GEOLOGIC RECORD OF FLOOD DYNAMICS ON THE RED RIVER OF THE NORTH, USA-CANADA

Dr. William D. Gosnold

Dr. Richard D. LeFever

Department of Geology and Geological Engineering

University of North Dakota

Grand Forks, ND

Dr. Stephen K. Boss

Department of Geology

University of Arkansas

Fayetteville, AR 72701

PROJECT SUMMARY

We propose a comprehensive, multidisciplinary, multi-institution program to investigate the geological record of flooding along the Red River of the North with the ultimate goal of resolving both flood frequency and flood magnitude along this river system. We will correlate existing data on stream discharge with preserved sediment layers corresponding to known floods. These correlations will serve as a basis for interpreting the geologic record of Red River flooding. The project will include two field seasons during which we will conduct detailed surveys of deltaic sediments of the Red River of the North in Lake Winnipeg and in selected oxbow lakes along the Red River flood plain in Canada and the United States using high-resolution seismic reflection and real-time global positioning system equipment. Following analysis of the seismic reflection data, an intensive program of coring will focus on recovering flood deposits from surveyed lakes. Results of this research will provide empirical data useful for defining future flood planning and flood mitigation strategies by providing better estimates of flood frequency and flood magnitude along this river system.

INTRODUCTION

The 1997 spring flood of the Red River of the North reached 16.9 m (55.6 ft) at USGS gaging station 05082500 at Grand Forks, ND-East Grand Forks, MN. This flood level is the highest ever recorded at the gaging station (1882-1997), and the flood waters devastated virtually all of East Grand Forks, Minnesota (six homes were not damaged) and approximately 80 percent of Grand Forks, North Dakota. Newspaper reports indicate that several earlier floods, in particular the flood of 1826, may have reached 3.2 m (10 ft) higher than the 1997 flood crest at Grand Forks-East Grand Forks.

Assessing the risk and mitigating the effects of future floods along the Red River of the North requires understanding both the maximum possible flood stage and estimating the probable frequencies of occurrence of various flood levels. However, the conventional method of flood frequency analysis appears inadequate for several reasons: 1) flood frequency analysis using probability theory assumes statistical homogeneity and independence, 2) flood frequency and magnitude vary in response to decadal- to century-scale climate (changing precipitation and temperature patterns, 3) flood frequency and magnitude vary in response to land use patterns over time, introducing an important human component to flood dynamics. We propose a comprehensive, multidisciplinary, multi-institution program to investigate the geological record of flooding (i.e., the paleoflood record) along the Red River of the North with the ultimate goal of resolving both flood frequency and flood magnitude (i.e., flood dynamics) along this river system. Results of this research will provide empirical data useful for defining future flood planning and flood mitigation strategies.

DESCRIPTION OF RESEARCH

To accomplish this, we will correlate existing data on stream discharge with preserved sediment layers corresponding to known floods. These correlations will serve as a basis for interpreting the geologic (i.e. prehistoric) record of Red River flooding. The project will include two field seasons during which we will conduct detailed surveys of deltaic sediments of the Red River of the North in Lake Winnipeg and in selected oxbow lakes along the Red River flood plain in Canada and the United States using high-resolution seismic reflection and real-time global positioning system equipment. Following analysis of the seismic reflection data, an intensive program of coring will focus on recovering flood deposits from surveyed lakes.

The basis for this research approach is that sediment properties such as layer thickness, organic content, particle size and distribution should correlate with several flood parameters such as flood stage, stream discharge and flood duration. The geophysical surveys will help identify those areas where flood sediments are most likely to be preserved as well as providing general information on the gross stratigraphy of deltaic and lacustrine deposits. Detailed analyses of cored deposits will permit sedimentological characterization of flood layers and provide samples for isotopic correlation of these layers from different lakes along the Red River floodplain.

At its terminus, the Red River of the North discharges into southern Lake Winnipeg and deposits sediment in a delta along the southern lake shore. These deposits have been accumulating continuously over the last 4,500 years (Last and Teller, 1983) and thus they contain the geological record of discharge events for the Red River and its tributaries. Other deposits important to this study occur in oxbow lakes along the river floodplain. These small lakes have little sediment input other than the Red River in flood stage, thus their stratigraphic record is especially relevant to the history of flooding and in distinguishing floods on the Red River and major tributaries such as the Assiniboine River. Sediment load in the Red River is considered unusually high for a low-gradient stream (Bluemle, 1997). Typical sediment loads at Grand Forks-East Grand Forks during the warm season are of the order of 1470 metric tons per day and during peak flows (e.g., major floods) sediment loads of 30,844 metric tons per day have been recorded (Bluemle, 1997).

The reasons that conventional methods for estimating flood recurrence interval are inadequate are that frequency analysis of hydrologic data entails the specific assumptions that the data are homogeneous and independent. That is, all observations are from the same population and the same event is not recorded more than once. In fact, the data currently in use are not from the same populations (Figs. 1,2). Two factors are at work. First, major floods occur during the spring snow melt, although lesser floods occur following unusually heavy summer precipitation events. These two flood seasons are different populations and during years of below normal snow fall, flood data from summer precipitation are used in the frequency analysis. This is not a rare occurrence. In 1991, 1993 and 1994, the maximum discharge recorded at the East Grand Forks gaging station occurred in July or August rather than in the spring. A simple analysis of the daily discharge data at the East Grand Forks gaging station can eliminate the summer flood data from the record.

Fig. 1. Combined spring/summer flood data for the Red River of the North (1904-1996). Fig. 2. Spring flood data (red) and summer flood data (green) for the Red River of the North (1904-1996).

The second factor is climate change. In the Red River Valley, average annual ground-surface temperature has increased 2.5 C and the average annual air temperature has increased 2 C during the past century (Gosnold et al., 1997). Precipitation dropped from 1890 until the mid-1930s and has been rising since 1950 (Groisman and Easterling, 1994). The effects of changing climate can be seen in the flood record since 1882 which suggest close coupling of flood dynamics and climate parameters (especially precipitation). For example, within the historical record, it is apparent that between 1882 and 1950, a severe flood (gage level >12.2 m) occurred approximately once every 6 years at Grand Forks-East Grand Forks. This corresponds to the interval of generally diminishing precipitation from 1890 until the mid-1930’s (Groisman and Easterling, 1994). Since 1950, however, Grand Forks has experienced a severe flood once every 3 years with an increasing trend toward greater frequency of large-scale flooding (Bluemle, 1997). This interval corresponds to generally increasing precipitation and rising regional temperature (Groisman and Easterling, 1994; Gosnold et al., 1997). Unofficial records indicate that between 1776 and 1885, the Red River flooded at levels exceeding 15.2 m (Bluemle, 1997; Hoyt and Langbein, 1955). However, climate parameters across this region related to this interval are poorly known.

Finally, changing human land use patterns can have impacts on flood dynamics. Construction of drainage ditches and rural road ditches drain the land surface more efficiently than in the past. Put another way, the lag time for water draining the land surface and entering the Red River is greatly reduced. In addition, with increasing urbanization near the river, actual runoff volume can increase substantially. Together, these two factors often affect river flood dynamics by increasing the "flashiness" of rivers - i.e., flooding is often more rapid, more severe and more frequent than under natural conditions.

EQUIPMENT AND FIELD PROCEDURES

Two seismic reflection systems will be deployed during this study. The first is an EG&G analog UNIBOOM system (Fig. 3). It produces an acoustic impulse with a peak frequency near 1 kHz, yielding vertical resolution of 15-20 cm and can image sub-bottom stratigraphy to depths up to 200 m in fine-grained sediments such as the glaciolacustrine/lacustrine deposits of the Red River Valley. This system is ideal for mapping gross characteristics of sedimentary sequences in Lake Winnipeg and oxbow lakes along the river’s course. The second seismic system is a Knudsen Engineering Model 320 B/P dual frequency (28 kHz/200 kHz) echo sounding system. Although the higher frequency of this system provides less sub-bottom penetration (15-20 m maximum), the better vertical resolution (2-3 cm) in the uppermost portions of the sediment pile should be excellent for imaging deltaic and lacustrine sediments deposited during the interval of interest (ca. 500 years) for this study. Both seismic systems are operated from the Research Vessel Ozark Traveler (Fig. 4), a trailer-portable, 24 ft pontoon boat operated by the University of Arkansas for lake and river research. From the geophysical data, detailed 3-D maps of deltaic deposits in the lake can be processed for use in reconstructing stratigraphic/depositional successions and identifying appropriate targets for core sampling.

Fig. 3. Uniboom seismic profiling equipment being towed from the University of Arkansas Research Vessel Ozark Traveler. Fig. 4. The University of Arkansas Research Vessel Ozark Traveler, a 24-foot pontoon boat for research on lakes and rivers.

Real-time navigation and positioning data will be provided during the seismic surveys by two Leica System 300 GPS receivers with radio communication. The two receivers provide sub-cm accuracy on base-lines 10 km or shorter (such as those proposed here). This will enable very accurate mapping of the sedimentary deposits and ensure precise location of core samples.

Previous seismic reflection and core sampling studies of deposits in Lake Winnipeg (Todd et al., 1998a, 1998b; Moran and Jarrett, 1998) address the gross seismic stratigraphy and detailed aspects of the sediment that may be observed by seismic reflection. These studies serve as a basis for designing a closely-spaced network of high-resolution seismic reflection profiles to be surveyed across the Red River delta in Lake Winnipeg and oxbow lakes along the Red River using real-time differential GPS positioning.

POTENTIAL IMPACT OF WORK

Detailed study of the paleoflood record of the Red River of the North will provide an empirical data base from which to evaluate the maximum credible flood. In addition, documentation of the frequency of varying magnitude flood events over an extended (geological) time interval should help refine estimates of the probability of floods at many scales.

Further, because the geologic record of flooding is coupled with decadal- to century-scale climatic variations, it will be possible to create different flood risk assessments for different climate scenarios. Clearly, flood risk during drier intervals would be different than flood risk during wet intervals and this should be considered in mitigation strategies.

REFERENCES

Bluemle, J.P., 1997, Factors affecting flooding in the Red River Valley, Proc. NDAS, v. 51, supplement 1., p. 17-20.

Brunskill, G. J., Elliott, S. E. M. & Campbell, P. (1980) Morphometry, Hydrology and Watershed Data Pertinent to the Limnology of Lake Winnipeg. Canadian Manuscript Report of Fisheries & Aquatic Sciences, No. 1556.

Brunskill, G. J., Campbell P. & Elliott, S. E. M. (1979) Temperature, Oxygen, Conductance and Dissolved Major Elements in Lake Winnipeg. Fisheries & Marine Service Manuscript Report, No. 1526.

Gosnold, W.D., Todhunter, P.E., and Schmidt, W.L., 1997, The borehole temperature record of climate warming in the mid-continent of North America, Global and Planetary Change, 15, p. 33-45.

Groisman, P. Ya., and Easterling, D.R., 1994, Area-average precipitation over the contiguous United States, Alaska, and Canada, in T.A. Boden, D.P. Kaiser, R.J. Sepanski and F.W. Stodd (Editors), Trends ‘93: A compendium of Data on Global Change. ORNL/CDIAC-65, Oak ridge Nat. Lab., pp. 786-798.

Hoyt, W.G., and Langbein, W.B., 1955, Floods, Princeton Univ. Press.

Last, W.M., and Teller, J.T., 1983, Holocene climate and hydrology of the Lake Manitoba Basin, in Teller, J.T., and Clayton, L., editors, Glacial Lake Agassiz, Geol.. Ass. Canada Special Paper 26.

Moran, K., and Jarrett, C.A., 1998, Lake Winnipeg sediment physical properties: interpretation, composite stratigraphic sections and calibration of acoustic reflection profiles, J. Paleolimnology (in press).

Todd, B.J., Michael, Lewis, C. F. M., Thorleifson L. H., Nielsen, E., and Last, W.M., 1998, Paleolimnology of Lake Winnipeg: Introduction and Overview of Special Issue, J. Paleolimnology (in press).

Todd, B. J., Lewis, C. F. M, Nielsen, E., Thorleifson, L. H., Bezys, R. K. and Weber, W., 1998, Lake Winnipeg: geological setting and sediment seismostratigraphy, J. Paleolimnology (in press).