An Evaluation of Alluvial Fan Agriculture

John J. Field

Floodwater farming on alluvial fans takes advantage of water that rapidly moves through desert basins following storms. Settlement patterns in the Marana Community indicate widespread dependence on cultivation by this method, primarily in Zones 1 and 4 of the bajada below the Tortolita Mountains on the east, but also in the areally limited portion of Zone 6 between the Tucson Mountains and the Santa Cruz floodplain (Figs. 3.8,3.9). Knowledge of floodwater methods comes in large part from Piman Indian practices of the late nineteenth and early twentieth centuries in topographic situations resembling those of Zone 1 (Bryan 1922, 1925, 1929; Castetter and Bell 1942; Underhill 1939). As described, this ak chin farming of the late historic period emphasized positioning of fields in ideal locations of natural overflow from drainages with a minimum of structural improvements for managing water. Additional studies of historic Piman agriculture (Nabhan 1983, 1986) and traditional farming methods in northwestern Mexico (Doolittle 1984, 1988) illustrate a wider latitude of more labor-intensive floodwater field systems that also have parallels in the Marana Community.



Approximately 390 square kilometers (150 square miles) of the lower bajadas in the surveyed area north of Tucson are composed of late Holocene alluvial fans (Figs. 1.8, 5. 1). These fans are formed primarily from the erosion and redeposition of older Pleistocene and early Holocene alluvial fans found farther upslope. The late Holocene fans are actively prograding onto floodplains and terraces of the Santa Cruz River. Two depositional facies are ubiquitous on the late Holocene fans: channel gravelly sand facies, and sheetwash silty sand facies. Facies are units of sediment deposited by a single depositional process (Reading 1978). By analyzing the distribution of depositional processes on an alluvial fan, paleoenvironmental reconstructions can be made to help assess the prehistoric agricultural potential of a particular fan.

The channel gravelly sand deposits are well-stratified, relatively well-sorted, loose, light colored sands containing up to 30 percent gravel (Fig. 5.2). Silt content is never greater than 5 percent. Agricultural drawbacks of coarse texture would be rapid infiltration of water to depths beyond crop roots and rapid evaporation of moisture content. The facies is typically 1 m to 2 m thick with sharp erosive bases and extends laterally for 50 m to 400 m before interfingering with silty sand deposits. The top of each channel unit also grades into sheetwash sediments.

The sheetwash silty and facies, in contrast, is composed of very poorly sorted, massive, slightly hard, yellow orange (10 YR 6/3) silty sands containing between 5 and 50 percent silt (Fig. 5.3). Moisture retention would be greater than for the channel gravelly sand facies because of the higher proportion of silt. Gravel content rarely exceeds 10 percent. Massive textures result from heavy bioturbation of originally laminated silts and interstratified gravel lenses. Silty sand deposits completely enclose channel units, and are frequently over 3 m thick where channel deposits are absent. Archaeological features are abundant and well preserved in this facies.

The coarse texture, erosive nature, and stratification of the gravelly sand facies indicate deposition in confined high-energy environments characteristic of ephemeral washes or channels seen on active alluvial fan surfaces. The high percentage of silt, excellent preservation of archaeological features, and heavy bioturbation are evidence for low-energy processes operating in the silty sand facies. Laminated silts and sands are characteristic of sheetflood deposits below the fan intersection point (Packard 1974). The intersection point on an alluvial fan is where channel depth becomes zero, below which the

Figure 5.1. Geomorphic map of the surveyed area north of Tucson.

Figure 5.2 Typical exposure of well-stratified channel
gravelly sand facies. Erosional contact with underlying
silty sand facies is visible at bottom
of trowel.

Figure 5.3. Typical exposure of sheetwash silty sand
facies, showing massive texture enclosing a gravel lens to the
right of the trowel.

stream flow becomes unconfined (Hooke 1967). This position equates with the location of natural water spreading described historically as preferred for Tohono O'odham ak chin (floodwater) farming (Bryan 1929: 449; Castetter and Bell 1942: 125).

By comparing depositional processes on active alluvial surfaces with those discussed above, a depositional model can be developed to reconstruct the paleohydrology of prehistoric surfaces. Portions of the fan accumulating sediment of a particular facies are revealed by surface indications of depositional processes. In stratigraphic profile, vertical variations in facies reflect changes in and document lateral migration of depositional processes through time at a single point.

Figure 5.4 is an aerial photograph of Cottonwood Fan displaying several active processes as well as inactive surfaces in Zone 1 of the Marana Community. Cottonwood Wash is well confined and channelized at the fan head. Farther down the fan a decrease in channel depth results in an increase in channel branches and a decrease in channel width. Below the intersection point where channel depth becomes zero, stream flow becomes unconfined and the aerial extent of deposition increases greatly. Sedimentary facies deposited in various reaches on the fan surface reflect the observed processes. Confined washes at the fan head are eroded into older fan sediments and are filled with gravelly sand channel deposits. The silty sand facies dominates below the intersection point. Similar longitudinal variations in depositional processes on alluvial fans are recorded elsewhere (Hooke 1967; Rahn 1967; Packard 1974; Bull 1977).

Subsurface longitudinal profiles constructed from evenly spaced backhoe trenches provide evidence for prehistoric distributions of depositional processes. Like surficial variations, channel deposits are most prevalent at the fan head and sheetflows dominate at the fan toe (Figs. 5.5, 5.6). Sections through Derrio Fan (Fig. 5.5) display a basal channel grading into siltv sands. Upward trends in depositional processes observed in vertical profiles are similar to downfan longitudinal variations and indicate a headward (upstream) migration of facies. As channels at headward portions of the fan begin to fill due to low and intermediate stage flows, the intersection point is restricted to higher portions of the fan (Packard 1974). As a result, sheetflow sediments are deposited above channel fill. When channels at the fan head become completely choked with sediment, evulsion or cutoff of the main trunk stream occurs, a lower portion of the fan becomes active, and the depositional sequence outlined above is repeated. Each sequence creates a fan lobe that becomes a small portion of the entire alluvial fan; each alluvial fan is comprised of several vertically stacked and laterally adjacent fan lobes.

Although deposition on all fans is consistent with the processes and model presented above, variability in fan configuration exists because of differences in drainage basin area and distance from the mountain fronts. Thicknesses of fan lobes, channel dimensions, and aggradation rates increase with increased drainage basin area or proximity to mountain fronts. Because channel dimensions and the corresponding ability to contain high flows increase with fan size, channel deposits on large fans extend to the fan toe (Fig. 5.5). Sheetflows, which occur where floodwater overflows established channels, are more easily induced on small fans because the smaller channels are unable to contain flow converging on the fan surface. Also, the intersection point on small fans is

Table 5.1. Drainage Characteristics of Selected
Washes In the Surveyed Area North
of Tucson

Figure 5.4. Aerial photograph of Cottonwood Fan in the Marana survey area. There
is an increase in channel branching and extent of deposits down the fan on both active
and inactive surfaces. Fan head is at top. Area in photograph is 1.2 km across.

closer to the fan head, causing proportionally greater deposition of the silty sand facies (Fig. 5.6). Regardless of drainage basin area, alluvial fans are larger, much siltier, and contain very few channel deposits when mountain fronts are a great distance from the fan head such as with Durham Fan near McClellan Wash in the far northern part of the surveyed area (Fig. 5.1, Table 5.1). In contrast, alluvial fans relatively close to the mountain front, as in the Marana Community, are much sandier and more channelized.

Figure 5.5. Longitudinal profile of Derrio, Fan in the Marana survey area.

Figure 5.6. Longitudinal profile of a small fan in the Marana
survey area, showing the dominance of sheetflow deposits.


Several factors influence the potential for direct agricultural use of floodwaters on late Holocene alluvial fans: location and areal extent of sheetflooding on active fan lobes, frequency and intensity of rains, and minimum discharge needed to induce overbank flow. When depositional facies are considered as records of water flow over an alluvial surface, the farming potential of each fan can be analyzed. Areas frequently inundated by sheetfloods provide the ideal conditions for floodwater farming of the type preferred in late historic times, because water evenly wets the entire active surface and the need for water diversion or labor-intensive irrigation is alleviated. Water residence time is much greater on sheetflow surfaces than in areas where flow and the corresponding opportunity for soil infiltration is confined to channels; denser vegetative growth is found in these areas than in adjacent channelized and depositionally inactive reaches of the fan (Packard 1974). As discussed above, both drainage basin area and distance from the mountain front affect the amount of fan area experiencing sheetflooding. Hence, fans far from the mountain front or fans with small drainage basins are most suitable for water distribution on floodwater fields with a minimum of human intervention and labor.

The entire surface of a fan is not active at any given time; conditions are conducive to floodwater farming only in active reaches. Inactive portions of the fan remain dry because large, deep channels contain flow and thus adjoining surfaces are isolated from water supplied by the drainage basin. By calculating the amount of area on each fan that is presently active, an estimate can be made of the relative amount of land available to floodwater use through natural water spreading during prehistoric times. Because the percentage of area actively aggrading varies from fan to fan depending on size, a sampling of both large and small fans was used (Table 5.2). By extrapolating the data presented in Table 5.2, 15.3 percent of the 390 square kilometers (150 square miles) of late Holocene fans is presently active. This area was perhaps greater in prehistoric times because recent channel cutting has reduced viable agricultural land.

Given equal volumes of incoming water, minimum bankful discharge calculations for channels on different alluvial fans establish the relative likelihood of overland flow (surface flooding) occurring on one fan as compared to another. Sheetflow is induced when channels become filled beyond capacity. As a result, fans with small channels tend to produce overbank flow more readily and therefore may be particularly convenient for floodwater techniques.

Table 5.2. Amount of Active Fan Area on Selected Fans
In the Surveyed Area North of Tucson


Table 5.3. Minimum Bankful Discharge
Calculations for Channels on Three
Fans in the Marana Survey Area

Calculations use the Manning equation:
Q = A(1/n)r2/3s1/2, where Q = discharge,
A = channel cross-sectional area,
r = channel depth, and s = channel slope.

Channel area, slope, depth, and roughness are needed to calculate bankful discharge. Minimum discharges for three channels in the Marana area were calculated (Fig. 5.7, Table 5.3). The relative ranking of agricultural potential through natural flooding determined from these measurements is consistent with other factors already discussed. The lowest minimum discharge occurs on the fan with the smallest drainage basin and highest percentage of sheetflow deposits. The highest minimum discharge occurs in the present channel of Derrio Wash, which is eroded into a late Holocene aggradational surface at the bottom of a permanently entrenched channel cut into Pleistocene alluvium. This situation stresses that the once arable aggrading surface on the floodplain of the wash bottom is presently isolated from all but the largest flood events.

Far from the mountain front, drainage basin area has a reverse influence. Floodwater potential may be high on fans far from the mountain front, where sheetflooding is

Figure 5.7. Locations of fan channel calculations in Table 5.3.

easily induced because of the low slopes and rapid loss of channel depth. Furthermore, large portions of the basin slope near the fan head are included in the drainage area supplying water for sheetflows; a majority of the drainage basin is located on the piedmont when the fan head is far from the mountain front (Table 5.1).


The frequency at which discharge conditions are reached and surface flooding occurs on alluvial fans is impossible to calculate without considerable data on rainfall patterns, duration, and intensity. Observations and data were collected during the summer of 1985 to aid in understanding the frequency and intensity of floods on different portions of the bajada along the two largest drainages of the Tortolita piedmont, Cottonwood and Derrio washes (Fig. 5.7). Although direct observations were difficult because of accessibility, data were gathered by monitoring geomorphic gauges and conducting personal interviews with local residents. Observations provide additional information on water resources.

Seven scour chains were placed in various locations along the length of Cottonwood and Derrio washes to see how a flood could differentially affect portions of the bajada. The four chains in the Cottonwood drainage were placed (1) at the mountain front, (2) on the upper bajada, (3) on the upper portion of the Holocene fan, and (4) downfan from the previous position, where extensive branching of the wash begins. Scour chains along the Derrio drainage were placed (1) at the mountain front, (2) at the channel fan head, and (3) at the Holocene fan head. The scour chains were made from a 5-foot length of link chain attached at one end to a piece of 7-foot rebar. The attached end is placed in a 5-foot hole standing straight up, so that when the hole is filled, the chain is completely buried by sediment. Two feet of rebar remains above ground for relocation. At times of flooding, sediment is washed away and the portion of chain exposed lies flat on the surface. During waning stages of the flow, sediment is redeposited over the chain. By measuring the depth to which the chain is covered and the length of chain lying flat, both the amount of deposition and erosion, respectively, can be determined. By subtracting the amount of fill from the amount of scour, a net sum of erosion or deposition can be obtained for each flood event.

The seven localities were revisited several times during the summer rainy season. Measurable scour and fill was recorded at all stations at least once except for the farthest downslope Cottonwood station, which is apparently too far downfan to be affected by moderate flows. Net scour and fill is nearly zero throughout the drainage systems, with total depth of scouring increasing toward the mountain front. Vegetation backed up behind the rebar gives a relative measure of depth of water flow and also appears to increase toward the mountain front. Through the 1985 summer season only two flow events affected the entire drainage basin; the 1985 summer rains were considered below average in length, intensity, and duration.

Interviews with local residents and personal observations provided some of the most useful information concerning flow in Cottonwood and Derrio washes. Discussions with residents revealed that the late August storm that washed out the Derrio mountain front station occurred only in the mountains; no rain fell on the bajada itself. Despite this, the county line road crossing Derrio Wash 10 km (6 miles) from the mountains was washed out the same night (Fig. 5.7). These observations indicate large storms in the mountains can induce significant flows on the bajada several miles from the mountain front, and that in some instances these flows may reach the Holocene fans near the Santa Cruz River. A resident also noted that pulses of sediment were occasionally washed out of the mountain canyon of Cottonwood Wash and were redeposited as a lobe of sediment in the wash floodplain at the mountain front. Migration of these pulses downstream with successive flood events would eventually supply sediment to active alluvial fans near the river.

On several visits through the summer, observations of stream flow out of Derrio Canyon at the mountain front (Zone 4 of the Marana Community in Fig. 3.9) reflected the flashiness and short duration of flow during the 1985 summer rainy season. During an initial visit to the canyon in April, water was flowing out of the mountain front and continuing along the flank of the mountains for over 2 km (1.2 miles). In June, before the summer rains began, flow continued past the mountain front for only 75 m before soaking into the sand and disappearing underground. Continued observations through the rainy season revealed that despite the summer rains, water flow in the canyon was found farther and farther upstream with each visit. These observations suggest that the rains do not appreciably add to the water table and thus do not affect the effluent nature of Derrio Canyon, although they may produce peak flow in short-lived floods.

The sudden and rapid nature of the rains result in flash floods that rapidly run off overland without soaking into the ground or recharging the water table. In contrast, winter rains are of a lower intensity and longer duration, which promotes saturation of the ground and replenishing of the water table. Because of these differences in rainfall patterns, upland perennial flow in Derrio Wash is at its peak at the end of the winter rains. Both perennial flow and summer floods may be used to produce crops with different techniques and in different locations. (A related discussion of temperature inversion and spring crops on upper bajadas is in Chapter 4.) Variations in rainfall patterns may also affect sediment supply in the drainage basins. During the lowintensity winter rains, sediment is supplied to the washes from adjacent hillsides but cannot be transported long distances downstream. Channel filling would be promoted and the creation of low terraces would be favored. In contrast, the high-intensity summer rains favor erosion, sediment transport, and redeposition where stream flow infiltrates sandy channels.


Information from interviews and scour chain analysis are helpful in understanding prehistoric settlement patterns and subsistence strategies along Derrio and Cottonwood washes. First, perennial flow in Derrio Canyon provides a year-round water supply, although slightly longer distances into the mountain canyon must be traversed in the summer and fall than during the rest of the year. Second, the late August flood that originated in the mountains but induced flow far down on the bajada has interesting implications for farming along these drainages. Successful floodwater farming is dependent on flash floods reaching the surfaces being farmed. If floods in the mountains can reach surfaces several miles away, the farming potential is much higher than if only local rains provide water. Channel fans aggrading on bottomland in larger washes during Hohokam occupation would have been flooded by storms similar to the one in late August; that flow did not overflow onto the current surfaces of the prehistoric channel fans because of the present entrenchment of active stream channels.

Small fans with drainages not extending to the mountains were heavily utilized by the Hohokam for farming. Floods originating in the mountains would not provide water to these smaller fans. However, heavy rains on the upper bajada can apparently flow the few miles needed to reach the fan heads on the lower slope.

Success of floodwater farming for individual locations is uncertain from year to year because of the spottiness and irregularity of intense rainfalls. Unpredictability of flow is greater for small fans with small drainage basins. Farming of individual fans is therefore risky on an annual basis. However, observations suggest that any given year may provide excellent opportunities for agricultural water on any alluvial fan if rainfall patterns are favorable.

Other factors can also hinder farming conditions on an alluvial fan. The development of arroyo networks results in channelization of flow and decreases the chances for favorable sheetflow. Such events are most likely at fan toes on the lower bajada where large arroyos along the river can impinge headward into the fan surface. The sudden migration of active fan reaches during large storm events may also hinder the agricultural potential of not only the old but also the new active surfaces. When shifting creates active surfaces farther from the source area, the distance water must travel is increased, thus lowering the frequency at which floods reach these active fan areas.

Although there are 390 square kilometers (150 square miles) of late Holocene alluvial fans in the surveyed area north of Tucson, only approximately 15 percent is active at one time. Within this fraction of fan territory, favorable conditions for floodwater farming without improvements are found where sheetflows dominate. Fans far from the mountain front or with small drainage basins have the lowest bankful discharges, greatest percentage of sheetflow deposits, and consequently offer agricultural advantages with this technology. These favorable conditions for floodwater farming are balanced against the unpredictability of rainfall patterns.

Next: Chapter 6 | Previous: Chapter 4

UAP Online Books  |   UAP Home Page

The University of Arizona Press, 2/10/01 5:03PM