H44A-01 INVITED
Colorado River Droughts Inferred from Tree Rings
Tree-ring reconstructions for the Colorado River at Lee Ferry, Arizona, have consistently indicated that the modern gaged flow record is unrepresentative of the past few centuries both in the long-term mean annual flow and the severity of extended hydrologic droughts. Information on past droughts can be considered most reliable for the past 500 years, when tree-ring site coverage is relatively dense. Alternative reconstructions using different modeling methods and basic tree-ring data differ in details but are consistent in the timing of periods of relative drought and wetness, and in identifying the late 1500s as the most severe multi-year drought of the last 500 years. Reconstructions for major tributaries emphasize the spatial coherence of droughts and wet periods over the UCRB. The Lee Ferry record of annual flows can be extended back beyond 500 years, with diminished confidence, by analysis of tree rings from dead standing trees and logs. A new reconstruction covering the period A.D. 762-2005 was generated by a series of reconstruction models with time-varying subsets of 11 recently developed tree-ring chronologies from the Upper Colorado River Basin (UCRB) as predictors. The percentage of variance of annual flow at Lee Ferry accounted for by these models decreases from 77 percent after A. D. 1365 to 60 percent before A. D. 1182, as the number of available tree-ring chronologies drops from eleven to three. Correlation analysis and cross-spectral analysis support the hydrologic interpretation of annual as well as decadal fluctuations in the extended reconstruction. The singular multi-decadal event is a drought in the mid-1100s, when the lowest 25- year mean is estimated to be just 84 percent of the modern observed long-term mean. Absence of high flows rather than presence of record-low flows is characteristic of this drought. Results suggest the mid-1100s drought was more severe in the western part of the UCRB than the well-documented North American mega-drought of the late 1500s. Greater site coverage and sample depth (number of trees) are essential to reducing the uncertainty of assessments of hydrologic drought severity in the early part of the tree-ring record.
H44A-02
The Long and the Short of it: Millennial-Scale and Contemporary Sediment Yields of Eastern Grand Canyon
Dams on the Colorado River were constructed to regulate and store water for irrigation and domestic consumption, but flow regulation has had major impacts on hydrological and ecological processes within the watershed. Intrinsic to understanding these impacts requires a quantitative knowledge of the sediment yields in regulated reaches. Unfortunately, little is known about the long-term sediment yields, and contemporary data were either estimated or measured over only a few decades and only at only a few locations and therefore may not reflect the longer cycles of erosion and aggradation within the mainstem Colorado River or within the tributary basins. Here, we present data that compares contemporary sediment yield derived from models of ungaged tributaries in eastern Grand Canyon with millennial-scale sediment yield data derived from cosmogenic 10Be to highlight the sediment delivery processes over the past several thousand years. Models of the contemporary sediment yield of eastern Grand Canyon suggest 349 Mg km-2 y-1 of sediment. Cosmogenic 10Be estimates of long-term (104 yr) yields are 227 Mg ky-2 y-1, only 65% of the contemporary estimates. The difference in these estimates most likely reflects a change in process during the effective range of cosmogenic 10Be (>10 ky). Hanks and Webb (2006) suggest that the bed of the Colorado River in this reach may have aggraded approximately 10 m as a result of a high debris-flow frequency from tributary canyons potentially related to the Pleistocene-Holocene climate change. Conversely, the cosmogenic 10Be data average the high contemporary yields with potentially lower yields of the early- Holocene and late-Pleistocene. If the contemporary sediment yields are extrapolated for the entirety of the Holocene, then the cosmogenic data suggest much lower average sediment yields prior to the Pleistocene- Holocene transition. If correct, these data suggest that Pleistocene-Holocene climate change increased sediment yields several fold. At a minimum, the face value the contemporary sediment yields are 1.5 times the long-term average and suggest that that climate change has had a significant impact on the sediment budget of the eastern Grand Canyon and probably the Colorado River basin.
H44A-03
Floodplain lakes and alluviation cycles of the lower Colorado River
The broad valleys along the lower Colorado River contain numerous bodies of still water that provide critical habitat for bird, fish, and other species. This chain of floodplain lakes is an important part of the Pacific Flyway - the major north-south route of travel for migratory birds in the western Hemisphere - and is also used by many resident bird species. In addition, isolated floodplain lakes may provide the only viable habitat for endangered native fish such as the razorback sucker, vulnerable to predation by introduced species in the main stem of the Colorado River. Floodplain lakes typically occupy former channel courses of the river and formed as a result of river meandering or avulsion. Persistent fluvial sediment deposition (aggradation) creates conditions that favor rapid formation and destruction of floodplain lakes, while long term river downcutting (degradation) inhibits their formation and evolution. New radiocarbon dates from wood recovered from drill cores near Topock, AZ indicate that the river aggraded an average of 3 mm/yr in the middle and late Holocene. Aggradational conditions before Hoover Dam was built were associated with rapid channel shifting and frequent lake formation. Lakes had short life spans due to rapid infilling with fine-grained sediment during turbid floods on the unregulated Colorado River. The building of dams and of armored banks had a major impact on floodplain lakes, not only by drowning large portions of the valley beneath reservoirs, but by preventing new lake formation in some areas and accelerating it in others. GIS analyses of three sets of historical maps show that both the number and total area of isolated (i.e., not linked to the main channel by a surface water connection) lakes in the lower Colorado River valley increased between 1902 and the 1950s, and then decreased though the 1970s. River bed degradation below dams inhibits channel shifting and floodplain lake formation, and the capture of fines behind the dams has prevented sediment infilling of the lakes. Bed lowering below dams and in artificially confined reaches could potentially dewater floodplain lakes, a process occurring at Beal Lake, a natural lake used for native fish restoration in the Havasu National Wildlife Refuge. Sedimentation near the upstream ends of reservoirs has created large areas of still water. One of the largest, Topock Marsh, is connected to the main channel, restricting its usefulness as a native fish nursery; other backwater areas are confined by bars that isolate standing water at tributaries.
H44A-04
Dynamics of Bottomland Geomorphology and Vegetation Along a Dammed, Arid Region River: Implications for Streamflow Management
In arid and semiarid western North America, floodplain forests dominated by native cottonwood and willow trees are highly valued as wildlife habitat and preferred recreation sites and are thus the focus of conservation efforts. The Bill Williams River harbors some of the most extensive native floodplain forests in the lower Colorado River region. Our work is aimed at understanding the dynamics of the Bill Williams River floodplain forests, in the context of pre- and post-dam hydrology and geomorphology. We have mapped bottomland geomorphology and vegetation using seven sets of orthorectified aerial photographs spanning more than 50 years. Two sets of photos (1953 and 1964) pre-date the completion of Alamo Dam, a large flood control structure; and three sets of photos (1996, 2002, and 2005) are from an era during which streamflow downstream of the dam has been managed to promote the establishment and survival of native floodplain forest. Comparison of the aerial photographs to LiDAR data collected in 2005 is providing a framework for quantifying changes in valley bottom morphology and estimating reach-scale changes in volumes of stored and evacuated sediment between 1953 and 2005. Furthermore, comparison of the extent of pre-dam active channel in 1953 with the extent of floodwaters from a regulated moderate flood in 2005 provides an approximation of the predominant patterns of aggradation and degradation in the system over this interval of time. Flood magnitude on the Bill Williams has been dramatically reduced since the closure of Alamo Dam in 1968, and low flows have increased considerably since 1979. Channels along the Bill Williams R. narrowed an average of 111 m (71 %) between 1953 and 1987, with most narrowing occurring after dam closure. Multiple regression analysis revealed significant relationships among flood power, summer flows, intermittency (independent variables) and channel width (dependent variable). Concurrent with channel narrowing was an expansion of dense floodplain vegetation, consisting primarily of native cottonwood and willow and non-native tamarisk shrubs. Moderate flood releases (~7000 ft3/s) from Alamo Dam in the early 1990's widened the river channel and resulted in the establishment of new woody vegetation. For the following nine years, relatively steady, low discharges were released from the dam, resulting in channel narrowing, extensive beaver pond creation, and dense vegetation growth. Moderate flood releases in 2005 again widened channels, destroyed beaver ponds, and created conditions suitable for new vegetation establishment. In addition to understanding the specific conditions along the Bill Williams River, our work should contribute to a more general understanding of connections between fluvial processes and floodplain vegetation, in the contexts of geomorphic response downstream of a large dam and efforts to manage streamflow for ecological benefits downstream.
H44A-05
Holocene Cyclical Switching of Colorado River Water Alternatively to the Sea of Cortez or to the Salton Sink
The former giant lake (ancient Lake Cahuilla) that intermittently filled the Salton Sink with a volume half that of Lake Erie has profound implications for the hydrologic and ecologic history of the Colorado River delta. Because the delta dams and isolates the sink from the Sea of Cortez (Gulf of California), the delta cone has a rare geometry that drains distributaries toward two unconnected termini: sea level on the south side and a fluctuating level in the Salton Sink on the north side. This level fluctuated in the Holocene between 85 m below modern sea level when the Salton Sink was dry and 12 m above sea level when occupied by successive incarnations of full Lake Cahuilla. Geologic and archaeologic records indicate that over the last 1300 years the Salton Sink cycled several times between dry and recurrently holding this 97 m-deep lake. At about 12 m above sea level the lake spilled southward across the delta to the Sea of Cortez most recently in the late 1600s or early 1700s. A simple model based on delta gradient can explain cyclical switching of the river from one side of the delta to the other. When Lake Cahuilla was dry or low, any northward flows off the delta would encounter greater gravitational potential and a steeper gradient compared to the south side. The floods of 1905-1906 dramatically demonstrated that flows down the steep north flank of the delta could cause channel entrenchment, in this case headward retreat of a waterfall 9 m high as fast as 30 cm/min. Except for human intervention, this rapid downcutting would have led to complete capture of Colorado River water until Lake Cahuilla filled. Similar entrenchment and capture events must have recurred many times during the Holocene and also earlier times. We infer that when Lake Cahuilla rose to its spillover level, the feeding distributaries silted in and lowered their grade enough to provide an impetus for the river to switch back to paths down the south side of the delta to the Sea of Cortez. Shut off from inflow, evaporation of 1.8 m/yr would dry Lake Cahuilla in a few decades, again lowering the base level below sea level and setting the stage for another cycle of northward diversion, downcutting, lake filling, and spillover. Fluctuating Lake Cahuilla must have profoundly impacted the ecology of the delta. Bones of razorback sucker and bonytail at shoreline archaeologic sites imply that when full, the lake served as a huge Holocene temporary fish incubator for these two now-endangered species endemic to the Colorado River. In contrast, the high evaporation rate of Lake Cahuilla and nearly 2 decades of average Colorado River discharge needed to fill the lake to 12 m above sea level would have deprived the south half of the delta of river water for long periods.
H44A-06
An Integrated Model for Evaluating Hydrology, Hydrodynamics, Water Quality and Ecology in a Coastal Desert Wetland
An integrated model describing hydrology, hydrodynamics, salt dynamics and vegetation was developed to predict the evolution of the Cienega de Santa Clara ("Cienega"), a non-tidal wetland located in the Colorado River Delta. The Cienega was created in 1977 when water intended for treatment at the Yuma Desalting Plant in Arizona was bypassed to a salt flat in the Delta. The continued delivery of this water is uncertain and thus, this model was developed to predict the effects of changes in the quantity and quality of inflow to the wetland over seasonal and annual timescales. The model is divided into four modules that run in sequence for each timestep, in which the results from one module are used to produce results in successive modules. The four modules are: (1) evapotranspiration, (2) water balance, (3) mixing/salt balance, and (4) vegetation response. Over the calibration period, 1993-2002, modeled results of wetland surface area, the fraction of the wetland covered in vegetation and salinity concentrations compare well to actual data. The model was used to run nine hypothetical scenarios, representing the range of inflow quantity and quality to the Cienega that could occur if the source of the inflow is altered, including the possible re-opening of the Yuma Desalting plant. Model results show that the Cienega ecosystem is more sensitive to changes in salinity than to changes in flow. However, in almost all cases, an increase in salinity and/or a decrease in flow would cause a significant decrease in wetland size and vegetation cover, compromising a large portion of the habitat currently available to wildlife at the Cienega.
H44A-07
WATER BODIES AND VEGETATION IN THE CALIFORNIA-BAJA CALIFORNIA BORDER REGION A REMOTE SENSORS PERSPECTIVE.
The California-Baja California border region although they share watersheds, similar climate and landscape, there is a big contrast in the vegetation cover and water bodies between the two countries as seen from remote sensors. There is a stronger signature of vegetation and larger number of water bodies in the California side. To do a quantitative estimate of these differences, a comparative analysis of vegetation and water bodies was perfomerd along a strip of 100 km from both sides of the border with remote sensing techniques using Landsat TM images from 1984 to 2006. The strong absorption of water to short wave infrared radiation captured by band 5 of TM Landsat sensor (1.55- 1.75 micrometers) is use to detect water bodies. The histogram segmentation technique was used with TM 5/1 band ratios reinforced with a shades prediction technique using the sun position and a digital elevation model. The aerial extent of detected water bodies is estimated. Also an analysis from 1972 trough 2002 of the Mexican portion of Colorado river delta will be presented, with emphasis on flood events induced by abnormal snowmelts and higher precipitations in the high basin; 250 Landsat image previews were collected , from which 157 were selected to integrate 63 scenes that provide a dynamic picture of the Colorado delta river over 30 years. A regression with the annual averages of inundated areas and annual water flow data from E.U. to Mexico was made with a correlation coefficient of 0.912. The normalized difference vegetation index (NDVI) was used to estimate the vegetation greenness in the agricultural valleys and in natural vegetated areas along the mountains on both sides of the border. The spatial distribution of the NDVI and the differences between zones with the same land use regime on both sides of the border is presented.
H44A-08
A Water Budget for Riparian Vegetation on the Lower Colorado River: the Myth of Water Salvage
For many years, river managers have envisaged large saving of water by clearing the exotic plant, saltcedar (Tamarix ramosissima) from western U.S. rivers. Early estimates of evapotranspiration (ET) by saltcedar ranged as high as 3-4 m/yr, and it was estimated that saltcedar on the Lower Colorado River used more water than Los Angeles. Furthermore, saltcedar was considered to have low habitat value, so clearing projects might enhance habitat value by allowing the return of more valuable native species. We have examined these assumptions based on recent evidence. Moisture flux towers set in dense saltcedar stands show that ET is moderate, ranging from 0.8-1.4 m/yr with a mean value of 1 m/yr over five studies on three rivers, similar to wide-area estimates from remote sensing studies. Projected over the 18,200 ha of dense saltcedar monocultures estimated for the Lower Colorado River riparian corridor in the U.S., the potential water saving would only be about 1 percent of the annual flow (assuming no replacement vegetation). A similar acreage of saltcedar monoculture exists in the Colorado River delta in Mexico, but these stands are supported by outflow of brackish water from the irrigation district rather than river water. The assumption of low habitat value is not supported by recent studies. For example, Hinojosa- Huerta (2006) found that saltcedar monocultures away from the river channel supported 65 percent as many bird numbers and 74 percent as many bird species as the best habitat type, mixed saltcedar and native trees in proximity to water, in the delta of the Colorado River in Mexico, and saltcedar provided equal habitat value as native trees for endangered willow flycatchers on Arizona and New Mexico rivers (Owen et al., 2005). Hence, the prospects for saving water without destroying habitat by clearing saltcedar are doubtful for this river system.