COCKFIELD FLOW

In the Mississippi Delta, Cockfield ground water flow begins at recharge areas to the northeast, and moves southwesterly, down regional dip, toward the embayment axis. Disruptions to this basic flow system occur in Washington County. The Monroe Uplift structure, by affecting dip, affects flow. Additionally, in the 20th century, the flow system was complicated by the pumping activities of humans. Instead of discharging downdip, some flow is now rerouted to discharge to wells.

A network of wells has been monitored for water level changes for several years by OLWR. Measured depths to water standing in each Cockfield well, corrected to feet above or below mean sea level, form a mappable potentiometric surface which describes the pressure head of the aquifer. The potentiometric surface for the Cockfield slopes to the pumping centers at Greenville (Figure 17). Elevations of water level ranged from 27' to 133' above sea level during the 1991-2000 period. The average low was 35', average high 114'. In the 2000 season, levels were below average, from 28' to 89' above sea level.

Figure 18 presents the flow directions derived from the mapped surface. From the character of the potentiometric surface, the following points are evident.

1. Flow to the southwest ("normal" flow) dominates in the eastern and southern margins. The gradient is gentle.
2. In the Greenville cone of depression, the gradient is steep and flow converges on a central point near the major city wells.
3. The gradient is low in an area east of Wayside.
4. From Wayside south to Avon, the surface is disrupted and elevations may rise significantly over short distances.

MRVA FLOW

Groundwater flow in the MRVA begins in recharge areas, largely at Delta margins in the Bluff Hills region, and along the Mississippi River in a narrow zone of influence. Ongoing research at OLWR indicates direct recharge from the Delta's ground surface and surface streams is restricted by the surficial clay layer which is dense and impermeable in most locations. Figure 20 illustrates the potentiometric surface of the MRVA, which is higher (67' to 118' above sea level in fall 2000) than that of the Cockfield in this area. The local pattern of flow is from highs at the Mississippi River southeastward toward the Sunflower River drainage.

SPARTA FLOW

OLWR has not engaged in detailed mapping of Sparta levels in this area for this study. An approximation of the Sparta gradient has been mapped from data gathered during regional monitoring of water levels (Figure 21). Data points are from measurements during 1995 and 1996. The Sparta potentiometric surface in declines gradually to the south.

VARIATIONS OVER TIME

The net change in Cockfield water levels from 1991 to 2000 as measured in the autumn surveys is mapped in Figure 22. Over this nine-year interval there was an average decline in wells of about 4 feet. However, the pattern was not a gradual decline in each well. OLWR observers found large variations in water levels. The most dramatic example was the change measured from autumn 1992 to spring 1993 (Figure 23), when the water level in the center of the cone of depression at Greenville rose by as much as 31 feet.

Some years were net recovery years, and in some years a net decline was recorded. The contrast is apparent in the fall-to-fall change from 1991 to 1992 (Figure 24) and from 1996 to 1997 (Figure 25). The first period included a wet winter. The second was not a drought year but rainfall was at or near normal levels and no record floods were recorded. Annual recovery or decline in water level clearly varies with the years selected for comparison.

Seasonal patterns typically reveal "rebound" effects from wetter winter-spring seasons. Recovery during the fall 1991 to spring 1992 winter (Figure 26) centered in the Greenville cone of depression. During the summer of 1991 (Figure 27) declines were recorded. Then, change recorded over the winter of 1992-1993 (Figure 23) again illustrated another wet interval, but with recovery amounts distributed in different locations.

Virtually no interaction between surface streams and the Cockfield aquifer had been anticipated, since the Cockfield does not appear at the surface in this area, and the Yazoo had been assumed to be more significant as a barrier between the MRVA and Cockfield. Yet some water level data suggests significant interaction between surface water and the Cockfield aquifer. Water level changes (Figure 26) over the winter of 1991-1992 appear to relate to stages of the Mississippi River. A map of changes from spring 1993 to spring 1995 (Figure 28) illustrates an increase in water levels, on this occasion toward the south and east. This period contained the 8th highest flood on record for the Sunflower River (4/25/95), which lies east and south of the study area. There was no corresponding record event on the Mississippi River during that time.

CONTINUOUS WATER LEVEL MONITORING

In addition to the periodic monitoring of water levels in the key well network, equipment was installed to observe water levels continuously. During 1993-1995, levels in three observation wells were recorded hourly. Each well was inactive, with no pumping of its own to disrupt the water levels recorded (Figure 29).

The first well, D-67, is situated in downtown Greenville, less than one-quarter mile east-northeast of the center of the cone of depression created by large withdrawals in city wells (approximately at well D-37). The second well, D-56, is also in the cone, but about 3.5 miles south-southeast of the center. Rural well G-257 lies near the edge of the cone of depression, ten miles south of the center, east of Wayside.

SURFACE WATER / COCKFIELD AQUIFER INTERACTION

As previously described, seasonal patterns of recovery in monitored wells suggested there might be some interaction with surface waters. Figure 30 adds, to a graph of water levels in the three continuously monitored wells, a line representing Mississippi River stages, measured daily at the bridge gage southwest of Greenville, 9.8 miles from the cone center. There appears to be a correlation between the seasonal variation in water levels in Cockfield water wells with the river stages.

The Mississippi River passes 2.5 miles west of well D-67 at its closest point. Lake Ferguson, not part of the active river, lies less than 3/4 mile to the northwest of this well. The second well, D-56, lies 3.25 miles east of the Mississippi, and well G-257 lies about 7 miles east of the river. In its course through the immediate area, the mapped bottom of the Mississippi River channel, while resting on alluvium, cuts deep enough to approach the top of the Cockfield. For example, at river mile 541, thalweg profile data recorded by the Corps of Engineers maps the deepest part of the river only 10 feet above the Cockfield's upper limit. Because the river bottom shifts continually and may locally downcut deeper than has been mapped, and because the basal alluvial sediments are highly transmissive, there is likely a direct hydraulic connection between the river and the Cockfield aquifer at some locations in the study area.

If, when comparing river and well graphs, the well level scale is exaggerated, the correlation of fluctuations can be more easily seen. Figure 31 superimposes the water levels over the river stage curve, with the well data scale magnified to more closely match the range of variation seen in the river curve.

To quantify these correlations, 8:00 a.m. well transducer readings of water depth were converted to elevation above sea level, then compared with the daily river stages which are recorded at 8:00 a.m., also converted to elevation relative to sea level. For each well, daily correlation coefficients were computed for water level elevation versus river stage elevation at both the Greenville gage (Hwy. 82 bridge southwest of town, downstream, 9.8 miles from cone center) and the Arkansas City gage (upstream, 15.3 miles northwest of cone center). Daily measurements for a two-year period (October 6, 1993 through October 6, 1995) were used for the computations.

As the graphing of transducer water levels suggested, D-56 shows the strongest correlation, D-67 less, and G-257 the least.

In addition to the same-day 8:00 a.m. comparisons, coefficients were computed for each well's water level versus the river stage from one day prior, two days prior, etc. The data generated by this process are graphed in Figure 32.

For D-67 near the center of the cone, the same-day readings (day zero) had the greatest correlation. All coefficients declined from 'day zero', gradually and evenly. By Day 15, coefficients had fallen to .49458 (vs. Greenville gage) and .50313 (versus Arkansas City gage). When the same-day records were processed on an hourly basis, the maximum correlations, .72470 and .72532, were those linking the gages at 8:00 a.m. and the well at 11:00 a.m., three hours later.

At D-56, the same-day correlations of .83815 and .83135 were not the maxima. Values increased gradually to peak correlations on day 9, at .86803 and .86599. In this well, the correlation with the Greenville gage was slightly stronger until day nine, when the lines crossed and diverged, declining gradually to .85140 and .85240 by day 15.

At G-257, the patterns were similar to those found for D-56, though more subdued. Correlations began at .49532 and .49011, then increased gradually to a maximum day 12, at .51704 versus the Greenville gage and .51684 versus the Arkansas City gage. The graphed lines crossed and diverged after day 12; and by day 15, correlations were .51277 and .51432.

Long term correlations

To determine whether the observed correlation between water levels in wells and river stages actually occurs within the brief 12-day lag window observed, the period of comparison was extended to 800 days. Correlation coefficients were computed, again using water levels from 10/6/93 through 10/6/95, versus gage readings ranging from zero to eight hundred days prior to the well measurement. Figure 33 illustrates the long-term curves plotted from these results.

At D-67, near the cone center, correlations decrease fairly steadily, from .71694 and .71921 at Greenville and Arkansas City respectively, to a minimum about six months later: -.32753 on day 161, and -.33845 on day 160. Coefficients ascend to reach new peaks of .55885 at 351 days, and .56338 at 352 days.

At D-56, the wells with the highest degree of correlation short term, correlations fall from peaks on day nine to lows of -.39010 on day 192 versus Greenville and -.40632 on day 194 versus Arkansas City. They rebound to peaks of .48853 after 376 days and .49524 after 377 days. These second peaks fall 367 and 368 days from the first peaks on day nine, a nearly perfect annual cycle.

At G-257, the well most distant from the river and cone center, correlations decline less steeply from the initial peaks on days 12 and 13, then fluctuate in the .25 to .2 range to about the 150-day point, then decline to lows of -.34280 at day 272 versus Greenville, and -.35615 at day 273 versus Arkansas City. A new set of peaks occurs at day 383 (.42629) and day 374 (.42129). A second, higher set of peaks is seen at day 441 (.50246 and .51105).

For all wells, the highest correlations between river stage and water levels in the wells occurred during the initial 12 day window. Correlations dropped off and did not approach the original values again until the following year, and never exceeded the correlations achieved in the initial 12-day window. The data illustrate a fairly rapid response in wells to rises in river stage. This is compatible with the positive correlation of river stages and water levels in the MRVA recently documented in alluvial wells in the Swiftwater area (Byrd, 2000, p.40). As in the Byrd study, the most rapid response was not found in the shallowest of the wells. In this study, same-day response was in well D-67, the cone center well, though its basal screen is 514' below the surface. The outlying wells, farthest from the river, reacted several days later, despite higher screens, at 411' deep (D-56) and 359' deep (G-257). This information supports Byrd's observation that vertical continuity is quite high in the MRVA, and suggests a similar phenomenon occurs in the Cockfield.

PREVIOUS		NEXT
	TOP