees growing at the lower elevation of Santa Rita Mountains were found to best explain the variance of Tucson climate. Principal component analysis was used in this study as it has the advantage of transforming the data into uncorrelated variables. Tree-ring indices were first analyzed by this procedure and the outputs subjected to a stepwise multiple regression analysis with the runoff records of Sabino Creek Watershed, Rincon Creek Watershed, and Rillito Creek Watershed. The reconstructed runoff data were accomplished for a period of 313 years. The runoff variance explained by these tree-ring models were 35% for Sabino Creek, 64% for Rincon Creek, and 34% for Rillito Creek. In general the prediction of runoff data using tree-ring indices is considered adequate for the Tucson Basin.