Sunteți pe pagina 1din 25

Society for American Archaeology

Hydraulic Engineering Aspects of the Chimu Chicama-Moche Intervalley Canal


Author(s): Charles R. Ortloff, Michael E. Moseley, Robert A. Feldman
Source: American Antiquity, Vol. 47, No. 3 (Jul., 1982), pp. 572-595
Published by: Society for American Archaeology
Stable URL: http://www.jstor.org/stable/280236
Accessed: 26/11/2009 01:52
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=sam.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access to
American Antiquity.

http://www.jstor.org

ANTIQUITY
AMERICAN

572

[Vol.47, No. 3,1982]

HYDRAULICENGINEERINGASPECTS OF
THE CHIMU CHICAMA-MOCHEINTERVALLEYCANAL
Charles R. Ortloff, Michael E. Moseley, and Robert A. Feldman
Of the many canal systems of the Chimuempirethe Chicama-MocheIntervalley(LaCumbre)Canalconnecting the Chicamaand Mochevalleys represents the highestlevel of technicalachievement.Thispaper examines
the engineeringskills of the Chimuas revealed by computeranalysis of the open channel flow design techniques they utilized.Analysis of agriculturalstrategies made possible by this canal and the surveyingskillsinherent to its use are examinedin detail. The presence of many trial canal paths toward the distal end of the
canal Indicate extreme difficultyin overcomingtectonically induced ground-slopechanges caused by fault
lines near the intervalleydivide.The canal was abandonedprior to completionof constructionand thus never
served to supply the Moche Valley with Chicamawater.
Irrigation agriculture requires a sophisticated combination of engineering knowledge, agricultural expertise, and political coordination. As such, it provides a measure of the degree to which a
civilization has progressed. The study presented here seeks to analyze the engineering aspects of
the Chicama-Moche Intervalley Canal, a massive canal project build about A.D. 1200 on the north
coast of Peru. This analysis assesses not only the knowledge of open channel flow design attained
by the Chimu engineers, but it also reflects on our current abilities to understand prehistoric
engineering.

GENERALCONTEXT
Coastal Agriculture

Fed by seasonal highland rains in the Pacific watershed of the Cordillera Negra, the lower
Moche and Chicama drainages cross the arid Andean coast lands which form part of the world's
driest desert (Lettau and Lettau 1978). The climate allows year-around agriculture supported
principally by irrigation from large-scale canal systems. Exceptionally high yields are possible
and are reflected in the fact that more than 80% of the lower Moche and Chicama irrigated lands
are currently under intensive, mechanized cultivation by agroindustrial complexes supplying the
international export market.
There has been growing pressure to increase agricultural production by expanding the amount
of land under irrigation, but this expansion has been inhibited by topographic irregularities and a
marked differential distribution of runoff relative to arable land. Topographic relief is extreme, in
part, because the Cordillera Negra fronts the Pacific seafloor, or Nazca Plate, which is actively
underthrusting the continental margin at an estimated rate of 10 cm per annum, producing a
young, mountainous terrain.
Broken topography and marked differences in the abundance of land and water require large
investments in any project designed to move runoff from a drainage where it is abundant across a
mountain divide to an area where arable land is abundant. An intervalley canal system designed
to connect the two drainages north of the Rio Chicama is currently under construction. Based
upon foreign financing and European engineering and technology, it has an investment cost of
millions of dollars that will be canceled only after decades of increased agricultural yields. The
CharlesR. Ortloff,GeneralElectric Company,Nuclear EnergyDivision,175 CurtnerAvenue, San Jose, CA
95125

Michael E. Moseley,RobertA. Feldman,Field Museumof Natural History,Chicago,IL 60605


Copyright (

1982 by the Society for American Archaeology


0002-7316/82/030572-24$2.90/1

REPORTS

573

scope of this project reflects the fact that intervalley systems are of necessity large-scale undertakings. This reclamation project is one of several multinational projects that intrude upon an
area formerly united by a single, far larger network of now-abandoned Prehispanic canals.
The five drainages above the Chicama-Moche connection share well preserved prehistoric intervalley canals forming the Lambayeque "megasystem," which Paul Kosok (1940, 1965) identified as a complex of Chimu-phase reclamation works. He argued that the Chicama-Moche link,
formed by the so-called "La Cumbre Canal," was sponsored by the Chimor Dynasty, for the purpose of bringing Chicama water to the imperial capital of Chan Chan and its environs. By 1450
A.D., Chimor rule extended from central Peru to near southern Ecuador, and Kosok believed the
Lambayeque region was incorporated into the empire as an ongoing megasystem developed by
older states in the area. If effectively farmed, the area encompassed by the now-abandoned megasystem would account for approximately one-third of all agricultural land ever reclaimed along
the entire Peruvian coast. Kosok overlooked a partially preserved irrigation linkage between the
Lambayeque system and the Chicama drainage that would have united what he considered to be
two distinct multivalley networks. He also thought that La Cumbre Canal had been a viable undertaking, when in fact it was never put into operation. However, given the geographical scope of
Kosok's pioneering research, these are small changes; his effort was not concerned with the
details about how much of the multivalley network was operational at any one point in time.
Preservation Patterning
Understanding the relative state of preservation of abandoned irrigation structures is central
to both the study of their hydraulic engineering, and to issues concerning the time of operation
and the degree of efficiency. Agricultural remains are not uniformly preserved in the Moche
Valley. Excavation has identified formerly functional canals lacking preserved fields and furrows, as well as furrowed fields lacking preserved canals; it is not always clear which features
functioned with others. These complications made La Cumbre Canal a useful case study because
the project was aborted as it neared completion and thus was not subjected to the modifications
associated with long-used systems.
Stretching over more than 70 km, construction went on over the full length of La Cumbre's
course. Masonry channel lining of the final phases is present in areas north of Quebrada del Oso
(Figure 1), while to the south the course is intermittently trenched but lacking channel leveling,
lining, and vertical alignment with completed aqueducts; it could not have carried water to field
systems laid out in the vicinity of Chan Chan. There is a general correlation between completeness of construction and degree of preservation. South of the Quebrada the trajectory of the canal
can be followed, but neither intended slope nor sidewall configuration can be accurately extrapolated on the basis of unfinished trenching (Pozorski and Pozorski 1981). It is the completed
masonry channel north of Quebrada del Oso that records Chimu channel design and forms the
focus of this study.
Due to their great length, intervalley canals in Peru are subject to differential preservation and
selective distortion generated by two environmental forces: El Nifio perturbations and tectonic activity.
El Ninio
A complex set of meteorological and oceanographical interactions, called "El Nino," combine
to disturb the normal desert conditions on the Peruvian coast (Wyrtki et al. 1976; Cromie 1980).
Besides disruption of marine life, strong Ninos often produce precipitation onshore. These rare
but recurrent torrential rains occur in the study area no more frequently than once per 15 years,
and often less than once per century (Nials et al. 1979; Moseley 1981b). In severe El Ninlo
downpours, runoff charges normally dry drainages or "quebradas," producing violent flash
flooding. Where Prehispanic canals cross large quebradas they have often been washed out,
whereas between drainages they have survived sheetwash and unchanneled erosion.

[Vol.47, No. 3,1982J

AMERICANANTIQUITY

574

_km

<~~~

Chcm
6km
Sa
odH

0t,

200CX

ur Crcetsa

Le?scans

Chicama

Hi~~~~tect~ion scaj

~~~~~~~~~~~~7*50'

420~~~~~~~~~~~A

La

Huanchaco

7 910'

Figure 1.

Cha

Ca

of Canals 0C
Pampalet

Paraleleranzal

79-

CnJ

ta5

|0

The Chicama-Moche Intervalley Canal (adapted from Kus [1972fl.

REPORTS

575

Tectonic Distortion
The Chicama-Moche watershed is experiencing tectonic displacement. Subduction of the
seafloor begins less than 200 km to the west and underthrusts the entire northern Cordillera at a
dip of c. 100 and a rate of c. 10 cm per annum. Ongoing uplift and subsidence are most easily
gauged along the coast, but substantially greater seismic activity occurs inland (Barazangi and
Isacks 1976), supporting the contention that relatively greater vertical displacement goes on in
the interior of the watershed (James 1971; Myers 1975). Tectonic displacement of the watershed
is associated with only weak tensile or compressional stress and most movements are in the form
of oscillation of crustal blocks without marked lateral shear (Zeil 1979). Uplift is evident in 2-, 3-,
and 8-m marine terraces at the mouth of the Moche Valley (Cossio and Jaen 1967:57), which must
have formed after about 3000 B.C. (Richardson 1981). The highest preserved structure is tentatively dated to about 500 B.C. (Nials et al. 1979:8), and the lower terraces are of more recent
Moche or Chimu time-phase origin.
Rates of tectonic activity relative to points and planes of crustal movement induce selective
stresses on hydraulic structures. The most detailed local record of ongoing tectonic motion during
the last two decades comes from the port city of Chimbote, three valleys south of the Rio Moche. A
great deal of landscape uplift is gradual and often involves downward oscillation. Between 1960
and 1970 land-to-sea-level relationships at Chimbote registered a total vertical movement of 18
cm, consisting of a 12-cm rise followed by 6 cm of subsidence, resulting in a net uplift of 6 cm for
the decade (Wyss 1978). Vertical landscape movements averaging 1.8 cm a year induce very different stresses upon a structure such as La Cumbre Canal than do seismic events which have
abruptly raised sections of the Andean coastline from 1 to 3 m (e.g., Darwin 1839:379; Herd et al.
1981). On May 31, 1970, an earthquake measuring 7.7 on the Richter Scale and centered c. 25 km
seaward of Chimbote triggered seismic compaction and spreading of surficial sediments that,
among other things, caused roadways and walls near the dock to settle by as much as 1 m. Farther
inland, movement of dry-slope debris shifted the foundations of a steel mill 30 cm down. Ground
shifting under a railway north of Chimbote caused twisting of the rails and slope changes of 1 to 2
ft per hundred, a tilt of more than 10 (Ericksen et al. 1970:21-23).
There is, unfortunately, no information on canal alterations either during the quake or from the
18 cm of gradual vertical movement during the previous decade. In the case of severe seismic
shock, liquefaction of earth-bank canals is expectable in channels transporting water at the time.
This process lacks, as yet, archaeological verification by excavation. However, seismic impacting
has tentatively been identified in one branch of the Lambayeque complex with possible correspondence in a Chimu primary canal of the Moche Valley (Shimada 1980). While La Cumbre
Canal would not have been subject to liquefaction, the course-especially
artificial earth-fill terraces and aqueducts-would
have been subject to alteration from sediment compacting and
spreading, as well as to tectonic distortion of crustal blocks.
In the case of gradual tectonic movement, unlined, earth-bank canals are probably self-adjusting to slow landscape slope changes when transporting sufficient water for channel erosion to induce the shape changes necessary for maintenance of subcritical flow (Ortloff et al. 1981). A better engineering record of flow design is preserved by lined channels where erosional modification
during use is controlled. Still better is an unused, lined channel which should reflect purely
theoretical design intent, unmodified by use-testing.
Lined versus unlined channels should also provide somewhat different records of cumulative
tectonic displacement. For example, on the south side of the Moche Valley excavation has identified formerly functional canal sections, which theodolite survey shows have zero slope or run
slightly uphill. It is not clear what percentage of slope alteration is the product of ongoing gradual
displacement or what was induced by individual seismic events. However, the cumulative slope
change represented by unlined channels should date after use, when erosional self-correction
ceased. Alternatively, the cumulative alteration of lined channels should begin after their construction and include time of use as well as abandonment. Thus, La Cumbre Canal should reflect
the effects of approximately 800 years of tectonic movement and seismic activity.

576

AMERICANANTIQUITY

[Vol.47, No. 3,19821

The points and planes of crustal movement induce selective slope distortion in a structure the
length of La Cumbre Canal because it crosses over and between oscillating bedrock blocks, some
exposed and others sediment-covered. Most of the blocks are graphically defined by the drainage
pattern in terms of quebradas following faults and interquebrada plains riding on block surfaces.
There is selectively greater stress where the canal crosses between blocks than where it rides
atop a block of interdrainage surface. Instrument sightings north of the intervalley divide show
that different stretches of the completed canal, from c. 0.5 to 8 km or more in length, have different gradients. Most sections have shallow declinations with slopes falling below 0.013 (13
m/km; downhill slopes are given as positive, uphill as negative). In other sections, the preserved
course has shallow uphill gradients with slopes falling below - 0.006. Although total cumulative
slope alteration over 800 years falls below the 1 to 2% warping (0.01-0.02) of the Chimbote rail
line during the 1970 seismic event, it is significant that the greatest uphill slope changes in La
Cumbre Canal occur not on interquebrada plains, but where the course contours up quebrada
sides in the process of crossing between blocks. This means that the greatest vertical distortion is
registered in crossing from the side of one block to the side of another. Horizontal displacement is
more difficult to gauge because of the washout of aqueducts and channels in quebrada midpoints.
Presumably, many washouts occur at points of greatest stress, which would be at fault lines.
CANAL DESCRIPTION
The Chicama-Moche Intervalley Canal has impressed various authors (Larco Hoyle 1945, 1946;
Lumbreras 1974; Von Hagen 1965) and has received study by Kosok (1965) and Farrington (1980;
Farrington and Park 1978), with Kus (1972) reporting in greatest detail. According to Kus, the intake and initial section of La Cumbre Canal are probably represented by the modern Sausal Canal
which feeds off the south side of the Rio Chicama at elevations between 300 and 350 m some 44
km inland from the river mouth (but see Pozorski and Pozorski [1981] for an alternative interpretation). The preserved La Cumbre channel begins about 17 km down the Sausal course at an
elevation of about 250 m (Kus 1972; Pozorski and Pozorski 1981) (Figure 1). From this point the
length of the channel is about 54 km to its juncture with the Moche Valley Vichansao Canal at an
elevation of c. 125 m, 3.1 km north of Chan Chan. From the canal junction the Chicama flow was
then to be channeled northwest to irrigate the plains of Esperanza, Rio Seco, and Huanchaco.
Chicama waters would have traversed upwards of 84 km to reach fields on the latter plain. From
the Sausal inlet to the outlet junction with the Vichansao the straight line distance is 42 km, but
the distance traveled by the intervalley course is twice this figure due to the extreme degree of
contouring necessary to negotiate broken topography. If the purpose of La Cumbre Canal was
simply to irrigate the plains west of Chan Chan it could have followed a shorter, more direct
course from the intervalley divide. However, Chimu strategy called for a substantially larger
labor investment and the creation of a switchback in the flow direction of the Chicama waters at
the head of Pampa Esperanza. This apparently was done to insure adequate irrigation on Pampa
Esperanza and thereby charge the aquifers that fed groundwater wells within Chan Chan and
supplied the city with potable water (Day 1974).
The pass between the Chicama and Moche valleys has an elevation of c. 230 m, and from the
canal intake up to this point, the critical engineering concern centered upon maintaining the channel at sufficient height to cross the divide. Conserving channel altitude required contouring the
course through high, very broken terrain. Once across the divide, maintaining high elevation was
not a paramount design factor, and the course was cut through lower, relatively unbroken terrain. Formed by wide alluvial fans, the intervalley pass is an area of low relief. The actual change
from broken to unbroken terrain occurred 7.5 km to the north at Quebrada del Oso, where the
canal leaves a long section of difficult topography. This point is also where completed construction of the La Cumbre ends.
On the southern flank of the Chicama Valley the canal passes through alternate areas of desert
pavement, stabilized sand drifts, and scattered foothills in the vicinity of Cerro Gasnape (Figure
1). Between La Cumbre and the modern Sausal Canal there are remnants of poorly preserved

REPORTS

577

Prehispanic canals which have received neither systematic survey nor excavation. Several can
be followed intermittently up to the Oso drainage and represent either precursors or abortive
channels of La Cumbre, while other channel remnants are probably early canals that irrigated
the south side of the Chicama. Along the flanks of the Chicama Valley where La Cumbre crosses
plains of porous, unconsolidated sediments there is often no masonry lining of the earthbank
channel. Judging from similar Chimu channels that functioned in the Moche Valley, the design
philosophy called for silt to accumulate during use to form nested layers of low porosity material
that would seal the canal bed. Continuous masonry channel lining begins along the western flanks
of Cerro Tres Cruces, where the canal overrides long stretches of bedrock as it turns out of the
Chicama drainage and heads for Quebrada del Oso. The massive aqueduct crossing this dry river
is now destroyed. Incomplete construction begins on the south bank, and a number of partially
trenched courses appear near the intervalley divide, reflecting alternative strategies of canal
placement.
Where masonry lining was employed there are often numerous pits downslope from the canal
where mining went on for suitably sized cobbles to face the channel. At several such stations,
piles of stone sorted for size are present, reflecting specialized construction that required certain
dimensional limits of materials, as will be discussed.
No fields are present along the uncompleted segment of the course. However, some field systems
were laid out adjacent to the completed section (Pozorski and Pozorski 1981). The largest complex
occurs on the plains immediately north of Quebrada del Oso and is associated with a Chimu administrative center (Kus 1972; Keatinge 1974). Other field areas are located in the regions of
Sausal, Cerro Lescano, and on the western flank of the Tres Cruces hill chain. In overview, it
seems that systematic layout of fields did not occur until the canal was near completion.
Quebrada Del Oso
The two largest drainages crossed by La Cumbre Canal north of the intervalley divide are
Quebrada del Oso and Quebrada Huascar (Figure 2). They bracket the Cerro Tres Cruces bedrock
hill chain, which is the most rugged topography negotiated by the canal. The Huascar drainage is
the older of the two quebradas and formerly incorporated the Oso Basin. Quebrada del Oso was
formed by stream capture along a major fault transecting the Tres Cruces block and the Huascar
Basin. A narrow, rock wall canyon breaches the hill chain. Along its course a 40-m "waterfall,"
representing either a bedrock structure or a secondary, perpendicular fault, divides the stream
between an upper-pirated basin and a lower outwash surface. For the stream capture to be sustained, any significant vertical displacement along the Oso fault requires relatively greater rise
along the north side of the canyon and/or proportionately greater subsidence along the south side.
The opposite pattern of relative movement would act to turn the stream back to its original
Huascar drainage. Stream capture began after formation of the highest two erosional terraces in
the upper Huascar Basin, but before formation of the lowest terrace and active floodplain. In the
outwash section of the Oso drainage, lithic stage occupation has not been found, but there are
scattered remains dating back to the first millennium B.C.
Approximately 2 km upstream from the Oso canyon, La Cumbre changes from a canal cut
through the sedimentary flanks of the Chicama Valley to an artificially elevated channel running
across bedrock hills with slopes often exceeding 600. Crossing the fractured stone face of the
Tres Cruces range ranks as one of the most prodigious large-scale construction projects-entailing precise survey and engineering-yet identified in the aboriginal New World. The channel is
supported by terrace-aqueduct structures standing upwards of 10 stories high (30 + m) that are
banked against the mountain slopes. The core of these structures is comprised of soil and cobbles
transported from the plains below, as well as angular fire-quarried rock from the mountain face
along the course. Wider at the base than summit, the structure exteriors are built up in masonryfaced terraces 1 to 2 m high.
Where the channel shifts from sediment to bedrock there is a shift from a downhill slope to a
neutral or slightly uphill inclination, which continues for about 0.5 km up to the largest bedrock

578

AMERICANANTIQUITY

[Vol.47, No. 3, 1982]

quebrada negotiated by the canal. Crossing the quebrada, the slope changes uphill to c. - 0.005 for
the last 1.5 km up to the Oso quebrada. This length of uphill slope channel is clearly bracketed by
bedrock faults. It is possible to profile both the last masonry channel and two underlying prior

4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...
. .........
*...1..e,,....,..t.'
.:.
.4.
:.
il',

Figure 2. Vertical aerial photograph of the zone around Quebrada del Oso. (1) Limitsof modern cultivation, (2) the Intervalley Canal (path marked by black diamonds);(3) lower Quebrada del Oso, (4) Oso canyon; (5) upper Oso basin, (6) Cerro Tres Cruces, (7) Quebrada Huascar. (InstitutoGeografico Militar,Lima:
Project AF 60-17, negative 1726).

REPORTS

579

courses over the length of this interquebrada span: essentially, the three superimposed channels
are parallel. The most economical explanation of the trichannel profile is that it records the effects of gradual, increasing north-to-south vertical displacement of the Tres Cruces bedrock block
which generated the Quebrada del Oso stream capture. In other words, there is a coincidence of
fault boundaries, direction of block movement, drainage pattern, and canal-slope distortion that
are mutually self-predicating in the context of ongoing geodynamics.
ENGINEERINGANALYSIS
Preservation patterning has important implications for interpretation of the archaeological
record, especially as it applies to engineering analyses. Anyone failing to take into consideration
the effects of preservation patterning can arrive at erroneous conclusions. We know that sections
of the Intervalley Canal now run uphill-the question is why. Were they built that way or has
their slope been altered by factors such as tectonic tilt, seismic compaction, and movement of the
unconsolidated sediments under the channel?
We have evidence, both prehistoric and modern, of tectonically induced slope changes equal to
or greater than those postulated for the Intervalley Canal, so tectonic distortion is a plausible explanation for the apparent slope errors. This explanation receives additional support when we
look at the channel itself. The clear evidence presented below of manipulation of the channel
geometry (also seen in the Pampa Huanchaco system [Ortloff et al. 1981]) shows sophisticated
knowledge of open channel flow, knowledge which is at odds with an interpretation that the apparent slope errors were due to sloppy engineering.
On the basis of both internal and external evidence, we feel entirely justified in extrapolating
the overall canal slope to that of the study section north of Quebrada del Oso, and then proceeding
with a detailed and sophisticated analysis. It should be remembered that the discussion that
follows is based on this estimated slope, but it also should be noted that the specific conclusions
are valid over a wide range of channel slopes and that the method of flow simulation has universal applicability.
Field Procedures and Methodology
Hydraulic engineering aspects of the Intervalley channel were studied for the zone immediately
north of Quebrada del Oso in order to determine the level of open channel flow design sophistication employed by Chimu engineers. The present analysis focuses exclusively on the hydraulic
characteristics of the uppermost late channel as it existed just prior to abandonment. In the
Quebrada del Oso section the masonry channel is well-preserved, and considerable data can be
collected without excavation. During survey, however, loose sand was removed to permit
measurement of canal surface geometry down to the most recent bottom. Profiles were subsequently excavated to determine earlier channel geometries, but these were not used for the present analysis.
Survey and recording began at the northern extant terminus of the now-destroyed aqueduct
(denoted as L = 0, Figure 1) crossing the quebrada. Proceeding in an upstream direction, we
recorded readings of channel bottom width (B), wall slope (Z), and Manning Roughness Factor (n)
at approximately 50-m intervals (Table 1; see Glossary for definitions of terms). Where special
features occurred (overflow chutes, wall roughness changes, channel geometry changes, etc.), the
50-m measurement interval was reduced and details of the feature recorded. With these same
procedures, a 1,606-m length of the Intervalley channel was surveyed upstream of the reference
station.
Since channel slope in the 1,606-m section has been altered by tectonic effects, an estimate of
slope at the time of construction is necessary in order to proceed with the analysis. Such an
estimate, when combined with Table 1 data, provides data for hydraulic calculations necessary to
assess the function of the channel cross section shaping associated with the 1,606-m canal length.
To this end, the present-day local slope is recorded at 50-m intervals and at feature locations. Us-

580

AMERICANANTIQUITY

[Vol.47, No. 3,19821

Table 1. Sample Intervalley Canal Parameters.

Station (m)
0 to 20 m
33
40
120
147
170
220
226
258
276
306
356
406
456
506
556
606
656
706
756
792
806
856
906
956
1006
1106
1156
1206
1256
1306
1356
1406
1456
1606

ih*

0.026
0.009
0.009
0.017
0.017
0.015
0.015
0.015
0.012
0.009
0.009
0.009
0.015
0.015
0.015
0.015
0.015
0.015
0.015
0.015
0.015
0.015
0.015
0.015
0.017
0.009
0.009
0.017
0.017
0.017
0.017
0.017
0.017
0.017
0.012

2
2
2
1.82
1.82
0
0
0
2.014
2.014
2.014
0.58
0.58
0.58
0.29
0.29
0
0
0
0
0.5
2
2
2
1
1
1
1
1
1
0
1
0
0
1

n
0.027
0.028
0.027
0.036
0.027
0.022
0.023
0.025
0.032
0.027
0.025
0.026
0.025
0.025
0.022
0.025
0.025
0.025
0.025
0.026
0.031
0.032
0.032
0.032
0.032
0.032
0.032
0.025
0.031
0.022
0.028
0.028
0.024
0.026
0.029

B (m)
3.603
2.201
3.658
2.896
3.962
3.962
4.267
1.524
2.643
3.048
3.048
5.791
5.791
5.791
7.925
7.925
7.925
7.925
7.925
7.925
4.572
0.506
0.506
0.506
1.219
0.914
0.457
1.372
0.914
2.438
2.134
2.438
2.438
4.267
2.438

Dn (m)

DC (m)

1.253

1.861

0.610
0.541
0.412
0.413

0.689
0.630
0.389
0.389

0.344

0.389

0.288
0.459
0.930

0.316
0.466
0.755

0.896

1.303

0.875

0.893

11.143
I0.856
0.954

0.981
0.981
0.981

0.498
(0.707
10.707
0.629

0.774
0.702
0.702
0.656

*Based on assumptions about design intent explained in "Field Procedures" section.

ing a range of assumed slopes, computer calculations of the hydraulic flows through the 1,606-m
canal length reveal the basic design intent of canal cross-sectional shaping. By next subtracting
the most representative theoretical slope from the present-day slope, we achieve an estimate of
the total angular distortion of this region over an 800-year time span. The net estimated local
slopes at the time of construction are then listed in Table 1. For an ancient theoretical slope of
-0.013 and for an average present-day slope of - -0.006 (Pozorski and Pozorski 1981) we see
that an approximate net slope change of - -0.019 (1.09?) has occurred over the 800-year interval. The -0.013 ancient slope value is close to that for critical flow in the channel; this value
represents the maximum possible flow rate through the channel, and as such forms an upper
bound to its flow rate. Additionally, the supply flow rate computed for the Intervalley Canal is approximately that of its intended destination canal system-the Esperanza and Huanchaco canals.
The computed theoretical Intervalley Canal flow rate is 4.67 m3/sec at the Oso station; the independently computed total flow rate to the coeval Esperanza and Huanchaco canals is 4.45
m3/sec (Ortloff 1981: Ortloff et al. 1981). The theoretical supply flow rate from the Intervalley
Canal under the assumption of critical flow is then very close to that of the canals intended to be
fed by the Intervalley Canal.
Two aspects of the engineering work behind the Intervalley Canal must be considered in the
analysis: first, the theoretical design intent in initial route survey and channel configuration and

REPORTS

581

second, the end product characterized by adjustments, reappraisals, and compromises of the
ideal that were necessary in order to fit the constraints of unexpected field problems.In the sections that follow, the design intent of the Intervalley is first given according to the signs and
signals in these works, which are recognizable to hydraulics engineers. A section is then
presented on the proximityto originalintent realized in the final construction.The design intent is
gauged by use of Table 1 results; these results contain several extrapolations and interpolations
of parameters across destroyed/erodedcanal sections. Solutions to the equations for fluid flow
through the channel given by Table 1 then reveals the design intent of the channel shape
throughoutthe 1,606-m stretch of the canal.
DESIGNINTENT
Basic Hydraulic Concepts

Fluid flow througha canal is governed by a large numberof parameters (ib,n, B, and Z). Additionally, the flow classification (subcritical,critical, and supercritical)tells whether a decrease in
canal cross-sectional area has a large upstream influence (subcritical flow) or a localized disturbance effect (usuallyin the form of a hydraulicjumpfor a supercritical flow). For example, a subcritical flow entering a local channel constriction(choke)would result in a water height change
that may extend for many kilometers upstream. For purposes of determining whether supercritical or subcritical flow exists in a channel, it is necessary to compute the theoretical normal
depth (Dn)and the critical depth (Dc)based on channel cross-section geometry,n and ib (Henderson 1966). If Dn> Dc, then subcritical flow exists; if Dn< Dc. then supercritical flow exists. For
subcritical flow the Froude number (Fr) is less than unity; for supercritical flow, Fr> 1. The
special case of Fr = 1 is denoted as critical flow (for which Dn = Dc also). The consequences of
maintainingFr close to unity result in a maximumflow rate for a given inlet-to-outletheight difference for a fixed channel area. The degree to which Chimuengineers consciously manipulated
all of the above-mentionedparameters and streamwise effects is then the basis for a large part of
the following discussion.
Calculation of Channel Flow

In technical terms, subcritical chokes are said to exist when the channel constriction has a
critical (Fr = 1) flow at the throat and an upstream subcritical (Fr < 1) flow. Calculations,based
on channel geometry field data (Table 1), of the effects of a choke must include upstream adjustment of the estimated water level by standard hydraulics calculation methods (Chow 1959;
Henderson 1966). Chokeswith upstream supercritical flows result in hydraulic jumpsstandingin
front of the choke. Calculation of theoretical water heights in this case was accomplished by
momentumand energy matching of the upstream and downstreamflow from the hydraulicjump.
From ib, B, and Z then, an averaging-marchingcalculation procedure (Chow 1959; Henderson
1966; Morris and Wiggert 1972) based on simultaneouscomputersolution of continuity,momentum, energy, and Manningequationswas initiated for streamwise water height and mean velocity
solutions. Beginningwith an initial estimate of flow rate (Q),we calculated the water height in the
1,606-m section of the canal. At sections where calculated water height exceeded canal wall
height, another lower Q estimate was made and the calculations restarted until the water height
was contained by the canal. In this manner the correct flow rate throughthe Intervalley Canal
was determined. Fromthis theoretical calculation the fluid mechanics of water flowing through
the ancient canal can once again be visualized. Further, by examination of the details of the
nonuniformflow throughthe many changingcanal cross sections, the originaldesign intent of the
Chimuengineers is revealed.
Due to large reaches of supercritical flow, upstream influence from these sections is nonexistent, so that selected portions of the canal flow may be calculated independently.Results of these
calculations are shown in Figure 3. In Figure 3a the bottomwidth (B)of the idealized trapezoidal
cross section is shown as a function of upstream length from the L = 0 reference point. Wide

582

AMERICAN ANTIQUITY

[Vol. 47, No. 3, 1982]

1524Mean slope
50 km length

1371
D

Mv

Hydraulic

jump

Suprc r it
choke

'b ,c
~~~~~~~~~~~~~~~~~~

1220 -

point

Dn>Dc
Subcritical

Dn<Dc
Supercritical

xz

1067 (overf low


c h u t e )~

^eD

914

z
0

C ri ti calI
seCtionl
Dec

?-

Critical

flow

Dc
~~~~~~~~~~Dn

I-

762

LU

iU

Flow

Direction

UJ 610

457

z -I

305~~~~~~~~~~~~~~~~~
305

Hydraulic
jump
D

152
Subcritical
choke

~~,0

I
3

.I

1.5 0 1.5 3m
Canal Width
(a)

Figure 3.

0 1 2 3
Froude
Number
(b)

0 0.01
ib
(c)

Intervalley Canal parameters in the study section north of Quebrada del Oso.

REPORTS

583

variations in Z ib, and n occur concurrent with large B changes. For example, at the 762-m
marker a steep-sided, narrow base (B = 0.42 m), deep trapezoidal section opens out to a shallow
trapezoid of wide base (B = 7.32 m). A cycle of trapezoidal base expansions and contractions appears evident from Figure 3a.
Calculations of water depths, mean velocity, normal and critical depth, local critical slope
(ib,c)' and Froude number reveal a conscious design objective of maintaining canal Froude
number close to unity (Figure 3b), thus producing an average critical flow in the canal. For a given
initial potential energy head represented by the absolute inlet height, the maximum flow rate
through the canal is achieved by maintaining a Froude number of 1 throughout. Thus, it appears
that the Intervalley system is carefully tailored to transport a known flow rate in such a manner
as to make that flow rate the maximum for the available potential energy. Further, by a series of
subcritical and supercritical chokes upstream of the Quebrada del Oso aqueduct, the Froude
number is continually varied about unity by changes in n, ib, Z, and the cross section so as to avoid
the flow instabilities associated with small disturbances at Fr = 1. For supercritical
Fr > 1 flows, a combination of n, ib, and cross-section geometry changes modulate the Froude
number by producing weak hydraulic jumps at stations 1219, 1036, and 306 to reduce the flow to
subcritical. In these locations the upstream Froude number never exceeds 2, so that total head
loss through the jump is small. For subcritical flows, the Froude number is increased by means of
slope and channel geometry changes (stations 1189 and 1036, for example). In the approach to the
now-destroyed aqueduct crossing Quebrada del Oso, a supercritical choke is employed to create a
large hydraulic jump at station 309 so that the flow crossing the large aqueduct is subcritical.
From a design point of view, the presence of a slow velocity flow across an aqueduct is preferable, as the aqueduct is made from unconsolidated fill material subject to wall erosion. The lowspeed flow then limits erosion of the canal walls and permits a narrow sidewall embankment to
contain the flow, conserving valuable labor by keeping the aqueduct width small. This type of
structural optimization by sophisticated open channel flow controls is entirely consistent with the
basic strategy of channel design for a maximum transport rate and points to a level of technical
sophistication previously unknown in the history of hydraulics among Precolumbian cultures.
It is interesting to note the effects of the canal cross-sectional shaping for slopes yielding noncritical flows. If the canal slope was higher than that assumed, then the choke systems at the
aqueduct work even more effectively. That is, a stronger hydraulic jump is formed and the flow
velocity downstream of the jump is further reduced, thus limiting flow-induced erosion even more.
Throughout the system the flow will tend more toward supercritical values, but with approximately the same flow rate as for critical flow. For shallower slopes, the choke section before the
main aqueduct again works to produce the same effect, but in a somewhat different manner. Subcritical flow passes through the choke opening, is accelerated through the throat, then increases
velocity until a weak hydraulic jump is formed in the expansion region. A low-speed flow is then
produced for passage over the aqueduct. Therefore, for both off-design cases the preaqueduct
choke system functions to produce low-speed flow through the large aqueduct. Overall, though,
the observed design is more consistent with critical flows, as determined by the computer simulations.
Computation of normal and critical depths along the canal indicates that Dn and Dc continually
interchange as maximum values despite wide variations in channel geometry (Table 1). This indicates a design strategy of achieving critical or near-critical flow throughout the channel length.
For the calculated flow rate (Q) value, the local critical slope ib, c is close to both the total length
mean slope and the local bed slope (Figure 3c), indicating that the design optimization toward unit
Froude number probably prevails throughout the entire canal length. At canal locations where
hydraulic jumps occur, the upstream Froude number is always less than 2, indicating a relatively
weak jump. From a design point of view this limits the energy lost due to turbulence and viscous
mixing, thus conserving the total head of the flow. In many places near-critical flow is achieved by
adjustments in ib, n, Z, and B, indicating that the designers understood the complex interrelationships of these parameters in controlling flow velocity.

584

ANTIQUITY
AMERICAN

[Vol.47, No. 3,1982]

Located at stations 1006 and 945 are overflow chutes near the top rim of the canal bank that
would limit flow to the maximum Q value of the canal. Any excess above the design Q is drained
off by these chutes. The chute at station 945 leads away from a partially destroyed section of the
canal with enough structure remaining to indicate that the chute ran from near the top rim of a
basin. This again indicates that only when the canal was over the design flow rate Q did the
chutes function.
It may be concluded that the Intervalley was purposefully designed to carry a specific Q value;
further, the system was designed to carry this Q optimally. Local variations in canal slope due to
unavoidable topographic features (which affect flow velocity) were countered in a manner indicating cognizance of the complex interrelationship between local channel geometry, roughness,
and bed and sidewall slope. The streamwise effects of subcritical and supercritical choking and
of velocity modulation by channel width changes appear to be well understood. Survey during the
1978 field season has led to the discovery of yet another choked-flow canal geometry preceding
an aqueduct (Figure 4). This system is located at point M in Figure 1, where an overflow chute is
also present.
It is interesting to note that destroyed canal sections (labeled D in Figure 3a) occur at locations
of the calculated hydraulic jumps (except for station 762). In the event of a massive rain, channel
flow could exceed the capacity of the overflow chutes to regulate water level, thereby creating
large Fr increases; the possibility of local overflow at hydraulic jump locations then is a distinctly
possible mode of canal bank failure at D locations.
Examination of the Manning Roughness Coefficient (n) as a function of canal length (Figure 5)
reveals local adjustments. Large variations appear in the vicinity of hydraulic jumps associated
with choke locations. Since n ranges over nearly a 100% change, it can be argued that wall roughness may have been used as an adjustment control on local Fr values, thus implying a knowledge
of this effect on flow velocity regulation.
In many places a lined trapezoidal cross-section channel is found with wide top walls, upon
which secondary walls are constructed. Calculations reveal that the topmost set of walls are not
interactive with the flow at any station. The use of such constructions is undoubtedly for the puron steep-sided aqueduct and terrace strucpose of walkways alongside the canal-especially
tures.
The total inventory of canal geometries available to Chimu engineers appear to range from
stone-lined, variable B trapezoids, with flush-mounted stones set in plaster, to unlined versions
with large Z and B variations. Sidewall erosion of the unlined shapes gradually leads to nearparabolic equilibrium cross sections.
Local theoretical flow velocities fall, for the most part, in the range of V = 1.67 to 3.05 m/sec.
After long-term stabilization of the canal bed in unlined sections, and given that no input material
is eroded from lined sections, most of the silt carried in the initial canal length is in the form of
suspended load. After transit of a sufficiently long distance it is likely that most of the suspended
load has settled onto the canal bed, whereupon bed load transport is the dominant means of sediment movement. That silt would not remain in suspension can be seen by a simple calculation. To
maintain sediment in suspension, a flow with a particle Reynolds number (Henderson 1966:412)
equal to -400, i.e., Re* = V* D80/v 400, is necessary. A calculation of Re* for the Intervalley
Canal reveals a value an order of magnitude less than 400 (based on a measured silt effective particle diameter D80 of 0.15 mm) indicating that a stable bed existed and no scouring occurred.
Thus, once settling had taken place the silt remained in the form of bed load. In the above expression, V* is the friction velocity and v the water kinematic viscosity.
Canal Use Strategies
The computed maximum flow rate (Q) for the Intervalley Canal is 4.67 m3/sec. Figure 6 shows
the hydrographs of the Moche and Chicama rivers, i.e., the seasonal river flow rate as a function
of month. The stippled area on the hydrograph represents the Intervalley flow rate subtracted
from the Chicama and added to the Moche Valley water supply. The first noticeable effect of

585

REPORTS

design intent upon the delivery of Intervalley waters to the Moche Valley is an extension of the
watering season in the vicinity of Chan Chan by about 3 months' time. If, for example, in midDecember, when the Moche flow rate is on the order of 7 m3/sec, adding the full Intervalleyflow

60
to aquaduct
FF

E
so

Fr<1

40

on
2

Hydraulic
jump

C overflow

30

Section
Section
AA
BB
CC
DD

Wi

(/)
Z

3:r.
o

EE

20

weir
Properties
eL
20.6
22.7
overflow
45
45
-4 5

FF

OL

Left

eR

Right

eR
28.7
24.7
weir
60
60

0.38m

from

--45
wall
wall

slope
slope

10

Choke
Supercritical
Ahead of Aquaduct
At Station
M (fig 1)

Fr> 1

0
B

Figure 4.

Base

width (im)

Supercritical choke at Station M.

bottom

586

AMERICAN ANTIQUITY

[Vol. 47, No. 3,1982]

1524

1371
Hydraulic
Jump
Location

1220

Hydraulic

Jump

Location

914
E

0
I

762

z
cr
w

457

Hydraulic Jump
Location

305

152

.02
.04
.03
.01
LOCAL BED SLOPE ib

Figure 5.

.02
.025
.03
MANNING
ROUGHNESS

.035
FACTOR

Canal bed slope (estimated) and Manning Roughness Factor.

REPORTS

587

effectively gives a water yield of 11.67 m3/sec to the Moche Valley. This flow rate would only be
available to the Moche Valley from its waters alone in mid-January,so any agriculturalplanting
done at that time could be advanced 1 month ahead with the inclusion of Intervalley waters.
Similarly,the water available from the Moche River alone in mid-Mayis nearly doubledby addi100

90

80
70
60
50//

Chicama

\\-

Flow

40
Chicama minus
Intervalley

30

// .&.

0 20
201

\\

\\Moche

Flow

plus
Intervalley

I5w...

Flow

B10

..... /.1\

o-J 9,
8

..g.
H \......
.\ . 1t~~~~~~~~
.......
^
_
_--------@

CC
1= 616_w

~~~~~~~~~~~~~~~~~~~~.........
.... ....

~~.......... _
. ........

....

OON.......
N D J

L_....... ...... . .......


_.._w
.......

... ...

.....

M A

MJ

..
........

..........
=w-

e.._..,
s.............
X . _.... _........

................,
.............
............_
,

Figure6... Hdograh
of
flow._sof the Ch-ama and Moche rivers
1.....
m
S
r t
(f..... ............,
.......ep
, the
~~~~~~~~~~~~~~~~~~~~.......
......
..........._
,
........ following
.........,,.....,........
................._
November
year),
showing ......
the intendedimpactofthe
I......
2 . of.the
...
s,
... ............C........
...... ..
.......
.........................................
,......
,........
,,
.~~~~~~~~~~~~.....
......................
~~~~~~~~~~~~~~~~~~......
.....
~~~~~~~~~~~~~~~~~.........
.....

.........
.........

,..................
.... .. ,,,...............

,,
.....
Figure..
.........

...D .

...

6....
Hydogap
of.... the
foown

of

th
yer)

flw

F
of

th

sho....te.nted.......ofth

.........
M...........A.........N

.....

Chicma....Mcherives..or...mot.........r

t
Interva.............ey..

Caa................

588

AMERICAN
ANTIQUITY

[Vol.47, No. 3,1982]

tion of Intervalley waters, to that equal to the maximum available from the Moche River alone at
the end of the watering season in mid-April. If all the Chicama waters are taken in June, then at
least the water flow rate obtained from the Moche River alone in mid-May can be sustained.
Therefore the agricultural production rate in mid-May from Moche waters alone can be sustained
at the same rate from Intervalley waters for 2 months longer (during this time the Moche River
flow rate drops off to levels insufficient for valley agriculture). Similarly, the mid-January Moche
River flow rate can be matched by the mid-December rate with the addition of Intervalley waters.
At the height of the Moche watering season in March there is a 15% flow rate increase available to the Moche Valley from Intervalley waters; this undoubtedly was to be used in ambitious
plans for field expansion to the Rio Seco, Esperanza, and Pampa Huanchaco areas. Calculations
based on Huanchaco Canal excavations have indicated that the theoretical flow-rate from the Intervalley is equal to the Huanchaco Canal input flow rate, thereby indicating a possible destination for Intervalley waters (Ortloff 1981).
The hydrographs shown in Figure 6 represent recent 40-year averages of maximum and
minimum flow rate values; over this time period three standard deviation departures from the
mean occur in 2.5% of the years represented, as a result of El Niniocoastal rains. For these years
each valley was subject to flooding conditions, and the Intervalley subject to severe erosion from
active quebradas flowing transverse to its course.
The Intervalley appears to have been designed with several options in mind. In addition to its
design capability to extend the watering season in the Moche Valley during low water times, the
Intervalley had the capability to take up to 6% of the Chicama flow during peak flow months. As
can be seen from Figure 6, the maximum Chicama flow rate can approach 80 m3/sec from March
to April. Unfortunately, the inlet of the Intervalley is now destroyed so that it is not possible to
determine if the same constant flow rate existed in the Intervalley Canal regardless of the flow
rate in the Chicama River. The Chimu nevertheless had the option to extract 6% of the Chicama
River flow when it probably could not have effectively been used for existing Chicama field
systems (in March) but could (in part) effectively supply the Huanchaco Canal system. Since the
Huanchaco field system was the intended benefactor of Intervalley (plus Vichansao) flow it would
be most beneficial only if the system were fully utilized. That is, if the strategy of Intervalley
operation was usage only during peak Chicama flow rate months, then only a month or two of use
would be expected for the Huanchaco system. If (assuming 100% delivery efficiency) the maximum flow rate of 4.67 m3/sec flowed through the Intervalley from mid-October to mid-June then a
full nine months' usage in the Moche Valley could be obtained from the Intervalley alone.
Excavations of the Huanchaco Canal (Ortloff 1981: Ortloff et al. 1981) reveal that parts of the
late system were never operational. This could indicate several possibilities: (1) the Intervalley
contribution was insufficient, requiring a contraction or abandonment of part of the Huanchaco
canal network and further reliance upon an expanded Vichansao contribution; (2) long-term
drought conditions prevailed and reduced Intervalley flow requiring a less extensive Huanchaco
canal system; (3) outside invasion or political changes drew attention away from canal construction and maintenance. Option (1) appears to be the most likely. As will be discussed subsequently,
the Intervalley probably never provided any water to the Moche Valley. This failure resulted in
the need for an amplified Vichansao Canal to supply the Huanchaco and Esperanza canals. Most
likely, the plans for Huanchaco expansion were contingent upon Intervalley waters supplementsegments built for this purpose therefore were never utilized. Furing Vichansao waters-canal
ther evidence for the hypothesis of dysfunction of the terminal stages of the Intervalley can be
seen in the incomplete G-O system (Figure 1) near the divide, as well as the unused field systems
north of the Quebrada del Oso.
One of the main reasons for failure of the Intervalley Canal appears to be tectonic warping of
the channel slope. Prime evidence of the effects of tectonically induced coastal uplift are seen in
canal segments north of Quebrada del Oso (Figure 1). South of this region, however, many canal
segments appear to branch away from the main stem canal. Each such canal path represents a
constant slope path across the then-present land surface; alternative canal paths adjacent to
earlier ones hence represent constant slope paths on a land surface distorted by tectonic activity.

REPORTS

589

As canal slopes are typically a fraction of a degree, a relatively small ground slope change can require the re-engineering and relocating of a canal from an original position.
The major path to the intervalley pass is uphill and remains incomplete by standards set by
upstream, fully lined canal construction. Thus, this final segment indicates the challenges of
canal engineering on a coastal plain subject to tectonic effects throughout the many years required for canal construction. In final analysis, the Intervalley was ultimately abandoned due to
difficulties with the southern portion. Although the Intervalley may have functioned off-design
before tectonic effects caused its abandonment, the input to the Huanchaco system was nevertheless negligible at best. This led to the expansion of the Vichansao as the main supply canal for the
Huanchaco system. Further evidence supporting option (1) is given in the next section.
Of the remaining options, (3) is contrary to the many years of stable Chimu coastal occupation
and successful military expansion. Option (2) is more difficult to test, as evidence for decreasing
water supply to the Rio Seco and Huanchaco systems over time can be given (Ortloff et al. 1981).
This too is attributable to tectonically induced canal slope changes and the concomitant water
flow rate decrease within the Moche Valley (Moseley 1981a).
By design, Intervalley waters would have certainly guaranteed ample water for the gardens
and wells of the elite class at Chan Chan from October to June. This may reflect the desire to keep
the capital city flourishing at the expense of valleys of lesser political importance. The potential
capability for Chan Chan's agricultural survival under adverse climate conditions, as well as extension of the Moche Valley watering season, point to the dominance of Chan Chan in dictating
policy to adjacent valleys; the Intervalley systems, by design, could have been used flexibly to implement the luxury status of Chan Chan's rulers as well as to guarantee continuance of this status
throughout dry years by literally draining the Chicama dry.
Engineering Reality
Sighting from the northern dune region toward the Intervalley pass reveals a line-of-sight slope
of - 0.007. The actual slope of this segment along the canal is, of course, less than the straight line
slope. Since there was little margin for error over difficult terrain, and considerable terracing
and aqueduct construction over the 20 km distance north of the Quebrada del Oso region were
necessary, it is foreseeable that considerable iteration and correction would characterize this
path. Examination of exposed canal profiles indicates a general trend toward decreasing the
local slope of the canal in response to the need for less elevation loss with distance. Since there is
sparse evidence for canal usage, such slope corrections are most likely a response to observations
of tectonic activity in this area. Major fault lines are located throughout this region and to the
south near Cerro Campana. Evidence of tectonic effects on canal systems in the immediate vicinity is shown by canal segment u-w (Figure 1). This segment shows extensive usage in the form of
heavily oxidized silt layers-yet its present slope is in the uphill direction! Overlying this segment
is a late canal of downhill slope, but indicating no usage. Obviously, significant tectonic effects
have occurred within the time frame of the building of these canals (Moche V to Chimu-Inca). The
fact that segment C-E of the Intervalley Canal runs uphill by current survey is indicative of the
groundslope movement in the region south of Quebrada del Oso.
To assess the relative age of the Intervalley Canal at and north of Quebrada del Oso, a series of
nine C-14 samples of associated organic matter were processed by the Desert Research Institute
of the University of Nevada. The organic material is associated with early, middle, and late canal
phases and, as such, bounds the earliest and latest times associated with canal building. Results
indicate dates from A.D. 960 to 1420 (minus and plus la) with a mean date of A.D. 1170. These
dates are compatible with Kus's findings for samples of organic matter taken in the region of
Cerro Sausal farther upstream along the Intervalley Canal (Kus 1972, 1974). The downstream
reaches of the canal indicate generally later dates, indicative of the many years of rework and
reconstruction associated with the southern tectonically more active reaches of the Intervalley
Canal. In the vicinity of Quebrada del Oso, for example, the height difference between lowest and

590

[Vol.47, No. 3,1982]

AMERICANANTIQUITY

<

E~~~~~
0~~~~~~~~~~~~~~~~~

-o
o

t=

--

~~ ~ ~ ~ ~
0~~~~~~~~~~~~~~~0

CL

7C

~~~~~~~~~~CD

I,

./
-

lX,ZCD

l_

4'

w_

II

C/

I,

~~~~~~~~CD
CD

CL~~~~~~~~~~~~~~~C
*

REPORTS

591

highest phase canals is on the order of 8 m, attesting to the magitudeof correctionrequiredby tectonically induced elevation changes in the southern reaches of the canal.
The Pampa Cabezonregion (Figures 1, 7) is laced with sequential canal paths indicative of the
response to groundslope changes in this southern zone. The fact that canal paths are abandoned
for other paths after partial construction (without the benefit of a trial flow through them)
represents the difficult process of engineering low slope canals througha geologically unstable
area; the fact that hundredsof years may have transpiredduringthe buildingprocess without apparent successful canal operation attests to the formidableproblemfacing Chimuengineers and
their determinationto obtain a solution.
Various strategies are represented amongthe canal paths of PampaCabezon.The easternmost
path would require a cut through ridge H-2 while the westernmost route would require alluvial
soil removal and low aqueducting.Apparently,while one path was in the process of construction,
resurvey revealed the necessity for a new canal path requiring considerable rework of the
upstream canal reaches, as evidenced by the many different elevation phases. With reference to
Figure 6 it appears that a straight line from A to B has been surveyed from the peak of H-2 to far
distant upstream and downstreampoints. The straight line path runs about 20 km and is meant to
complete the approach to the intervalley divide. The region near the low ridge interruptingthe
A-Bpath is instructive as to the methodsused to survey canals. Since a straight line appears to be
surveyed through the hill it may be that large canal segments are laid out for the "feasibility"
survey. The topographicdetails are then examined for local corrections to maintainthe slope. Options 1, 2, and 3 represent paths reflecting the difficulty of approach to the upliftingdivide. Path
D, in its approach to Quebradadel Oso, is some 12.3 m higher than any correspondingnorthside
canal segment;further, south of hill H-2 this segment runs uphill. Paths C and D then run through
a bowl-shapeddepression centered about the ridge, with each end of these canals at higher elevations. Clearlypaths C and D, once considered viable optionson a differently contouredlandscape,
are presently unworkable.Path A-Bis presently downhill and connects onto the last phase of the
canal on the north side of Quebrada del Oso. This southside path retains the same slope as the
average upstream canal segments and is obviouslythe last of the attemptsto cross the intervalley
divide. Near point A', and for all downstream reaches to the divide, the canal goes uphill and is
therefore unworkable.The progression from paths C and D to A represents successively greater
penetration in the southward direction of workable canal segments; however, none appear to
have successfully negotiated the intervalley divide. It appears then that the system was abandoned and the Intervalley never reached its design potential.
Due to the large number of slope and cross-section geometry changes throughoutthe canal
length, obvious difficulties would be encounteredin crossing the pass with slope sufficient to permit near-critical flow. As the slope decreases in the southern reaches of the canal, the tendency
toward subcritical flow increases with large attendant upstream influences, and canal flow rate
decreases if this influence extends to the inlet point. The fact that the canal cross-sectional area
and hydraulic radius may increase while flow rate remains constant indicates that the flow may
be critical when the slope becomes shallower. This optionexercised upon the A'-Bsegmentwould
permit critical flow to exist to the Pampa Cabezonregion, but further extension would require
massive trenchingand widening up to the divide. That such an optionwas never exercised attests
to the labor investmentinvolved-clearly, trenchingto a depth of 15 m for 10 kmwould provide a
workable canal (for a while). The prospect of yet-to-comere-engineeringand rework clearly gave
caution to the Chimuengineers to consider abandonmentof the total system (as indeed provedthe
case).
In retrospect it is clear that the multiplicityof canal paths and strategies in this region was undone by small ground-slopechanges. Foreknowledgeof nonfunctionby resurvey led to alternate
path selection under the constraints of slope and canal cross-sectional area necessary to produce
critical flow in the canal. Any deviation from near criticality in the canal flow would tend to
choke the entire system and render it useless with regard to levels of flow sufficient to justify the
labor expenditure involved in its construction.Had the canal been choked by a low-slope downstream stretch, the entire canal would then have operated in a subcritical mode; to obtain the
same flow rate as the critical mode, the entire system would then have to have been considerably
widened, requiringterrace and aqueduct additions. Clearly,the hydraulic difficulties associated
with the southern region near the divide are critical to the operation of the system in the design

592

AMERICANANTIQUITY

[Vol.47, No. 3, 1982]

mode. When it emerged that these problems could not be surmounted, the entire system must have
been abandoned out of necessity.
Special Features
Several engineering features appear unique to the Intervalley Canal. At the base of a
25-m-deep, 30-m-long straight aqueduct (now destroyed) located about 5 km north of Quebrada del
Oso are the vestiges of a porous culvert. Since the aqueducts block passage of quebrada floodwater they serve as inadvertent dams. The combination of water pressure and seepage into the
aqueduct fill leads to failure if water is allowed to stand behind the "dam." Through the base of
this aqueduct, size-graded large stones are piled transverse to the direction of the canal. Such a
construction would have served as a porous path to transmit water through the base of the
aqueduct. To date, this is the only such example found.
A survey of 10 aqueducts immediately north of the Quebrada del Oso zone reveals several interesting features. For aqueducts on the order of 30 m or less, a plan view indicates straight-line
construction; for aqueducts averaging 60 m, a plan view indicates curved-line construction. Curvature is toward the up-quebrada direction, similar to an arch dam. Such a configuration will add
additional strength to resist the hydrostatic pressure of water behind the aqueduct, due to the
compressive force component acting toward the aqueduct center through the fill material. Most
of the high canal terraces have extensive stepped-stone facings. In addition to maintaining slope
stability, the facings prevent the tendency of the seepage flow to wash out fine soil particles at the
point of exit from the embankment. In general, embankment slopes are on the order of 45 for terraced canals.
One further feature located near point M, Figure 1, consists of a narrow-width zone preceding
a sharp canal bend. The narrowing is believed to induce a local hydraulic jump upstream of the
bend, thereby reducing flow velocity and erosional effects in the canal bend. A bypass canal segment is found in the same region. Although this construction required massive filling of a
quebrada, nevertheless the canal was shortened by some 200 m in length.
A number of lining types have been observed along the length of the Intervalley Canal. As mentioned earlier, many canal segments have been laboriously lined, only to be filled in for new canal
phases within the same bed. Elevation of a canal is the result. Among the lining types are cobbled
walls with/without a clay-rich plaster fill for the interstices between the cobbles. An unusual
calcium sulfate lining occurs for about 2 km to the north of Quebrada del Oso. Further north in the
dune region, a segment of lining consists of large basaltic slabs tightly fitted together. Approximate face area is on the order of 1 m2 for each slab. The most common lining appears to be quarried angular cobbles stacked tightly together on the sidewalls of the canal. Stone lining appears to
run throughout the entire canal length to the north of Quebrada del Oso; to the south, lining appears intermittently. This further indicates that the process of ongoing construction on the south
side failed to achieve a final version meriting stone lining.
In terms of new technology attempted, the Intervalley appears to be unique in that terracing
and large aqueducts are employed extensively. Although Intervalley Canal slopes are approximately 100 times greater than those of the Vichansao Canal (Figure 1), small errors in slope
calculation on the Intervalley nevertheless lead to large height deficiencies. For example, an error of 0.01 in slope can lead to a height shortfall of 12 m after the 70-km length of the Intervalley
is traversed. In terms of labor investment the Intervalley Canal represents several orders of
magnitude over that of the Vichansao Canal. In terms of technical sophistication it should be
noted that the Vichansao is basically a subcritical system requiring gigantic cross-sectional area
to provide adequate water supply to the Esperanza and Huanchaco systems. The fact that the
Vichansao is subcritical and the Intervalley critical in design intent indicates that the designers
understood how to minimize labor investment and obtain maximum flow rate in return. Were the
Intervalley a subcritical system, the labor investment in construction for the same output flow
rate would have been an order of magnitude greater than that for the existing system. Here,
sophistication in hydraulic design clearly saves many valuable man-years of construction time
and also leads to maximum possible agricultural production through large supply rates of water.
Clearly, labor was not a commodity to be misspent even in ancient times.
The presence of complex choke/overflow systems on the Intervalley is without precedent

REPORTS

593

elsewhere in the Chicama-Moche Valley canal complex and represents considerable sophistication in open channel flow design. In terms of design intent it may be surmised that the Intervalley
possessed an inlet designed to always flow full, i.e., the inlet was placed deep into the Chicama
riverbank. In this manner, the design flow rate could be achieved and the sophisticated hydraulic
controls made to function on-design.
SUMMARY AND CONCLUSIONS
Since the Intervalley Canal appears to be designed with knowledge of the critical slope
necessary to transport water at a given optimal flow rate, it may be supposed that extensive
surveying and calculation preceded its design. This implies that such variables as the absolute
height from sea level of the Chicama Valley intake and the Moche Valley outlet points were
known, and that slope measurements to less than 30' (0.009) were routine along canal length.
Slope changes about the mean critical slope due to topographic variations produced the most
sophisticated hydraulic countermeasures to keep the canal flow critical. It should be remembered
that Chimu engineers could have selected any slope they wished for the Intervalley-the fact that
they appear to have chosen a slope to transport water optimally speaks of foreknowledge of this
technology. The intake point is some 44 km up the Chicama Valley-precise survey of the intake
point in difficult foothill regions and determination of its absolute height from sea level are difficult problems even with present technology.
The sum total of this hydraulic and surveying expertise points to a most advanced technological
society; indeed, the level of understanding of the complex relation between wall roughness, bed
slope, hydraulic radius, and subcriticality and supercriticality as evidenced in the Intervalley
were not known until the late nineteenth century in Europe and America (Rouse and Ince 1963;
Sprague de Camp 1974).
It appears that the Intervalley Canal was not completed at its distal end due to severe problems
of tectonic movement. At present, the southern uphill canal segment of the Intervalley stands as
testimony to ongoing coastal uplift effects. It appears that on or before A.D. 1400 the Intervalley
was abandoned without ever reaching its potential. Although parts of the Intervalley may have
briefly functioned in a marginal off-design mode, nevertheless the absence of bed silt in the southern reaches is a clear indication that water never passed through this zone.
Although the techniques used by Chimu engineers to design canals are now lost, the fact nevertheless emerges that even by twentieth-century standards the Intervalley is close to optimum
design, i.e., for transport of 4.67 m3/sec the same slope would be selected for critical flow. The
use of local ib, n, Z, and cross-section variations to regulate Froude number show an understanding of hydraulic controls. The sum total of both surveying and hydraulics expertise points to a
systematic science of observing, recording, and generalizing among Chimu engineers. Had the Intervalley been designed for subcritical flow, then the system would have had a much wider and
deeper cross-section to obtain the desired flow rate; had the canal been designed for supercritical
flow, sidewall erosion would have been a problem and channel width would have had to be large.
Also the required steep slope would have proven impractical to construct through the mountains.
Only by a critical flow design, then, can both labor and hydraulic efficiency be served.
The Chimu thus appear to have had engineering skills worthy of maintaining an advanced
agricultural system. Such skill implies a specialized branch of government devoted to collecting,
analyzing, and abstracting hydraulic observations into a theoretical base which could be used for
systems design. That the strategy of the Intervalley usage must necessarily be coupled to
engineering design implies technical liaison activities coupled with Chimu administrative functions, much the same as in modern governments. Since flexibility of use of the Intervalley is a
design feature, it is possible to envision multiple use strategies that could have been chosen
depending upon climate, workforce levels, internal and external politics, and agricultural
strategies. All of these conclusions suggest an advanced technology-based Chimu government and
shed light on the techniques and strategies used by Chimu leadership to exercise dominance over
wide areas of the north coast of Peru.
Acknowledgments. The research underlyingthis paper was supportedby National Science Foundation
grants BNS76-24538 and BNS77-24901. Assistance in the field was provided by members of the Programa
Riego Antiguo, including Genaro Barr, Alfredo Narvaez, and Fred Nials.

594

AMERICANANTIQUITY

[Vol.47, No. 3,1982]

The authors wish especially to acknowledgethe commentsand criticisms of Shelia Pozorskiand Thomas
Pozorski.Their manuscript(Pozorskiand Pozorski1981) forced us to reevaluate some of our assumptionsand
to make them more explicit. We also gratefullyacknowledgethe inclusionof data fromtheir manuscript.The
responsibilityfor our interpretationsof those data, however, rests squarely with us.
GLOSSARY
Bed Slope (Ub):
For a channelbed at an angle 0 with the horizontal,then 'b = tan 0. Downhillslopes are given
as positive numbers,uphill as negative.
CriticalDepth (Dc):Water depth at which the Froudenumberis unity.
D80: Equivalentparticle grain diametersize such that 80% of all particles have diametersless than this size.
Flow Rate (Q):Numberof cubic meters of water flowing by a fixed reference station in one second.
FrictionVelocity(V*):Defined as (To/p)1/2 where To is the shear stress at the fluid-wallinterface and P is the
fluid density.

FroudeNumber(Fr):Defined as Fr = Vl (gD)1/2where V is the mean water velocity, g the gravitationalconstant, and D is the hydraulicdepth.
Hydraulic Jump:Transition zone between a supercritical and subcritical flow characterized by a height
increase and velocity decrease of the subcritical flow.
ManningRoughnessFactor (n):Empiricalcoefficient related to channel wall roughness.
NormalDepth(Dn):Water depth at which uniform(constantdepth)flow exists in a channel obtainedby solution of the Manningequation, incorporatingthe bed slope for the head loss term.
ReynoldsNumber:Defined as Re = VL/I,v
where V is the mean water velocity, L is a characteristic length,
and v is the kinematicviscosity.
REFERENCESCITED
Barazangi, M., and B. L. Isacks
1976 Spatial distribution of earthquakes and subduction of the Nazca plate beneath South America.
Geology 4:686-692.
Chow, V. T.
1959 Open channel hydraulics. McGraw-Hill, New York.
Cossio, A., and H. Jaen
1967 Geologia de los cuadrangulos de Puemape, Chocope, Otuzco, Trujillo, Salaverry y Santa. Boletin
No. 17. Servicio de Geologia y Mineria, Lima, Peru.
Cromie, W. J.
1980 When comes El Nifio? Science 80 1(3):36-43.
Darwin, C.
1839 Narrative of the surveying voyages of his majesty's ships Adventure and Beagle, between the years
1826 and 1836, describing their examination of the southern shores of South America, and the Beagle's circumnavigation of the globe, Volume III, Journal and Remarks. Henry Colburn, London.
Day, K.
1974 Walk-in-wells and water management at Chan Chan, Peru. In The rise and fall of civilizations, edited
by J. Sabloff and C. Lamberg-Karlovsky, pp. 182-190. Cummings Publishing Co., Menlo Park, Calif.
Ericksen, G. E., G. Plafker, and J. F. Concha
1970 Preliminary report on the geological events associated with the May 31, 1970 Peru earthquake.
United States Geological Survey Circular 639. Washington, D.C.
Farrington, I. S.
1980 The archaeology of irrigation canals, with special references to Peru. World Archaeology 11(3):
287-305.
Farrington, I. S., and C. C. Park
1978 Hydraulic engineering and irrigation agriculture in the Moche Valley, Peru: c. A.D. 1250-1532.
Journal of Archaeological Science 5:255-268.
Henderson, F. M.
1966 Open channel flow. Macmillan, New York.
Herd, D. G., T. L. Youd, H. Meyer, J. L. Arango, W. J. Person, and C. Mendoza
1981 The great Tumaco, Colombia earthquake of 12 December 1979. Science 211:441-445.
James, D. E.
1971 Plate tectonics model for the evolution of the Central Andes. Geological Society of America Bulletin
82:3325-3346.
Keatinge, R. W.
1974 Chimu rural administrative centers in the Moche Valley, Peru. World Archaeology 6:66-82.

REPORTS

595

Kosok, P.
1940 The role of irrigation in ancient Peru. Proceedings of the 8th American Scientific Congress 2:169178. United States Department of State, Washington, D.C.
1965 Life, land and water in ancient Peru. Long Island University Press, New York.

Kus, J. S.
1972 Selected aspects of irrigated agriculture in the Chimu heartland, Peru. Unpublished Ph.D. dissertation, Department of Geography, University of California, Los Angeles.
1974 Irrigation and urbanization in prehispanic Peru: the Moche Valley. In Yearbook of the Association
of Pacific Coast Geographers, Volume 36. Corvallis, Oregon.
Larco Hoyle, R.
1945 Los mochicas. Sociedad Geografica Americana, Buenos Aires.
1946 A culture sequence for the North Coast of Peru. Handbook of South American Indians, Bureau
of American Ethnology Bulletin 143, 2:149-175. Washington, D.C.
Lettau, H. H., and K. Lettau
1978 Exploring the world's driest climate. Institute for Environmental Studies, Report 101. University of
Wisconsin, Madison.
Lumbreras, L. G.
1974 The peoples and cultures of ancient Peru, translated by B. J. Meggers. Smithsonian Institution Press,
Washington, D.C.
Morris, H., and J. Wiggert
1972 Applied hydraulics in engineering. The Ronald Press Co., New York.
Moseley, M. E.
1981a The dynamics of agrarian collapse in coastal Peru. Ms. in possession of the author, Field Museum
of Natural History, Chicago.
1981b Patterns of settlement and preservation in the Viru and Moche valleys. Revised version of a paper
presented at Burg Wartenstein Symposium No. 86, August 16-24, 1980.
Myers, J. S.
1975 Vertical crustal movements of the Andes in Peru. Nature 254:672-674.
Nials, F. L., E. E. Deeds, M. E. Moseley, S. G. Pozorski, T. G. Pozorski, and R. A. Feldman
1979 El Ninio: the catastrophic flooding of coastal Peru. Field Museum of Natural History Bulletin 50(7):
4-14 (Part I), and 50(8):4-10 (Part II).
Ortloff, C. R.
1981 La ingenieria hidraulica Chimfu(parte I): el sistema de canales La Cumbre. In La tecnologia en el
mundo andino, edited by H. Lechtman and A. M. Soldi, pp. 91-111. Universidad Nacional Aut6noma de
M6xico, M6xico, D. F.
Ortloff, C. R., M. E. Moseley, and R. A. Feldman
1981 Hydraulic engineering practice among the coastal Chimu: Part II, the Pampa Huanchaco canal
system. Ms. in possession of the authors, Field Museum of Natural History, Chicago.
Pozorski, T., and S. Pozorski
1981 Reinterpreting the Chicama-Moche Intervalley Canal: comments on "Hydraulic Engineering Aspects of the Chimu Chicama-Moche Intervalley Canal." Ms. in possession of the authors, Section of Man,
Carnegie Museum of Natural History, Pittsburgh.
Richardson, J. B., III
1981 Modeling the development of sedentary maritime economies on the coast of Peru: a preliminary
statement. Annals of Carnegie Museum 50(5):139-150.
Rouse, H., and S. Ince
1963 History of hydraulics. Dover Publications, New York.
Shimada, I.
1981 The Batan Grande-La Leche Archaeological Project: the first two seasons. Journal of Field Archaeology 8:405-446.
Sprague de Camp, L.
1974 The ancient engineers. Ballantine Books, New York.
Von Hagen, V. W.
1965 The desert kingdoms of Peru. Weidenfeld and Nicolson, London.
Wyrtki, K., E. Stroup, W. Patzert, R. Williams, and W. Quinn
1976 Predicting and observing El Nifio. Science 191:343-346.
Wyss, M.
1978 Sea level changes before large earthquakes. Earthquake Information Bulletin 10(5):165-168.
United States Geological Survey, Reston, Virginia.
Zeil, W.
1979 The Andes, a geological review. Beitraige Zur Regionalen Geologie der Erde, Band 13. Gebruider
Borntraeger, Berlin.

S-ar putea să vă placă și