Hydraulic effects of changes in bottom land vegetation on three major floods, Gila River, in southeastern Arizona

Abstract

Changes in bottom-land vegetation between December 1965 and October 1972 apparently caused significant differences in stage, mean cross-sectional velocity, mean cross-sectional depth, and boundary roughness at peak discharges of three major floods in an 11.5-mile (18.5 km) study reach of the Gila River. The first flood, which had a peak flow of 39,000 ft3/s (1,100 m3/s), occurred in December 1965 when the dense bottom-land vegetation was dormant. The second flood, which had a peak discharge of 40,000 ft3/s (1,130 m3s), occurred in August 1967 when the vegetation had large amounts of foliage; however, the vegetation had been eradicated in the upstream half of the study reach prior to this flood. The third flood, which had a peak discharge of 80,000 ft 3/s (2,270 m3/s), occurred in October 1972; the vegetation had been eradicated in the whole study reach prior to this flood. Compared to the 1965 flood, the large amounts of foliage in the uncleared half of the reach during the 1967 flood apparently caused a 7 percent decrease in mean velocity, a 6 percent increase in mean depth, and an 11 percent increase in the Manning roughness coefficient at peak stage. Compared to the 1965 flood the clearing of the study reach apparently caused a 25 percent increase in mean velocity, a 15 percent decrease in mean depth, and a 30 percent decrease in the Manning roughness coefficient at peak stage in the 1967 and 1972 floods. The mean velocities of the three peak flows were relatively low where large parts of the flows moved across the meandering stream channel; the Manning coefficients and the mean depths were relatively large in these segments. After the first flood, scour was noted at seven of the nine cross sections in the study reach. After the second flood, fill was observed at all the cross sections, and, after the third flood, scour was observed at six sections. From 1964 to 1972, there was a net scour at .only one section, section 7, where the mean crosssectional velocity was relatively large for the three floods. Effects of changes of bottom-land vegetation on scour and (or) fill could not be determined. INTRODUCTION Saltcedar (Tamarix chinensis Lour1) has created problems along many streams in the arid and semiarid regions of the United States. Since about 1930 the plant has spread rapidly, consumed large amounts of water, and, in many streams, created potential flood hazards (Robinson, 1965, p. 1). The problems intensify as the demand for water mounts, the need for reducing flood 1Also referred to as Tamarix pentandra and Tamarix gallica. hazards grows, and at the same time the areal extent and density of the plant increases. Management of the saltcedar is necessary to lessen the magnitudes of the problems. As a remedial measure saltcedar has been eradicated along several streams in the western United States. The effectiveness and the side effects of this measure are not well documented. The flood plain of the Gila River in southeastern Arizona is an area where the vegetation has been managed. The low-benefit, deep-rooted vegetation, mostly saltcedar (Tamarix chinensis Lour) and mesquite (Prosopis juliflora var. velutine (Woot.) Sarg.), was replaced with a beneficial short-rooted grass (Culler, 1965, p.33-38). The saltcedar and mesquite trees are known to increase both the resistance to flow and the stability of the flood-plain boundary. Therefore, replacement of these trees with grass is likely to cause changes in rates of erosion and deposition, and to cause changes in channel width, depth, sinuosity, gradient, roughness, and even channel location. The main purpose of this report is to describe the apparent differences in hydraulic characteristics of the Gila River during three major floods owing to changes in bottom-land vegetation. The types of change in vegetation relevant to this study are seasonal increase in foliage and plant eradication. The hydraulic parameters studied are stage, mean cross-sectional velocity, mean cross-sectional depth, and the Manning roughness coefficient at peak discharge. Changes in the mean altitude of the bottom land as a result of the floods also are described. The floods occurred in December 1965, August 1967, and October 1972, with peak discharges of 39,000, 40,000, and 80,000 ft3/s (1,100,1,130, and 2,270 m3/s). These floods have a return interval of about 17 and 50 years, and they were the largest in the study reach since 1917 (Burkham, 1970, figs. 16 and 23). Discussions, descriptions, methods, and analyses presented in this report deal with averages, lumped