In Arizona and the Southwest, Volume 34 (2004)

Hydrology and Water Resources in Arizona and the Southwest, Volume 34 (2004)

Item Type text; Proceedings

Publisher Arizona-Nevada Academy of Science

Journal Hydrology and Water Resources in Arizona and the Southwest

Download date 06/10/2021 17:37:08

Link to Item http://hdl.handle.net/10150/296609 Volume 34

HYDROLOGY AND WATERRESOURCES INARIZONA AND THE SOUTHWEST

Proceedings of the 2004 Meetings of the

Hydrology Section Arizona -Nevada Academy of Science

April 12, 2004,. Midwestern University Glendale, Arizona DEDICATION

This Proceedings is dedicated to the committed students, professors, and professionals in hydrology and watershed management who have dedicated their efforts to one of the most important aspects of natural resource management and science in the South- west. They are to be thanked for taking time out from their busy schedules to share their research at this meeting. This Proceedings is also dedicated to Dr. Peter Ffolliott, University of Arizona, who has been a long -time inspiration and catalyst for the Hydrology Section, and who is President -Elect of the Academy.

iii ORDERING INFORMATION

This issue can be obtained in hard copy as long as supplies last from Gerald J. Gottfried USDA Forest Service Tonto National Forest 2324 E.McDowell Road Phoenix, AZ85006 602 -225 -5357 ggottfried @fs.fed.us

iv CONTENTS

Water Balance in Upper Lake Mary 1 Assefa Desta and Aregai Tecle

A Water Budget for Emory Oak Woodlands of Southeastern Arizona: An Expansion of the Initial Approximation 11 Peter F. Ffolliott

Spatial Evaluation of Precipitation in Two Large Watersheds in North- Central Arizona 15 Boris Poff, Assefa S. Desta, and Aregai Tecle

Predicting Event Based Peak Discharges Resulting from Thinning and Wildfire for the Upper Rio de Flag Watershed, Flagstaff, Arizona 21 Duncan S. Leao and Aregai Tecle

Use of the Analytic Hierarchy Process in Forest Budget Allocation in Durango, Mexico 39 Gustavo Perez - Verdin and Aregai Tecle

Soil Loss Following the Rodeo -Chediski Wildfire: An Initial Assessment 51 Pablo A. Garcia- Chevesich,Peter F. Ffolliott, and Daniel G. Neary

Water Consumption of Common Plants in the Southwest U.S. 57 Aregai Tecle

Geomorphology of Small Watersheds in an Oak Encinal in the Peloncillo Mountains 65 Daniel G. Neary and Gerald J. Gottfried

v INTRODUCTION

The Spring 2004 meeting of the Hydrology Section of the Arizona- Nevada Academy of Science took place at Midwestern University's Glendale Campus, Glendale, Arizona, on April 12, 2004. The Hydrol- ogy Section organizers wish to extend their thanks to Ms. Louella Holter with the Bilby Research Center, Northern Arizona University, for her editorial assistance on this publication. Dr. Dan Neary, USDA Forest Service, Rocky Mountain Research Station, was the chairperson for this year's Hydrol- ogy Section meeting. Dan Neary Flagstaff, AZ

vii WATER BALANCE IN UPPER LAKE MARY

Assefa Desta1 and Aregai Tecle1

Upper Lake Mary is one of many sources ofcreased population in the city, the height of the drinking water for the city of Flagstaff, Arizona,dam was raised in 1951 to double the capacity of but it is the only surface water. The other sources the lake. The lake is 5 miles long, and a half mile are ground water in deep aquifers. On average, the and 300 feet wide at the widest and narrowest amount of water withdrawn from Lake Mary has sections of the lake, respectively. The water surface accounted for 46 percent of Flagstaff's water level of the lake varies widely, depending on supply since its construction in 1941, but only 29 annual precipitation and surface runoff. At full percent during the relatively dry period of the last capacity, when the water surface tops the spillway, 10 years. The lake also serves as a recreational the surface area of the lake reaches its maximum facility for the people of Flagstaff and others corn- size of 876 acres, with 15,620 acre -feet of capacity. ing from outside the city as well as a habitat for When full, the lake near the dam is 40 feet deep, numerous wild animals in the area. Its proximity and it averages 17.9 feet deep over the entire area. to town, good accessibility, water availability Upper lake Mary has various environmental throughout the year, and suitable location make and resource functions. In addition to being a Lake Mary one of the best areas for viewing wild- source of drinking water for Flagstaff, it serves as a life and birds (Coconino National Forest 1999). recharge basin to the underlying aquifer and its Upper Lake Mary and its twin, Lower Lake surface water is used for fishing and boating. The Mary, are located in the lake Mary graben, a lake is also a habitat for various wildlife species down- dropped block between the Anderson Mesa and an excellent area for picnicking, sunbathing, fault to the north and the Lake Mary fault to the wildlife watching, and nature photography. How- south. Precipitation onto the surface and runoff ever, this study is concerned only with estimating from the surrounding watershed are the main the water balance in the area. inputs into the lake. Evaporation, seepage into the ground, and withdrawal for drinking water are the WATER BALANCE COMPONENTS main outputs from the lake. A water balance analysis is one of the best ways In this study, we use a water budget approach to characterize the hydrology and water resources to characterize the hydrology of the lake and to de- of an area, because the approach includes all termine the amount of water available, by monthaspects of hydrology and other important factors and by year, for various purposes. The basic as- that affect the system (Warren et al. 2002). The sumption in the water budget calculation is thatwater balance is expressed in the form of an equa- the total amount of water entering Lake Mary is tion that describes the relationship between inputs, equal to the total amount of output from the lake outputs, and any change in storage. Furthermore, plus any change in the lake storage. the input and output parts are expressed in terms of many variables representing the different DESCRIPTION OF THE STUDY AREA factors that contribute to each part. For Lake Mary, Upper Lake Mary is a long and narrow artificial the water balance equation is expressed as follows: lake located in the high plateau region of north - P +R- E- G -Q =AS [1] central Arizona, about 10 miles southeast of Flag- staff (see Figure 1). It was created in 1941 by con- where structing an earthen dam to provide a source of P= the amount of precipitation falling on the water for the city of Flagstaff. In response to in- surface of the lake R = the amount of runoff entering the lake 1School of Forestry, Northern Arizona University, Flagstaff from the surrounding watershed 2 Desta and Tecle

Upper Lake Mary and its watershed

Coconino National Forest

Figure 1. Location of Upper Lake Mary within the Coconino National Forest in Arizona.

E= the amount of evaporation from the lake Precipitation surface Precipitation comes in the form of rain or snow G = the amount of ground water seepage from that falls directly on the lake surface. The snowfall the wetted surfaces of the lake is converted to equivalent water in the water Q =the amount of lake water withdrawal for budget analysis. There are no specific precipitation the city of Flagstaff data available for the Lake Mary area. The data AS = the change in the lake storage. used here reflect the 1950 to 2003 precipitation record obtained from the National Weather Service inputs archive recorded at the nearby Flagstaff airport. The inputs in the Lake Mary water balance The surface area of the lake for each month can be equation are direct precipitation on the lake and estimated from the stage versus area relationship surface runoff into the lake from the surrounding curve shown in Figure 2 (Blee 1988; Ward et al. watershed. Most lakes have subsurface inputs that 1995). Then, we determined the monthly average come in the form of springs. However, due to the amount of water from precipitation by multiplying unique geologic formation with offsets in thearea, the area of the lake by the corresponding precipita- there is no significant subsurface contribution to tion depth, as shown in Table 1. Lake Mary. Water Balance in Upper Lake Mary 3

100 200 300 400 500 600 700 800 900 Lake surface area in acre

Figure 2. Relationship between lake stage and surface area (Blee 1988).

Table 1. Estimating the amount of water from direct precipitation falling on the surface of Lake Mary.

Precipitation Precipitation Average Lake Lake Surface Precipitation Month (inches) (feet) Stage (ft) Area (acres) Voume (ac -ft)

January 1.99 0.166 24.12 560 92.87 February 2.09 0.174 24.42 565 98.40 March 2.24 0.187 27.00 650 121.33 April 1.28 0.107 30.22 730 77.87 May 0.70 0.058 30.18 728 42.47 June 0.50 0.042 29.45 725 3021 July 2.48 0.207 27.86 693 143.22 August 2.83 0.236 26.75 648 152.82 September 1.96 0.163 25.70 610 99.63 October 1.59 0.133 24.97 595 78.84 November 1.72 0.143 24.47 566 81.13 December 1.93 0.161 24.02 558 89.75 Sum 21.31 1.776 - - 110853 4 Desta and Tecle

Surface Runoff loam and shallow sandy loam soils that have low Surface runoff from the watershed area around organic matter content but usually are high in clay the lake is the main input into the water budget content and with a fair amount of wood or forest equation for Lake Mary. The watershed area surface cover. Table 2 shows the amounts of sur- around the lake, which was determined using a face runoff estimated using the SCS method de- Digital Elevation Model (DEM), is estimated to be scribed above. about 34,043 acres. We used the USDA Soil Con- servation Service's (SCS) curve number method in Outputs equation [2] to estimate the amount of runoff The output components of the water balance produced from precipitation events falling on the equation [1] are evaporation, groundwater seep- watershed. age, and water withdrawal by the city of Flagstaff. Estimates for the values of these variables are (P - Ia)2 obtained either from the literature, by using equa- [2] Q-- Ia + S tions, or from real data obtained from the city of Flagstaff Water Treatment Plant. where Q = the excess rainfall or direct runoff from a Evaporation storm event (inches) According to Blee (1988), the evaporation loss P = depth of precipitation (inches) from the water surface in Upper Lake Mary can be Ia = initial abstraction before ponding (inches) estimated using a mass transfer equation devel- S = potential maximum retention (inches). oped by Marciano and Harbeck (1952). The mass transfer value, which is directly proportional to the The amount of the initial abstraction, Ia, is approxi- product of wind speed (U) and the vapor pressure mated by 0.2S for use in equation [2]. The result is gradient (eo -ea), is expressed in equation [7]: shown in equation [3]. E = NU(eo - ea) [7] (P - 0.2S)2 [3] where Q P +0.8S E = evaporation loss (in inches) The value of S in turn is estimated from a curve N = mass transfer coefficient number (CN) using equation [4] U = wind speed at some height above the water surface (in miles per hr) S 10 [4] CN eo = saturated vapor pressure that corresponds where CN is a dimensionless curve number de- to the temperature of the water surface (in fined such that 0 _< CN <_ 100. For impervious sur- Hg) faces, the value of CN = 100, whereas it is less than ea = vapor pressure of the air at some height 100 for other surfaces. above the water surface or shore (inches/ The curve number is usually readily available in Hg). the form of tables for average antecedent moisture The constant N in equation [7] is related to other condition (AMC II). Then, either a conversion table independent evaporation factors, such as energy or (USDA Soil Conservation Service 1972) or equa- some other water budget component. For Upper tions [5] and [6] below (Chow et al. 1988) can be Lake Mary, an empirical equation that was devel- used to convert the AMC II values into antecedent oped by Koberg and Ford (1965) was used to deter- moisture for dry conditions (AMC 1) and for wet mine the amount of evaporation from the lake. To conditions (AMC III). obtain the data necessary for estimating evapora- 4.2CN(II) tion, a raft was anchored at each end of the lake, CN(I) = 10 - 0.058CN(11) containing instruments to measure the mass transfer variables of vapor pressure and wind speed as 23CN(II) CN(III) = well as temperature -an anemometer, a psychrom- 10 +0.13CN(II) eter, and a water temperature probe. A water The CN value for the average antecedent moisturetemperature correction factor to correct fetch, the condition (AMC II) for the study area is estimated distance between the upwind shore and the point to be 77. This assignment is made on the basis ofof measurement, and an atmospheric stability cor- the area having soil group C, described by clay rection factor were used in constructing the mass Water Balance in Upper Lake Mary 5

Table 2. Monthly average runoff from the Upper Lake where Mary watershed (34,043 acres). G=ground water seepage from the lake (ft) Runoff Runoff H1 =lake stage (depth) at the beginning of the Month Depth (ft) Volume (acre -ft) water budget period (ft) H2 =lake stage at the end of the water budget January 0.042 1418.44 period (ft) February 0.022 737.59 March 0.029 992.91 P=precipitation on the lake surface (ft) April 0.003 85.11 E=evaporation rate (ft) May 0.000 0.00 water withdrawal from the lake(in feet). June 0.006 198.58 Q = July 0.017 567.37 The change in storage (H1 - H2) was determined August 0.025 851.06 from recorded changes in lake stage. The amount September 0.019 652.48 of evaporation (E) was measured at the rafts in- October 0.017 567.38 stalled in Upper Lake Mary. The amount of water November 0.038 1304.96 withdrawal (Q) was measured by the city of Flag- December 0.020 680.85 staff persónnel at the water treatment plant. Pre- Sum - 8056.73 cipitation depth (P) was measured by the rain gage at the upper end of the lake. The stage /seepage relationship in Figure 4 was established to estimate the long -term seepage loss transfer equation to determine the evaporation rate from the lake (Blee 1988). It was constructed by on each raft from May to October of 1969 and 1970, plotting the results obtained from the short -term the time of year with the largest amount of evapo- water budgets. The average seepage loss volume ration occurrence. The evaporation rate computed from the lake for each month of the year can be for the lake was weighted by the lake surface area estimated from this graph. each raft represented, as shown in equation [8]. E1A1 + E2A2 Water Withdrawal by the City of Flagstaff ET [8] A1+ A According to the data on the monthly amount 1 2 of water pumped from Lake Mary by the city of where Flagstaff during the years 1949 through 2001, the ET = average lake evaporation rate (inches /hr) amount of water pumped varied with the amount E1 = evaporation rate at from raft 1 (inches /hr) of water coming into the lake and the seasonal E2 = evaporation rate at from raft 2 (inches /hr) water demand. The data show that, on average, Al = lake surface area covered by raft 1 (acres) 2363.38 acre -feet of water per year has been A2 = lake surface area covered by raft 2 (acres). pumped from the lake during the last 52 years. Table 3 shows the amount of water pumped each Figure 3 gives the average amount of monthly month during the last 5 dry years, indicating the evaporation determined for Lake Mary. As shown effect of drought on the demand for Lake Mary above, the largest amounts of evaporation occur water (City of Flagstaff Utility Department 2002). during the warm summer months of May through October. RESULTS AND CONCLUSION Annual inflows and outflows have been deter- Seepage Losses mined using several methods and the available Seepage losses from Upper Lake Mary were de- data. The water balance for each month -that is, termined from several short -period water budget the change in the lake volume -is determined by estimates; the periods ranged from 3 to 60 days. subtracting the total average outputs from the total The water budgets were computed only for peri- inputs into the lake. Table 4 shows the monthly ods in which there was no surface flow into the average values of the different hydrologic condi- lake. All the variables in the water budget of equations contributing to the water balance in Upper tion [1] were measured except seepage, which was Lake Mary. The last column in Table 4 and Figure computed as a residual using equation [9]. 5 show the monthly changes in the amounts of water in Lake Mary. G= (H1- H2) +P -E -Q [9] 6 Desta and Tecle

400 350 300 i250 200 150 i100 50

, °, 3Y; é Month

Figure 3. Monthly evaporation loss (in acre -ft) from Upper Lake Mary (Blee 1988).

Figure 4. Relationship between lake stage and seepage loss developed froma short-term water budget (Blee 1988). Water Balance in Upper Lake Mary 7

Table 3. Average monthly water withdrawal from Upper Lake Mary during 1999 -2003.

Average Percentage Actual Annual Avg Actual Average Withdrawal from Water Pumped Monthly Month (1999 -2003) [2] Total [3] 1941-2001 (ac -ft) [4] Volume = [3] * [4]

January 266.17 4.37 103.23 February 93.81 1.54 36.38 March 246.67 4.05 95.72 April 269.30 4.42 104.49 May 1125.60 18.47 436.70 June 1299.78 21.33 504.31 July 913.07 14.98 2363.38 354.12 August 687.22 11.28 266.53 September 421.12 6.91 163.33 October 190.69 3.13 73.96 November 262.18 4.03 101.73 December 318.13 5.22 123.43 Sum 6093.73 - 2362.93

Table 4. Average monthly values of the different water balance components in Lake Mary.

Precip. Runoff Using Change on Lake Evap- With- SCS Method Runoff in Surface oration Seepage drawal Volume Storage Month (ac -ft) (ac -ft) (ac -ft) (ac -ft) (in) (ft) (ac -ft) (ac -ft)

January 92.87 56.00 263.5 103.23 0.50 0.04 1418.44 1088.58 February 98.404 70.60 255.2 36.39 026 0.02 737.59 473.80 March 121.33 124.60 341.0 95.72 0.35 0.03 992.91 552.92 April 77.87 243.30 420.0 104.49 0.03 0.01 85.11 -604.81 May 42.47 327.60 427.8 436.70 0.00 0.00 0.00 -1149.63 June 30.21 374.60 396.0 504.31 0.07 0.01 198.58 -1046.12 July 143.22 297.40 372.0 354.27 020 0.02 567.37 - 313.08 August 152.82 270.00 334.8 266.62 0.30 0.03 851.06 132.46 September 99.633 208.40 294.0 163.39 0.23 0.02 652.48 86.323 October 78.838 185.90 285.2 73.99 0.20 0.02 567.38 101.13 November 81.127 82.50 267.0 101.73 0.46 0.04 1304.96 934.86 December 89.745 55.80 260.4 123.43 0.24 0.02 680.85 330.97 Annual 1108.53 2296.80 3916.9 2364.27 - - 8056.73 587.29 8 Desta and Tecle

1500

1000

500 m rn

J o -500

ca .0 U -1000

-1500 Z a) co m C tii iii To co 2 á C 2 O C co aEi E 0 8 CD Z p Month

Figure 5. Monthly change in storage in Lake Mary obtained using the water balance method.

The water balance calculation shows that the Deepening the lake by at least 17 ft, which is the annual average amount of water in Upper Lake average depth, and sealing 64 percent of the lake Mary is 587.29 acre -feet. This is only about 3.76 bottom would avoid 50 percent of the evaporation percent of the maximum capacity of the lake. This loss and 81 percent of the seepage loss. Since the amount is so small that it is not adequate even to lake is located within the USDA's Coconino Na- satisfy the supplementary drinking water needs of tional Forest and was built by the city of Flagstaff, the city of Flagstaff. Under this kind of situation, it it is managed jointly by both the Forest Service and would be unwise to pump any water from the the city. There is an agreement between the two lake. Because of this the amount of water with-not to lower the lake depth below 18.3 ft. This drawn from the lake varies from year toyear helps to avoid overwithdrawal while ensuring its depending on weather conditions in the area. In continuous availability for recreational activities, some years, little or no water is pumped from the wildlife habitat, and other benefits. This water bal- lake to avoid total loss of the lake duringsevere ance study estimates the average water availability drought periods. Table 4 shows that the largestin Upper Lake Mary for domestic and other uses. amount of water losses from the lake are through However, the results do not show the water evaporation and seepage. Both types of losses balance characteristics during wet and dry periods. amount to 72 percent of the total output. The rest is Evaluating the water balance on a yearly basis can withdrawn for drinking water by the city of Flag-help estimate the amount of water availability for staff. However, the withdrawal does notoccur withdrawal under different climate conditions. when the lake is low during dry periods. Hence further studies to find ways to reduce the amount ACKNOWLEDGMENTS of evaporation and seepage are appropriate. Re- This study was partially funded by a grant from ducing the loss will save water that can be used for the State of Arizona Proposition 301 (ERDENE) the increasing population in the city of Flagstaff and by the USDA Forest Service, Rocky Mountain and to prevent the lake from drying up during Research Station grant #01 -W- 11221606 -214. drought. Water Balance in Upper Lake Mary 9

REFERENCES Blee, J. W. H. 1988. Determination of evaporation and Marciano, J. J., and G. E. Harbeck, Jr. 1952. Mass -transfer seepage losses, Upper Lake Mary near Flagstaff, studies, in water -loss investigations, Lake Hefner Arizona. U.S. Geological Survey. Water Resources studies, technical report. U.S. Geological Survey Pro- Investigation Report 87 -4250. fessional Paper 269. Chow, V. T., D. R. Maidement, and L. W. Mays. 1988. USDA Soil Conservation Service. 1972. National Engg Applied Hydrology. McGraw -Hill, New York. neering Handbook, Section 4, Hydrology. U.S. De- City of Flagstaff Utility Department. 2002. Water and Oartment ofAgriculture.U.S. Government Printing Waste Water Operation Plan, report to the Water ffice, Washington DC. Commission, Flagstaff, Arizona. Coconino National Forest. 1999. Flagstaff /Lake Mary Ward, A. D., and W. J. Elliot. 1995. Environmental Hy- ecosystem analysis. Available online at http: / /www. drology. CRR Press, Washington, DC. fs.fed.us/r3/coconino/nepa/flea_minerals.html. Warren, V., Jr., and G. L. Lewis. 2002. Introduction to Koberg, G .E., and M. E. Ford, Jr. 1965. Elimination of Hydrology. Harper Collins College Publishers, New thermal stratification in reservoirs and the resulting York. benefits, in contribution of the hydrology of the United States, 1964. U.S. Geological Survey Water - Supply Paper 1809 -M. A WATER BUDGET FOR EMORY OAK WOODLANDS OF SOUTHEASTERN ARIZONA: AN EXPANSION OF THE INITIAL APPROXIMATION

Peter F. Ffolliottl

An initial approximation of an annual water bud- Infiltration was the most problematic of the get characterizing the Emory oak woodlands of budget components examined. Nevertheless, it southeastern Arizona (Ffolliott 1999) was pre- was estimated from plot data presented by Beutner sented in the Hydrology Section of the Arizona - et al. (1940) that about 80-85 percent of the annual Nevada Academy of Science at the 2000 meeting. net precipitation infiltrates into the soil in the Storage points and the flows of water within and "average" year. However, the average year rarely between these storage points were examined in the occurs in the Emory oak woodlands. Fluctuations context of the results of hydrologic investigations in rainfall regimes typically result in a few very available at that time (Ffolliott 2000). Analysis of wet years interspersed among several average this water budget indicated that transpiration was years and below average years. the largest component, a finding that is generally Transpiration of Emory oak trees had been esti- expected (Brooks et al. 2003; Chang 2003). Esti- mated by the sapflow velocity method (Swanson mates of transpiration rates of thinned coppice on and Witfield 1981; Swanson 1994) in an earlier the rootstocks of harvested Emory oak trees have study (Ffolliott et al. 2003). The estimate of trans- been made since this initial approximation, allow- piration by a stand of mature trees (60 years and ing an expansion to be made of the water budget older) was about 50 percent of the net annual for these woodland ecosystems. precipitation. Mature trees and stump sprouts collectively stocking a partially (more than 75 %) BACKGROUND PERSPECTIVE harvested stand transpired 85 percent of the net Earlier investigations of the hydrologic charac- annual precipitation; the large number of post- teristics of the Emory oak woodlands in southeast-harvest stump sprouts occupying the stand was ern Arizona allowed the initial approximation of the assumed reason for this finding. Transpiration an annual water budget by examining key water rates for other woody plants, grasses, forbs, and storage points and flows within and between these succulents found in the Emory oak woodlands are storage points. Components of the water budget unknown. cycle represented in this approximation included By applying a set of basic operating rules, and interception and stemflow, infiltration, stormflow, using the estimated long -term frequency of "large" and transpiration. storms in an "average year" in interpreting these Extrapolating results of a study of rainfall distri- rules ( Ffolliott 2000), it was estimated that approxi- bution in Emory oak woodlands by Haworth and mately 5-10 percent of the net annual precipitation McPherson (1991) to a stand -basis using known is converted into stormflow in the average year. tree distributions and the estimated long -term Most of this stormflow is generated by relatively frequency of large and small storms in a year led few rainfall events. Snowmelt runoff (when it to an estimated canopy interception value of 10occurs) also flows into stream channels passing percent of the annual gross precipitation. The through the woodlands; however, this flow path- magnitude of litter interception is unknown, but it way was not considered in this water budget. was considered negligible because of the sparse buildups of litter on the soil surface. Stemflow was TRANSPIRATION OF THINNED COPPICE also considered insignificant. Net precipitation, Estimates of annual transpiration by thinned therefore, was estimated to be 90 percent of thesprouts (coppice) on the rootstocks of harvested annual gross precipitation. Emory oak trees have been made (Shipek et al. 2003, 2004) since the initial water budget approxi- 1School of Natural Resources, University of Arizona, Tucson. mation was presented. These latter estimates of 12 Ffolliott

transpiration were also obtained by the sapflow 450 mm, a value between the reported ranges of velocity method. This more recent information annual precipitation values in the Emory oak complements the findings from the earlier study of woodlands and that used in illustrating the initial Emory oak transpiration (Ffolliott et al. 2003) and approximation (Ffolliott 2000), into the respective furnishes a basis to estimate the amount of water water budget components. The estimates of water that is lost by the transpiration process afterstorage points and flows of water within and implementing selected coppice thinning practices. between the storage points shown in Figure 1 are A summary of the transpiration rates for all of the presented only to indicate the relative magnitudes Emory oak stand conditions studied to date is of these water storage points in relation to the presented in Table 1. Parenthetically, the similarity annual gross precipitation amount selected. The of the transpiration estimates for the mature trees ranges of absolute values for these components are and the stump sprouts stocking the partially unknown. harvested stand and the stand of unthinned root- Water losses not shown in Figure 1 include stocks averaging 4.5 dominant post -harvest stump evaporation from water surfaces, soil evaporation, sprouts is attributed largely to the overriding and watershed leakage. Evaporation from water dominance of the post- harvest stump sprouts surfaces is a minor loss in the Emory oak wood- occupying the partially harvested stand. lands of southeastern Arizona because of the It is apparent from Table 1 that the thinning of general absence of large water bodies. However, post -harvest Emory oak coppice has the potential soil evaporation could be a significant part of the to decrease the amount of water lost by transpira-water budget in the interval between large rain- tion in harvested stands and, therefore, increase storm events when the relative humidity is low the amount of water potentially available to and wind velocities are high. Knowledge of the recharge groundwater aquifers, produce overland geology for the watershed in question provides the flow, or increase the growth of other plants in best evidence of losses to leakage. comparison to the amount of water available in stands of unthinned rootstocks. Incorporating SUMMARY these latter estimates of transpiration into the Estimates of annual transpiration by thinned initial approximation of the annual water budget Emory oak stump sprouts resulting from earlier provides expanded insight on the impacts of harvests of fuelwood are available to complement management practices. the findings from the earlier investigation of Em- ory oak transpiration and, additionally, to furnish ANNUAL WATER BUDGET a basis to estimate the amount of water loss by The annual water budget expansion presented transpiration after implementing coppice thinning in this paper was developed by partitioning an practices. Incorporating these estimates of transpi- arbitrarily selected gross precipitation amount ofration into an approximation of the annual water budget provides insight on the impacts of selected management scenarios on the hydrology of these Table 1. A summary of the transpiration rates for woodlands. selected Emory oak stand conditions. ACKNOWLEDGMENTS Stand Condition Percents This research was supported by funds provided to the International Arid Lands Consortium by the Mature trees 50 USDA Forest Service and USDA Cooperative Partial (over 75 %) harvest 85 States Research, Education, and Extension Service, Complete harvest and by the Arizona Agricultural Experiment Unthinned stump sprouts 85 Station. Thinned to 3 residual stump sprouts 65 Thinned to 2 residual stump sprouts 55 Thinned to 1 residual stump sprout 40 1Percent of net annual precipitation Water Budget for Emory Oak Woodlands 13

GROSS PRECIPITATION (450 mm)

INTERCEPTION (45 mm)

NET PRECIPITATION (405 mm)

INFILTRATION STORMFLOW OTHER WATER LOSSES (325 -350 mm) (20 -45 mm)

TRANSPIRATION (155 -345 mm)

Figure 1. Annual water budget for selected management practicesin the Emory oak woodlands of southeastern Arizona. 14 Ffolliott

REFERENCES CITED Beutner, E. L., R. R. Gaebe, and R. E. Horton. 1940. Haworth, K., and G. R. McPherson. 1991. Effects of Quer- Sprinkledloy runoff and infiltration experiments on cus emoryi on precipitation distribution. Journal of the Arizonadesertsoils. Transactions of the American Arizona -Nevada Academy of Science, Proceedings Geophysical Union 2: 550 -558. Supplement 26: 21. Brooks, K. N., P. F. Ffolliott, H. M. Gregersen, and L. F. Shipek, D. C., L. F. DeBano, G. J. Gottfried, and P. F. DeBano. 2003. Hydrology and the management of Ffolliott. 2003. Water use by rootstock of Emory oak watersheds.IowaStatePress, Ames. 574 pp. coppice. In Assessing Capabilities of Soil and Water Chang, M. 2003. Forest hydrology: An introduction to Resources in Drylands: The Role of Information Re- water and forests. CRC Press, Boca Raton, Florida. 393 trieval and Dissemination Technologies. Proceedings of the Conference and Workshop, pp. 167 -173. Inter- Ffolliott,P. F. 1999. Encinal woodlands in the southwest- national Arid Lands Consortium, Tucson, Arizona. ern United States. In Ecology and Management of Forests, Woodlands, and Shrublands in the Dry land Shipek, D. C., P. F. Ffolliott, G. J. Gottfried, and L. F. Regions of the United States and Mexico: Perspectives DeBano. 2004. Transpiration and multiple use man- for the 21st Century, edited by P. F. Ffolliott and A. agement of thinned Emory oak coppice. USDA Forest Ortega -Rubio, pp. 69-81. Centro de Investigaciones Service, Research PaperRMRS-RP-. 8 pp. Biologicas del Noroeste, S.C., La Paz, BCS, Mexico. Swanson, R. H. 1994. Significant historical developments Ffolliott, P. F. 2000. An annual water budget for Emory in thermal methods of measuring sap flow in trees. oak woodlands: An initial approximation. Hydrology Agricultural and Forest Meteorology 72: 113-132. and Water Resources in Arizona and the Southwest Swanson, R. H., and W. A. Witfield. 1981. A numerical 30: 37-41. analysis of heat pulse velocity theory and practice. Ffolliott, P. F., G. J. Gottfried, Y. Cohen, and G. Schiller. Journal of Experimental Botany 32: 221-239. 2003. Transpiration by dryland oaks: Studies in the southwestern United States and northern Israel. Jour- nal of Arid Environments 54: 638-644. SPATIAL EVALUATION OF PRECIPITATION IN TWO LARGE WATERSHEDS IN NORTH -CENTRAL ARIZONA

Boris Poff,1,2 Assefa S. Desta,1 and Aregai Teclel

The USDA Forest Service established the Forest Service had in place to collect rainfall data Beaver Creek Experimental Watershed Pilot in the experimental project (see Figure 1). The Project in north -central Arizona in 1957 andobjective of the Beaver Creek Experimental operated it until 1982. After the Forest Service Watershed Pilot Project was to determine the discontinued the project in 1982, Northern Arizona impacts of forest manipulation on water yield. The University's School of Forestry continued toultimate desire was to increase the amount of monitor and do research in two of the largest water available from the forest regions of Arizona watersheds, known as Woods Canyon and Bar M. for the growing population in the state. From 1982 to 1995 only streamflow data are available for these watersheds. A weak correlation METHODOLOGY between streamflow and precipitation for both The data used in the evaluation process for Woods watersheds was determined by analyzing the Canyon (WS19) consisted of precipitation data USDA data from 1961 to 1982, using spatial from 39 rain gauges in the Beaver Creek Experi- interpolation methods in ArcGIS. However, this mental Watershed for the time period from 1961 to study is a work in progress and the authors are in 1981. Streamflow data from the Woods Canyon the process of using different spatial interpolation stream gauging station for the same time period methods in ArcGIS to determine better regression were used as well. The data are available at a Web functions between streamflow and precipitation. site maintained by the Rocky Mountain Research The objective of this study is to evaluate existing Station and the University of Arizona in Tucson historical data using newly available techniques (http: / /ag.arizona.edu /OALS /watershed /index.h and technology to obtain new results. Methods tml). The data used in the evaluation process for used so far, in ArcGIS, are four different spatial Bar M (WS20) consisted of the same precipitation interpolation methods, namely Inverse Distance data used for Woods Canyon from 39 rain gauges Weighting, Radial Basis Function, and Global and in the Beaver Creek Experimental Watershed from Local Interpolation. 1961 to 1981, and other 1961 -1990 precipitation data from seven additional weather stations east BACKGROUND AND STUDY SITE and south of the Beaver Creek Watershed. This Originally, between 1957 and 1962, the USDA became necessary because the boundaries of Bar M Forest Service set up 20 pilot watersheds in the extend farther east than the location of any of the Beaver Creek Experimental Watershed in north- 39 Beaver Creek rain gauging stations. The stream- central Arizona. Of the 20 watersheds, 18 ranged flow data from the Bar M stream gauging station from 27 to 824 ha. The remaining two watersheds, were used for the time period of 1961 -1981. The known as Bar M (Watershed 20) and Woods Can-streamflow and precipitation data from the 39 yon (Watershed 19), are much larger with areas of Beaver Creek Experimental Watershed rain gaug- 6620 and 4893 ha, respectively (Baker and Ffolliott ing stations were downloaded from the watershed 1999; see Figure 1). Seventeen of the watersheds, management Web site given above. Rain data from including the largest two, are located in the the additional seven weather stations were taken ponderosa pine forest ecosystem. This study uses from the Global Historic Climate Network data, data from 39 of the 72 rain gauges the USDA National Climate Data Center, National Oceanic and Atmospheric Administration, U.S. Department of Commerce. 1USDA Forest Service Rocky Min. Research Station, Flagstaff A Trimble GeoExplorer3 handheld GPS unit 2School of Forestry, Northern Arizona University, Flagstaff was used to determine the boundary for the two 16 Poff, Desta, and Tecle

Coconino NF

e 1 Rain -gauging Stations

' d

Figure 1. The location of the study sites and of the 39rain gauges used in this study, in the Beaver Creek Experimental Watershed, Arizona.

watersheds (Poff et al. 2005). The GPS datawere assumption necessary to using IDW is that this recorded in the Universal Transverse Mercatorphenomenon is modeled as a spatially continuous (UTM) system and projected using the Northdataset. Even though precipitation valuesare American Datum (NAD) of 1927. usually collected as discrete point data, it is safe to Precipitation data were converted into milli- make this assumption with such a dataset. meters where necessary (NOAA data were already Even though the data tend to be normally dis- in mm) and the rainfall values (1961 -1981)were tributed, there is one outlier, which can be seen in averaged for each calendar month for each rain the histogram Voronoi map. Including this outlier gauging station. This produced 12 average values in the dataset produced a Kurtosis value of 5.21 for each station -one for each month of theyear. and a skewness coefficient of 1.00. The Voronoi These average values were imported into ArcMap, map also indicates clustering of values around where their UTMs were calculated. After the several locations, which given the pattern of pre- gauging stations were spatially referenced, Inverse cipitation is to be expected. Distance Weighting (IDW), Radial Basis Function The average rainfall value for August for (RBF), and Local and Global Interpolation methods Woods Canyon was used for a more in -depth in- were used to create a precipitation interpolation vestigation after the interpolation output for each map for each month. IDW, which is a local inter-month using the IDW method had been reviewed. polator, weighs closer points greater than those Because these values were collected during the farther away, applying the fundamentalassump- monsoon season they provide adequate data for tion for spatial interpolation (ESRI 2001), which is statistical analysis. the first law of geography: Everything is related to Global and local trends such as latitude everything else, but near things are more related (global), as well as elevation and aspect (local),can than distant things (Tobler 1970). The second influence the spatial distribution of precipitation. It Spatial Evaluation of Precipitation in Watersheds 17

Figure 2. A bar graph showing the mean volume of precipitation for each month for the two watersheds(converted from cu m back to average mm distributedover the entire watershed).

is safe to assume that elevation and consequently watershed resulting in a total of 12 values for each temperature are factors causing a trend, given that watershed (see Figure 2). Next these valueswere the gauging stations used to collect this informa- used to predict streamflow basedon precipitation tion are distributed along the elevation gradient ofvolume calculated by IDW for each watershed. the Mogollon rim. The results of the linear, quadratic, andcubic IDW, local, and global polynomial interpola- regression functions are shown in Figures 3 and 4. tions were run with a power of 1 forreasons of simplicity. The neighborhood searchwas set to RESULTS include at least 10 neighbors for all interpolations, The global polynomial interpolation is the whereas we included 15 for the Radial Basis least exact interpolation, because ituses the entire Function (RBF), the IDW, and the local polynomial dataset to make its predictions and shows values interpolation. We selected 39 neighbors (or all that are different from the measured values. How- gauging stations) for the global polynomial inter-ever, it does show a trend in the level of precipita- polation. These helped tocompare the different tion. It indicates that precipitation increases from methods to each other at thesame (default) set- southeast to northwest. This trend is also indicated tings whenever possible. by the local polynomial interpolation, whosepre- The 24 IDW output maps (12 each for Woods dicted values more closely resemble the measured Canyon and Bar M) as well as the two watershed ones. The results obtained using both the IDW and polygons were converted into raster layers witha RBF, better represent the measured data than those resolution of 50 x 50 m cells. This allowed theuse obtained using the polynomial interpolations. of 3D spatial arithmetic to calculate precipitation Even though the results in the formertwo methods volume in cubic meters for each month foreach are relatively similar to each other, thereare 18 Poff, Desta, and Tecle

O Observed - Linear -- Quadratic - - Cubic

StreâiriÍlow inm3

Figure 3. Linear, quadratic, & cubic regression curves of fitting the WoodsCanyon data; the cubic line has the best fit.

Q Observed Linear -- Quadratic - - Cubic

I 1 1 1 I 1 I 0.00 500.00 1000.001500.00 2000.00 2500.00 3000.00 Streamflow in m3

Figure 4. Linear, quadratic, and cubic regressioncurves of fitting the Bar M data. The cubic line has the best fit. Spatial Evaluation of Precipitation in Watersheds 19

noticeable differences. These differences are due to cells to accurately calculate the average precipita- the way each method calculates the interpolation. tion for different time periods for each of these The r2 values of the three different regression cells. Local or global polynomial interpolation can functions are given in Table 1. The cubic regression be used to identify trends such as elevation or functions show the highest r2 values for both temperature gradients. All methods of interpola- watersheds -0.60 for Bar M and 0.59 for Woodstion can also be used to identify global or local Canyon. outliers, which, in turn, can influence our data if not taken into consideration. In other words, CONCLUSION interpolation is a tool that can be successfully The r2 values generated by the regression applied in today's decision -making processes. functions are not high enough to be very reliable. However, they show that there is a correlation between streamflow and precipitation and that Table 1. The r2 values for linear, quadratic, and cubic re- streamflow data can be used, in this case, to gressions for the two watersheds. "predict" precipitation for the years and months for which precipitation data are missing. In the Type of Function Woods Canyon Bar M future, the authors will use the above methods to determine the relationship between streamflow Linear .359 .205 and precipitation; hence this study is considered a Quadratic .548 .521 work in progress. There are other geostatistical Cubic .558 .601 techniques to be used, such as Kriging and Cokriging, as well as the transformation of data and the elimination of outliers. Another possibility REFERENCES CITED would be to run these techniques on a month by Baker, M. B., Jr., and P. F. Ffolliott. 1999. Interdiscipli- month basis instead of the mean values for 20 nary land use along the Mogollon Rim. In History of years. The goal is to use such techniques to find a Watershed Research in the Central Arizona High- lands, compiled by M. B. Baker, Jr. USDA Forest more reliable correlation (higher r2 value) between Service General Technical Report RMRS- GTR -29, Ft. streamflow and precipitation in order to "predict" Collins, CO. precipitation based on streamflow data with a ESRI. 2001. Using ArcGISTM Geostatistical Analyst. ESRI, Redland, CA. higher accuracy and reliability for the water years Poff, B., D. S. Leao, A. Tecle, and D. Neary. 2005. Deter- 1982 -1994. mining watershed boundaries and area using GPS, Interpolation in general, but IDW or RBF in DEMs, and traditional methods: A comparison. Pro - ceedings of the 7th Biennial Conference of Research particular, can be very useful geostatistical tools on the Colorado Plateau. USGS Plateau Field Station, for predicting values in areas where adequate data Northern Arizona University, Flagstaff, November are not available or when it is difficult to predict 2003. University of Arizona Press, Tucson. Tobler, W. 1970. A computer movie simulating urban values in the past. Precipitation is a prime ex- growth in the Detroit region. Economic Geography ample. Areas can be divided into relatively small 46(2): 234 -240. PREDICTING EVENT BASED PEAK DISCHARGES RESULTING FROM THINNING AND WILDFIRE FOR THE UPPER RIO DE FLAG WATERSHED, FLAGSTAFF, ARIZONA

D. S.Leao1and A.Tecle1

The upper Rio de Flag watershed in Flagstaff, Binkley 2000; Robichaud et al. 2000; Robichaud Arizona (Figure 1) is drained by an ephemeral 2000; Robichaud and Hungerford 2000; DeBano et stream, the Rio de Flag, and its tributaries. During al. 1998). The effects of forest disturbance through storms, the Rio de Flag gathers surface flows from burning on stream peak discharges are highly a number of parallel sub -basins that descend more variable and complex. In the Southwest, post -fire than 5000 feet from the top of the San Francisco peak discharge increases of 500 -9600 percent are Peaks to the city of Flagstaff. The steepness of the common due to intense monsoon rainfalls at the San Francisco Peaks induces rapid downstream end of the summer fire season, steep terrain, shal- movement of surface water runoff, resulting in low skeletal soils, and water repellency (Robi- flooding along the Rio de Flag within the city (Hill chaud et al. 2000). et al. 1988). During the winter, spring, and mon- To decrease the risk of a stand -replacing fire, soon (summer) seasons of the last two centuries, forest thinning operations are being conducted in peak discharges entering into downtown Flagstaffthe watershed within open forest stands and in the have caused extensive property damage (Murray wildland -urban interface (Fort Valley Ecosystem 1974). Restoration Project 2002). Many studies have For a period of 11 years (1969 -1980) the city of shown that forest thinning can increase water yield Flagstaff and the USGS operated three crest -stage and peak discharges from watersheds (Nyland gauging stations in the 32,000 -acre Upper Rio de 1996; Gottfried 1991; Tecle 1991; Baker 1984; Brown Flag watershed to determine discharges with et al. 1974). A long -term study of forest thinning in recurrence intervals of 2, 5, 10, and 25 years (Hill et the Beaver Creek Experimental Watershed, located al. 1988). Although the gauging stations are no within 30 miles southeast of the Rio de Flag water- longer in operation, they provided important in- shed, showed significant increases in stream dis- formation to city planners who anticipated future charge (Brown et al. 1974). Baker (1984) showed a development in the watershed and surrounding 63 percent increase in annual water yield in the areas. ponderosa pine vegetation type after intensive tree While development continues throughout the removal. In addition, Burton (1997) reported a 66 watershed, there is another threat to the health of percent increase in peak discharge after thinning in the watershed and downstream property. This Brownie Creek, Utah. The causes for the increased threat comes from several years of drought, cur- flow are thought to be more saturated soils, soil rent and future insect outbreaks, and heavy compaction, road construction, and snow accumu- accumulations of fuels that can lead to a possible lation and snowmelt in the thinned areas (Chang stand replacing wildfire and the destruction of 2003). Gottfried (1991) reported that in mixed coni- homes and other properties in the wildland -urbanfer stands within the Thomas Creek watershed in interface as well as in the forest. In addition, be- the White Mountains of Arizona, annual runoff cause fire is known to decrease the rate of waterincreased by 45 percent and mean peak discharges infiltration into soil and thus cause an increase inincreased by 60 percent in thinned areas in the surface runoff, large -scale flooding is an immedi- winter. These increases were attributed to lower ate post -fire concern (Neary et al. 2003; Fisher and evapotranspiration and greater snow accumulation and melting rates than in pre -thinned and har- 1Northem Arizona University, Flagstafff vested stands. According to Ffolliott and Fogel 22 Leao and Tecle

Figure 1. Location of the upper Rio de Flag watershedwithin Arizona and Flagstaff.

(2003), the relationships between watershedman- and endangered species, diseases and pathogens, agement practices such as thinning and peak dis- riparian resources, and fire hazard. Asa result, charge are difficult to isolate and quantify. Peak forest thinning goals include the removal ofsup- discharges can decrease or can remain unchanged pressed and intermediate trees toopen up forest after thinning, depending on thinning intensity, stands and restore the forest condition tonear pre - location in the watershed, regrowth of vegetation, European densities, structure, composition, and and climatic patterns (Chang 2003). function (Fort Valley Ecosystem Restoration Project 2002). In addition, prescribed fire will be Study Objectives incorporated on 2990 acres after certain standsare Our objective was to compare the predicted thinned (Newbauer, personal communication peak discharges under various simulationsof 2004). The current rate of thinning in the forestis severe wildfire with a simulation of forest thinning less than 1000 acres per year,or approximately 3 in the Rio de Flag watershed using the U.S. Army percent of the total watershed area, due to defer- Corp of Engineers, Hydrologic Engineering Cen- rals for wildlife, research projects, and inoperable ter's Hydrologic Modeling Software (HEC -HMS; areas. In addition, managers are thinning stands to Scharffenburg 2001). We also compared the scale varying degrees of density. Standtreatments of the predicted peak discharges relative to actual include light, moderate, and intense thinningto historic floods in Flagstaff. We expected modeling approximately 120, 90, and 60 sq ft of residual results to show that wildfire produces significantly basal area respectively. Newbauer (personal higher peak discharges than forest thinning during communication 2004) predicts that treatments in larger storm events. Also,a wildfire over the entire the watershed may not be completed until 2009. watershed could produce peak discharges ofa To err on the side of not underestimating peak magnitude never seen before in Flagstaff. discharge, we assumed that a total of 10,000acres in Fort Valley was treated to the intense level with- Forest Thinning in a short time period of 3 years. Reasons for this Scenarios for forest thinning incorporated the were (1) 10,000 acres is the expected area of treat- current stand treatments in 10,000 acres of the Fort ment, (2) water yield should increase incases Valley Experimental Forest as a part of the Greater where basal area reduction is greater than 20per- Flagstaff Forests Partnership. Thepurpose of these cent and understory vegetation recovery is limited treatments is to facilitate the restoration of ecosys- (Chang 2003), and (3) forest thinningactivities tem health as it relates to tree vigor, threatened must cover large areas for any impacton peak Predicting Peak Discharge in Rio de Flag Watershed 23

discharge to occur (Hewlett 1982). Therefore,we surface runoff from the thinning and three wildfire assumed that if about 30 percent of the watershed scenarios under dry, average, and wet moisture would be thinned promptly, the subsequent peak conditions. The NRCS models are categorizedas discharges from thinning would not be influenced event, lumped, empirical, or fitted parameter by factors such as the reestablishment of under- models (Scharffenburg 2001). We used NRCS story vegetation or human development. methods because they are easy to apply,are the most widely used by hydrologists, andwere Wildfire developed to evaluate downstream impacts from Forest conditions in the area indicatea potential various management treatments (Woodward et al. for an ignition of a wildfire within the watershed 2002). The NRCS Curve Number Loss Model, or outside of it (Neary et al. 2003). The probability taken from Scharffenburg (2001) and NRCS (1986), of a certain size event occurring dependson fac- estimates precipitation excess as a function of tors such as weather, fuel conditions, and topogra-cumulative precipitation, soil cover, landuse, and phy. Severe wildfires can consume much of the antecedent moisture using the following equation: forest vegetation, reducing both rainfall interception by the forest canopy and evapotranspiration Q= (P- Ia)2 /P -la +S (Robichaud and Waldrop 1994). While currentwhere Q = accumulated precipitationexcess (run- drought conditions increase the probability fora off); P = accumulated precipitation depth; IQ= severe fire, it is extremely difficult to predict a initial abstraction (initial loss); and S= potential specific fire event and the degree at which vegeta- maximum retention, a measure of the ability ofa tion and soils are affected. watershed to abstract and retain storm precipita- The effects of fire on peak dischargesare mostly tion. Until the accumulated rainfall exceeds the determined by fire severity and post -fire precipita-initial abstraction, the precipitationexcess, and tion regime, making it impossible to determinehence runoff, will be zero. changes in peak discharge from all conceivable An empirical relationship of IQ and S is situations (DeBano et al. 1998). Therefore,we created three generic wildfire scenarios related to IQ =0.2S. watershed area burned at the highlysevere level. The maximum retention, S, and watershed charac- These scenarios were one -quarter,one -half, and teristics are related through an intermediatepa- the entire watershed burned under asevere, stand - rameter, the curve number (CN) where replacing fire event. S= (1000 /CN) -10. METHODS The curve number for a watershedcan be esti- The upper Rio de Flag watershedwas first mated as a function of land use, soil type, andan- delineated using a 10 m Digital Elevation Model tecedent soil moisture. The curve numbers for this (DEM) obtained from the Arizona Regional Image study are shown in Table 1. Curve numbers in Archive (http: / /aria.arizona.edu /; accessed in AMC II can be converted using the following 2004), and the CRWR Preprocessor extension in equations: ArcView (ESRI 2002; Maidment and Djokic 2000). CN(I) = 4. 2CN(II) / 10- 0.05SCN(II) Three subwatersheds gauged by the U.S. Geologi- cal Survey (USGS) for 11 years were also deline-and CN(III) = 23CN(II) 110 + 0.13CN(II). ated in order to help calibrate the surface runoff Hydrologic soil groups are defined in Table 2. model. Watershed characteristics suchas soil type, Antecedent moisture conditions (AMCs), which stream patterns, vegetation type, slope, and land reflect seasonality and 5 -day antecedent precipi- use characteristics were obtained in the form of tation conditions, are described in Table 3. shapefiles from the Arizona Land Resource Infor- Although findings from Johnson (1998) show mation Service and Coconino National Forest that the use of NRCS methods onsome watersheds databases. After all watershed data were gathered, in excess of 96.5 can be appropriate, Ponce (1989) we used the Army Corp of Engineers surfacerecommended using careful judgment whenap- runoff modeling software, HEC -HMS. With this plying NRCS methods in excess of this limit and software we employed Natural Resource Conser- suggested subdivision of larger watersheds. The vation Service (NRCS, formerly Soil Conservation area of the subwatersheds in this study is sufficient Service) methods (Scharffenburg 2001; NRCS 1986; to determine peak discharge because the variabil- McCuen 1982) to determine the event -based ity in curve number for the largest subwatershed 24 Leao and Tecle

Table 1. Runoff curve numbers for AMC II (NRCS 1986).

Hydrologic Soil Group Hydrologic Impervious Cover Type Condition Area ( %) A B C D

Forests Poor 45 66 77 83 Fair 36 60 73 89 Good 25 55 70 77 Pasture, grassland, or range2 Poor 68 79 86 89 Fair 49 69 79 84 Good 39 61 74 80 Meadow - continuous grass, no grzing 30 58 71 78 Farmsteads 59 74 82 86 Streets and roads 98 98 98 98 Paved 76 85 89 91 Gravel 72 82 87 89 Dirt Commercial and business 85 89 92 94 95 Residential lots 1/8 acre or less 65 77 85 90 92 1/4 acre 38 61 75 83 87 1/3 acre 30 57 72 81 86 1 /2 acre 25 54 70 80 85 1 acre 20 51 68 79 84 2 acre 12 46 65 77 82 1Poor: Forest litter, small trees, and brush are destroyed by disturbanceor regular burning.Fair: Evidence of grazing but no burning, and some forest litter covers the soil. Good: No evidence of grazing, and litter andbrush adequately cover the soil. 2Poor: < 50% ground cover or heavily grazed withno mulch. Fair: 50-75% ground coverand moderately grazed. Good: > 75% ground cover and lightly grazed.

Table 2. NRCS hydrologic soil groups (Scharffenburg 2001). (29.9 sq mi) is relatively small, which fails to dis- courage application of NRCS methods over large Minimum areas (Hawkins, personal communication 2004). Soil Infiltration Group Description Rate (in /hr) Description of Subwatersheds A Deep sand, deep loess, aggregated 0.30-0.45 It is important to point out that most of the peak silts discharge data and general watershed description B Shallow loess, sandy loam 0.15-0.30 information we gathered for the three subwater- C Clay loams, shallow sandy loam, 0.05-0.15 sheds comes from the USGS Water Resources soils low in organic content, soils Investigations Report 87 -4210. In this report Hill et high in clay al. (1988) give regression functions for peak dis- D Soils that swell significantly when 0.00-0.05 wet, heavy plastic clays charges, historically estimated and recorded peak discharges, vegetation characteristics, channel characteristics, and hydrologic soil groups. The three subwatersheds gauged by the USGS used for Table 3. Definitions of the three AMC (Antecedent Mois- this study are Hidden Hollow, Schultz Canyon, ture Condition) groups (NRCS 1986). and Crescent Drive.

Total 5-day antecedent rainfall (in) Hidden Hollow AMC Group Dormant Season Growing Season The Hidden Hollow watershed encompasses 29.9 square miles (Figure 2). As the largest sub- I <0.5 <1.4 watershed, Hidden Hollow consists of a large II 0.5-1.1 1.4-2.1 forested and rural valley bounded by mountains to Ill >1.1 >2.1 the north and south. The main stream channel is Predicting Peak Discharge in Rio de Flag Watershed 25

Mies 0 1.25 25 5 7.5 10

Figure 2. Map of the Rio de Flag watershed showing the threesubwatersheds and their juxtaposition with respect to the San Francisco Peaks, Flagstaff, Fort Valley, and Interstate40 and Highway 180. 26 Leao and Tecle

approximately 8 miles long and has an average valley and stream. The gentle slopes and pervious slope of 2.4 percent. About three- quarters of the soils along the stream produce low runoff. vegetation in the area is ponderosa pine and pine /oak forest. Meadows covered with grasses Surface Runoff Modeling and herbaceous plants are the next -abundant The construction of a surface runoff model vegetation type. In the higher elevations, mixed begins with the selection of model components. conifer and subalpine conifer stands are present.These are then assembled as parts of the overall Land use in the watershed consists mostly of small model, following a logical sequence that resembles ranches, scattered residential areas, and national that of the natural processes. Rainfall and snowfall forest land. The amount of impervious area in the are considered first, followed by hydrologic ab- watershed is estimated, using Geographic Informa- stractions, subwatershed hydrograph generation, tion Systems (GIS) software, to bé approximately 2 stream channel routing, and hydrograph aggrega- percent. Hydrologic soil groups range anywhere tion in a cascading manner at the stream network from B to D, with B soils dominating areas along confluences (Scharffenburg 2001; Ponce 1989). the stream. Soils in group B have a moderate to Surface runoff modeling using HEC -HMS re- high rate of infiltration when thoroughly wet (Van quires three general components: a basin model, a Mullein et al. 2002; NRCS 1986), and according to meteorological model, and a control specification. Hill et al. (1988), this watershed producesvery The basin model requires information suchas little runoff. watershed area, infiltration capacities, NRCScurve numbers, surface imperviousness, lag time, base- Schultz Canyon flow, and channel roughness or routingparame- Schultz Canyon, which encompasses 6.4square ters. The meteorological model requires informa- miles, drains a rural pine forest on the steep south- tion such as storm type and duration, and rainfall east slope of the San Francisco Peaks (Figure 2). depth. When basin and meteorological models are The main tributary to the Rio de Flag is about 6complete, a control specification can be set and an miles long with an average slope of 5.6 percent. event is generated to determine the event dis- The Schultz Canyon tributary meets the main Riocharge. The control specification depicts the de Flag stream channel near the northwest Flag-resolution of the hydrograph calculation (Ponce staff city limits. Despite steep slopes, little runoff 1989). The control specifications we used were at occurs from Schultz Canyon. The vegetation types 15 minute intervals. are similar to Hidden Hollow vegetation. Land use in the watershed consists of scattered residential Establishing Current Conditions areas and national forest land. The impervious Using information from Hill et al. (1988) and cover in the watershed is estimated at 1 percent. GIS data, we determined NRCS parameters to be Hydrologic soil groups are of the B and C typesused in the HEC -HMS model. Table 4 shows and are highly pervious. watershed parameters using NRCS methodology for current conditions, the intense thinning treat- Crescent Drive ment, and the three wildfire scenarios. Watershed The Crescent Drive subwatershed encompasses parameters were also subdivided into three AMCs: 16.9 square miles and includes the areas south of AMC I for dry conditions, AMC II for averagecon- Hidden Hollow and Schultz Canyon, and anarea ditions, and AMC III for wet conditions to reflect intervening between these two watersheds (Figure seasonality and 5 -day antecedent precipitation 2). The intervening area is forested with thesame conditions (NRCS 1986; Ward and Elliot 1995). vegetation as Hidden Hollow and Schultz Canyon. Testing the model under all AMCs is significant to Housing developments and urban developmentthis study because (1) return intervals for dis- are abundant at the lower end of the watershed. charge and rainfall events published by Hill et al. The main channel length is about 4 miles withan (1988) and the Army Corp of Engineers (2000) do average slope of 2 percent. Land use in the water- not coincide (Tables 4 and 5), and (2) AMC is a shed includes more than 1000 acres of residential consistent factor in explaining deviations from the and urban development; the rest is national forest central trend of runoff. For example, precipitation land consisting of roads, trails, and recreationfrom the previous 5 days may help explain the areas. The impervious cover in the watershed is reason why a rainfall event with a 25 yr return estimated at about 4 percent. Soil groups are most- interval produced a 100 yr flood in September ly C, but B soils are commonly found along the 1923. Predicting Peak Discharge in Rio de Flag Watershed 27

Table 4. NRCS parameters used for the HEC -HMS surface runoff model for the current watershed condition, wildfire scenarios, and the intense thinning treatment. The parameter values were generated for all three AMCs. The three wildfire scenarios are severe burning to the entire area, to half, and toa quarter of the watershed. Intense thinning treatment is thinning all 10,000 acres of the Fort Valley area in the watershed to 60 sq ft of basal area.

Initial Loss (inches) SCS Curve Number Transform: Impervious - SCS Lag AMC I AMC II AMC III AMC I AMC II AMC III ness ( %) Time (min) Current Condition Hidden Hollow 1.97 1.75 125 33 52 68 2 120 Schultz Canyon 2.7 2.5 1.6 31 50 66 1 75 Crescent Drive 1.97 1.75 1.25 33 53 69 4 100 Wildfire (entire) Hidden Hollow 1.35 1.05 0.85 54 73 79 2 100 Schultz Canyon 1.7 1.5 1.1 51 70 77 1 70 Crescent Drive 1.35 1.05 0.85 56 74 78 4 85 Wildfire (half) Hidden Hollow 1.72 1.49 1 40 60 73 2 110 Schultz Canyon 2.6 2.1 1.4 38 58 70 1 72 Crescent Drive 1.72 1.49 1 41 61 72 4 92 Wildfire (quarter) Hidden Hollow 1.9 1.62 1.15 36 56 71 2 115 Schultz Canyon 2.65 2.3 1.45 34 54 70 1 74 Crescent Drive 1.9 1.62 1.15 37 57 71 4 95 Intense Thinning Hidden Hollow 1.94 1.7 1.2 39 59 70 2 117 Schultz Canyon 2.7 2.5 1.6 32 51 68 1 75 Crescent Drive 1.95 1.73 1.22 38 58 71 4 98

Parameters for the watershed were first deter- storm events (Riley 1998). We used lag time as the mined for the current conditions under AMC II time interval from the maximum rainfall rate to the and then adjusted according to historic peak dis- peak rate of runoff (Viessman and Lewis 1996). charges on record. The basic procedure in calibrat- The amount of impervious area was determined ing the model was one of selective parameter value using total area of development and average manipulation and comparison of simulated flow impervious values for the area taken from McCuen values against recorded peak discharges. For (1982). The impervious portion in developed areas example, if the simulated discharge was less than ranges from 12 to 85 percent. Baseflow estimates recorded, the value of initial loss was adjusted to a were not necessary because the streams in the smaller value to increase the simulated runoff. Westudy area are ephemeral and do not have base- used initial loss to describe surface storage, inter-flow (Hill et al. 1988). After determining flow ception, and infiltration in inches prior to runoffcharacteristics under the current conditions, we (Ward and Elliot 1995). used watershed parameter values that explain Other parameters were manipulated as well. An forest watershed conditions under thinning and important consideration in determining the param-wildfire for input into the NRCS model. eters for the current watershed condition is the spatial modification of the watershed since the last Thinning Scenario large flood in 1993. Since 1993, there have been Based on the NRCS hydrologic soil complexes numerous housing and road developments in the for forest land in the fair condition vegetation area. As a result, impervious areas in the studycover class, the curve number was increased from area have increased. This may reduce infiltration the current condition to reflect the possible change rates and lag times, resulting in higher peakin runoff characteristics in the thinning scenario discharges than previous floods having similar(Chang 2003). It is important to note that the soils 28 Leao and Tecle

Table 5. Historic peak flows and complementing rainfall depth with corresponding return interval (RI),antecedent soil moistures depth (ASM) from previous 5 days,antecedent moisture condition (AMC), andevent type in Fort Valley.2 Peak discharges ranged from3 to 1200 cfs (1920 -1993).

Peak Flow Event Rainfall Date (cfs) /RI (yr) Depth (in) /RI (yr) ASM AMC Event Type 02/20/1993 900/50 2.5/25 1.04 II Rain on snow 03/12/1982 240/10 2.3/25 0 I 07/26/1980 Snowmelt 104/5 0.8/2 0.34 I Rain 05/21/1979 90/5 0.6/2 0 I Rain 04/02/1978 128/10 0.75/2 1.07 I 05/15/1977 Rain 10/2 0.45/2 0.53 I Rain 02/09/1976 40/2 1.1/2 1.64 III Snowmelt 04/08/1975 10 /2 0.3/2 0.38 I Rain on snow 04/03/1974 3/2 0.11/2 0.43 I Rain 04/28/1973 235/ 10 0.6/2 0 I Rain on snow 09/30/1971 10/2 1.52/5 0.33 I Rain 08/30/1970 10/2 0.15/2 0.07 I Rain 08/05/1963 300/25 0.5/2 1.33 I Rain 03/24/1960 11/2 0.45/2 0 I Rain 04/20/1958 56/2 0.15/2 0 I Rain 03/04/1938 600/50 1.45/5 1.4 III Rain 09/18/1923 1200/100 2.33/25 1.5 II Rain 02/22/1920 600/50 1.42/5 1.48 III Rain on snow 1ASM is the precipitation depth within 5 days leading up to the storm. ASM determines AMC.AMC I = Dormant season ASM less than 0.5 in; growing season ASM less than 1.4 in. AMCII = Dormant season ASM between 0.5 and 1.1 in; growing season ASM between 1.4 and 2.1in. AMC III = Dormant season ASM greater than 1.1 season ASM greater than 2.1 in. in; growing 2Precipitation depths and eventtypes are for the Fort Valley weather station.

in study subwatersheds do not belongto one soil cause an increase in the velocity of overland flow group, and the current thinning treatment is not that enters stream channels (DeBanoet al. 1998). conducted in the entire watershed. Asa result, the The curve number generated for wildfire burned curve number generated is an estimation of what areas was based on NRCS hydrologic soil com- may actually occur in the watershed from intenseplexes for forest land in poor condition because thinning. Other parameters suchas initial loss and wildfire can destroy much of thevegetation, lag time were decreased accordingly. The Schultz leaving the soil bare and dry. In dry soilcondi- subwatershed received minor adjustmentin its tions, water repellency occurs at shallow depths curve number because the actual thinning taking (Robichaud and Hungerford 2000). place in the subwatershed involves mostlyurban interface thinning, not open forest thinning.As a Meteorological Model result the curve number was increasedby 1 to 2 According to Viessman and Lewis (1996) and points because thinning affectsa minimal area. Ponce (1989), an appropriate model of rainfall distribution pattern for the study Wildfire Scenarios area is the Type II SCS 24 hr rainfall distribution (Figure 3),which Parameters for the wildfire scenarios in Table4 we used for this study. The Type II rainfall distri- were decreased for initial loss and lag time andbution is representative of stormsystems that increased for the curve number. The initialloss occur in the southwestern United States (Ponce was used as the main factor in determining wild- 1989). Twenty -four hour, basin -wideprecipitation fire impacts because experimentsperformed by depths of 2, 5, 10, 25, 50, and 100yr return periods Robichaud (2000) suggested that infiltrationrates for the area were obtained from the U.S.Army can decrease up to 40 percent, and this value was Corp of Engineers feasibility report for the Riode used to estimate the initial loss. Lag timewas Flag (2000). Figure 4 shows the 24 hrprecipitation decreased because a lack of vegetation would depth for each of the six return periods. Thedepth Predicting Peak Discharge in Rio de FlagWatershed 29

1- 0.9 - 0.8 - 0.7 - 0.6 - 0.5 - 0.4 - 0.3 - 0.2- 0.1 - 0-

Figure 3. SCS Type Il rainfall distribution model for a 24 hour storm (ordinates takenfrom McCuen 1982).

át d 4- o 3.5 - Ço 3- 2.52- - Lc1.5- °-L 1 - = 0.5 - 4 2 -year 5 -year 10 -year 25 -year 50 -year100 -year Frequency

Figure 4. Twenty-four hour, basin -wideprecipitation depths for 2, 5, 10, 25, 50, and Arizona (Army Corp of Engineers 2000). 100 yr return periods in Flagstaff, 30 Leao and Tecle

of rainfall ranges from 1.3 inches for a 2 yr storm is greatest at approximately 260cfs in the wildfire event to 3.4 inches for a 100 yr storm event. scenario where the entire watershed isburned. Also, the discharge is greater Reconstruction of Historical Floods with among all wildfire Respect to Thinning and Wildfire scenarios under wet conditions foran extended period of time, compared with thinningand cur- To reconstruct historical floods withrespect to rent condition scenarios. This extended thinning and wildfire we employed the discharge basin and is a common trend shown by allhydrographs in meteorological models describedpreviously; every return period. however, we used the actualstorm depth from In Figure 6, a 5 yr precipitation past storms (Table 5) instead of storm over a return period watershed area completely burned bya wildfire depths. One historical storm from eachstorm type produces peak dischargesup to 1100 cfs in wet was chosen. Storm events selected include heavy conditions. However, a 5 winter runoff from the February yr storm over the same 1993 flood, wildfire burned area underaverage and dry con- intense monsoon moisture fromthe July 1980 ditions produces a discharge flood, and a tropical storm from near 600 and 200 cfs the September respectively. Following a 5yr storm, a wildfire that 1923 flood. After choosing historicalstorm events burns over one -quarter of the watershed for the meteorological model, produces we computed the slightly greater discharge than thethinning treat- peak discharge using HEC -HMSbased on the ment. appropriate parameters in the basin model (Table In Figure 7, a 10 yr storm showsa difference in 4). For example, in the 1923storm event we used peak discharge among all scenarios 2.33 inches of precipitation and AMCs, as the storm depth with wildfire and wet conditionsproducing the (meteorological model) and lossrates, curve higher discharge. A wildfire burn numbers, and lag times for AMC over the entire II under thewatershed and a wet AMCproduce peak dis- thinning and wildfire scenarios.Reconstruction for charge of about 2500 cfs. The the February 1993 storm post -wildfire peak was a special case for discharges would range between400 and 1300 cfs modeling because antecedent snowmeltoccurred if the AMC is averageor dry. The peak discharges prior to the storm period, resultingin baseflow. from intensely thinned areas Therefore, for simulation range between 300 cfs purposes we included under average or dry AMC and 900cfs under the baseflow in the model at therate of 6 cu ft/sec /sq wet AMC. mi. This rate, which was suggested by the Army In Figure 8, a 25yr return period storm pro- Corp of Engineers (2000) in theirreproduction of duces a peak discharge the February 1993 flood, is derived more than twice the his- from estimated toric peak discharge in soils ofaverage AMC and components of shortwave radiation, groundheat, almost four times greater in soils of and convection -condensation melting. wet AMC and after a wildfire that burned theentire watershed. Again, a thinning treatment would produce RESULTS a peak discharge slightly higher than thatunder the cur- Surface runoff estimations using the HEC -HMS rent condition and slightly lower than wildfirethat modeling system produced consistentresults for burned one -quarter of the watershed. all return period storms. Figures 5 -10 show peak In Figure 9, a 50 yr storm producespeak dis- discharge hydrographs for 2, 5, 10,25, 50, and 100 charges of great magnitude. The yr return period storm events under three storm produces AMCs the highest peak discharges followingall wildfire and under current conditions,thinning treatment, scenarios under all AMCs. The peak and wildfire scenarios. The discharges hydrographs usually produced under these conditionsrange from 700 reached their peak flow rates 10to 20 hr after the to 6500 cfs. A 50 yr storm after intensethinning onset of a storm event. Also, dischargescontinued and under current condition would for approximately 36 hr after produce hy- the onset of the drographs that have maximum peakdischarges of storm. It is important to note that forinterpreta- 3300 and 2800 cfs, respectively. tion, AMC discharges in all hydrographs are the Figure 10 shows a 100yr storm producing peak same as AMC II discharges unless the hydrograph discharges from a watershed entirelyburned by shows otherwise. wildfire to be about 4.5 times Figure 5 shows that the peak discharges greater than the his- under toric discharge in theaverage AMC and 6.5 times AMC I and II for a 2yr magnitude storm are the greater in the wet AMC (Table 5). The same, at less than 200 cfs. The 2 yr peak discharge peak discharges following wildfire burnrange from 1000 to Predicting Peak Discharge in Rio de Flag Watershed 31

Current Condition Thinning Treatment Wildfire (one -quarter) 300 300 300 ÿ 250 w 250 250-` v 200 200 200 á: 150 150 150 ó 100 100 100 w 50 50 50 o 0 0 0 1020 30 40 0 10 20 30 40 010 20 30 40 Time (hours)

Wildfire (half) Wildfire (entire area) 300 300 250 250 200 200 Average 150 Dry 150 - --- Wet 100 100 50 50 0 0 0 10 20 30 40 0 10 20 30 40

Figure 5. Hydrographs from a 2 yr storm showing peak discharge less than 260cfs following the worst scenario of entire watershed burn.

8000 cfs under dry to wet AMC and small to large DISCUSSION fire extent. The peak discharges produced under The results of hydrologic modeling for the Rio intense thinning and current condition and wet de Flag watershed may have some important AMC are 1300 and 1100 cfs, respectively. implications for decision makers. These modeling The reproduction of three historical floodspro- results provide information on the amount of duced mixed results. In Table 6, peak dischargesdischarge coming from the watershed and entering from rainfall falling on top of melting snow anda the city of Flagstaff under existing conditions. tropical storm (Feb 1993 and Sep 1923 respectively) Results also serve as the basis for watershed man- were significantly increased from their historical agement decision making by pointing out the amount after wildfires. Following the 1923 event, possible outcomes for various levels of area that there was no increase in peak discharge attributed the forest burns. This will help determine the loca- to intense forest thinning. However peak dis-tion and level of treatment to conduct to prevent charges from the 1993 event were slightly higher wildfires, and will provide needed informationon after thinning the forest. Model results for the July the hydrologic effects of the treatments. 1980 flood were significantly less than the actual Increases in discharge following wildfire and event, by more than 475 cfs, and did not change thinning were similar to peak discharges reported throughout all scenarios. by Robichaud et al. (2000), Baker (1984), and Gott- 32 Leao and Tecle

Current Condition Thinning Treatment Wildfire (one- quarter) 1200 1200 1200 s 1000 1000 1000 800 800 800 600 .600 600 c400 400 400 . 200 200 . 200 0 0 10 20 30 40 0 10 20 30 40 0 1020 30 40

Wildfire (half) Wildfire (entire area) 1200 1200 1000 1000 800 800 !+ Average 600 600 Dry 400 400 Wet 200 200 0 0 0 10 20 30 40 0 10 20 30 40

Figure 6. Hydrographs from a 5 yr storm following a wildfire that burned the entire watershed, producing significant flows of about 1100 cfs under wet antecedent soil moisture conditions (AMC) while storms after intensely thinned forest and following fires of lesser extent under all AMC types produce smaller peak discharges of less than 600 cfs.

Table 6. Estimated peak discharge using NRCS methods in HEC -HMS software for historicalstorm events under wildfire, thinning, and current condition scenarios.

Reconstructed Storm Discharge (cfs)

Scenario Feb 93 Jul 80 Sep 23

Wildfire (entire area) 3143 119 3021 Wildfire (half) 1665 119 1815 Wildfire (quarter) 1280 119 1308 Thinning 1126 119 1175 Current condition 968 119 1010 Predicting Peak Discharge in Rio de Flag Watershed 33

Current Condition Thinning Treatment Wildfire (one -quarter) 2500 2500 2500 w 2000 U 2000 2000 1500 1500 1500 1000 1000 1000 0 I' w 500 500 ¡'% 500 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Time (hours)

Wildfire (half) Wildfire (entire area) 2500 2500 2000 2000 Average 1500 1500 Dry 1000 1000 - - -- Wet - - Historic Peak 500 500 (Sep -1923) o 0 0 10 20 30 40 010 20 30 40

Figure 7. Hydrographs for a 10 yr storm show the peak discharges to be different under varying treatment scenarios and antecedent soil moisture conditions (AMC). The peak discharge following a wildfire that burned the entirewater- shed is approximately 2500 cfs. Other scenarios under average and dry AMC produce discharges between 300 and 1400 cfs after a 10 yr storm.

fried (1991). Precipitation from storms following trees as well as the understory in the affected area, wildfire and intensive forest thinning have the whereas forest thinning only removes trees desig- potential to yield peak discharges never recordednated for cutting and has less impact on the over- before. As the hydrographs show, peak discharges story and understory vegetation (Neary et al. 2003; after wildfires are very high compared to those DOE 2000). after treatments. Our analysis also shows that the Two factors that influence the values and inter- impacts of intensive forest thinning on floodingpretation of the modeling results are AMC and are small compared to those of wildfire. As such, return period. Table 5 shows that the majority of less intensive forest thinning distributed over 5 flood events in the Rio de Flag occurred when the years or more should have less impact on flooding. AMC was dry, and none of them were greater than The higher peak discharges after wildfire are due a 25 yr storm depth. The less frequent and larger to the fire's impact on soils and vegetation. Water - floods occurred over average and wet AMCs. repellent soils are common after severe wildfires, Consequently, the larger floods have a higher whereas forest thinning only increases soil bulk likelihood of occurring during tropical storms or density. Furthermore, severe wildfires may kill all winter snow and snowmelt events, at less than or 34 Leao and Tecle

Current Condition Thinning Treatment Wildfire (one- quarter)

4000 4000 4000

3000 3000 3000

a; 2000 2000 2000

If:1000 1000 1000

0 0 0 010 20 30 40 0 10 20 30 40 010 20 30 40 Time (hours)

Wildfire (half) Wildfire (entitre area)

4000 4000 Average 3000 3000 Dry - - - - Wet 2000 2000 - - -- Historic Peak (Sep -1923) 1000 1000

0 0 0 10 20 30 40 0 10 20 30 40

Figure 8. Hydrographs from a 25 yr storm over a wildfire that burned the entire watershed, producingdischarge values greater than 2-4 times the historic peak discharge of 1200 cfs under average and wet antecedent soil moisture conditions (AMC). A similar storm over intensely thinned forest and any AMC type produces discharge at least 200 cfs below those over a wildfire that burned one -quarter of the watershed.

equal to25 yr storm events. Storms over averagerest of the treatment scenarios. Storms over other and wet AMCs have resulted in some of the largest scenarios would yield the same discharge regard- historical floods (see Table 5). However, most ofless of the type of AMC. This is probably due to the peak discharges during historical flood events the high infiltration capacity of soils in the water- in the period of data measurement occurred overshed. Also, forested watersheds reduce peak dry AMCs. In the event that one of the six treat- discharges for storms of low intensity and short ment scenarios occurs within the watershed, we duration, but cannot prevent the occurrence of recommend that scientists and decision makers floods from storms of higher intensity and long interpret our modeling results as a range of pos-duration over a large area (Chang 2003). A 10 yr sible discharge values over dry and wet AMCs. storm over a wildfire burned area can produce For smaller and more frequent storm events peak discharge greater than that of the 1923 peak such as 2 yr storms, peak discharges after wildfire discharge of about 1200 cfs. According to Gifford and thinning do not seem to have great impact. et al. (1967) the occurrence probability for a 10 year According to the model, a wildfire will increase storm is during a period of one week in early the amount of discharge by 60 cfs compared to theAugust. All peak discharges from a 100 yr storm Predicting Peak Discharge in Rio de FlagWatershed 35

Current Condition Thinning Treatment Wildfire (one -quarter)

6000 6000 6000

4000 4000 4000 r + ó 2000 i iI w_..__1. 2000 2000 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Time (hours)

Wildfire (half) Wildfire (entire area)

6000 e 6000 n n Average 1 1 I 4000 4000 - - --W 2000 _.._.._ Historic Peak 2000 (Sep -1923)

0 010 20 30 40 010 20 30 40

Figure 9. Hydrographs from a 50 yr storm showing a gradual increase with changes in treatment from currentcondi- tion to wildfire over the entire watershed. Dischargeafter intense forest thinning is below the smallest and above the current condition. wildfire scenario

event exceed 1100 cfs. Moreover, the peak dis- with several feet of water. It isestimated that charge following the worst -case scenario of wild- damages could total about $93 million ifa 100 yr fire over the entire watershed couldbe 2 to 6.5event, such as the September 1923 flood (Army times greater (depending on AMC) than the 1923 Corp of Engineers 2000). Under the wildfireand flood. Discharges in that range may cause exten- thinning scenarios, Flagstaff could experiencea sive damage over parts of Flagstaff, especially if flood of greater intensity and volume.According Flagstaff continues to encroach further intothe to the modeling results shown in Table 6,two floodplain. The most likely periods for theoccur- possible historical eventspose the greatest risk to rence of a 100 yr storm event are during the Flagstaff if similar events occur in the future. These months of July, August, December, January, andextreme events are rainfall on top of meltingsnow February (Gifford et al. 1967). Althoughthe and a severe tropical storm. From the model,both probability of a 100 yr event occurring isslight, it events produced discharges between 1200 and is essential to not underestimate unpredictable 3100 cfs following wildfire. Stormsover intensely weather patterns due to future climate changes. thinned forest produced discharge lessthan 1200 In previous flooding, large portions ofFlagstaff cfs. Costs from a flood caused byan extreme storm have suffered damages and have beeninundated event after a wildfire may significantly exceed $93 36 Leao and Tecle

Current Condition Thinning Treatment Wildfire (one- quarter)

8000 8000 8000 .°, 6000 6000 6000 fg 4000 4000 i 4000 i¡ 2000 2000 2000

0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Time (hours)

Wildfire (half) Wildfire (entire area)

8000 8000 - 6000 6000 - Average Dry 4000 4000 - - - -- Wet -. -. - Historic Peak 2000 2000 (Sep -1923) o 0 0 10 20 30 40 0 10 20 30 40

Figure 10. Hydrographs from a 100 yr storm showing peakdischarges after a wildfire burned the entire watershed between 2 and 6.5 times greater than the historic peaks in allantecedent soil moisture conditions (AMC). Under average AMC the peak discharge from the current condition is equivalentto the historic mark of 1200 cfs and slightly higher at about 1500 cfs in the intensely thinned forest scenario.

million, and the costs of flooding froman extreme Another requirement of the model is theas- storm after forest thinning could possibly be less sumption of uniform precipitation distribution than $93 million. across the watershed. Precipitation does not fall The utility of the HEC -HMS model for predict- uniformly over an area because elevation and local ing an event -based peak discharge in theupper micro -site conditions play an important roleon the Rio de Flag watershed may havea number of distribution of precipitation. This is especially the weaknesses that need proper consideration. Be- case in Flagstaff because of the rugged topography cause the model is based on NRCS curve numberalong the slopes of the San Francisco Peaks and methods it may not have an adequate mechanism seasonal and local variation from frontal and to account for the high degree of ground water convectional storm activity. However, tropical and percolation in the area. The model indirectly winter storms typically cover largeareas and are attempts to account for the abundant faults and usually of long duration and lower intensity than fissures within the watershed throughsome summer storms (Beschta 1976). Consequently, this adjustment of the runoff curve number. However, may explain why the model underestimated the the latter may not adequately explain the rapidreconstructed discharge of the July 1980 storm infiltration rate in some areas of the watershed. event. Predicting Peak Discharge in Rio de Flag Watershed 37

Finally, SCS methods to predict runoffare a The Rio de Flag watershed is constantly chang- result of empirical data collectedover much of the ing its structure due to the dynamic environment United States, and developed for smallwater-around Flagstaff. We can expect variationsin sheds. As a result, local conditions for the study hydrologic phenomena over time, and continuous area may not be completely or accurately ac-monitoring of management practices, landuse, counted for. and the hydrology of the watershedare essential. In addition, small -scale collaborativeprojects to MANAGEMENT RECOMMENDATIONS bring the community together should be imple- Several recommendations should be carriedout mented. These projects could include improving to improve upon watershed modeling in the Rioland management and developmentpractices, de Flag. Other methods should be used to calibrate planting riparian species, removing exoticspecies, the current conditions of the watershed andthen and creating or enhancing greenbelts and wetlands compare the precision of each method to real -time along the stream channel (Riley 1998; Williamset storm data. This will facilitate finding the best al. 1997). method for use in the HEC -HMS modeling software. Also, older gauging stations should be LITERATURE CITED reinstated and new gauging stations should be Baker, M. B., Jr. 1984. Effects of ponderosa pinetreat- installed in smaller watersheds usinga paired - ments on water yield in Arizona. Water Resources watershed approach (Viessman and Lewis 1996). Res. 22: 67 -73. Beschta, R. L. 1976. Climatology of the ponderosapine The three watershed gauging stations (Hidden type in central Arizona. Technical Bulletin 228. Col- Hollow, Schultz Canyon, and Crescent Drive) lege of Agriculture, Agricultural Experiment Station, should be reinstated because 11years of gauging is University of Arizona, Tucson. 24 pp. not adequate to derive large flood events, and 11 Brown, H. E., M. B. Baker Jr, J. J. Rogers, W. P. Clary, J. L. Kovner, F. R. Larson, C. C. Avery, and R. E. Camp- years is a short period of record compared to the bell. 1974. Opportunities for increasing water yields long -term meteorological data collected for Flag- and other multiple use valueson ponderosa pine forest lands. Research Paper RM -129. USDA Forest staff. Paired watersheds will allow scientiststo Service, Rocky Mountain Forest and Range Experi- capture before and after effects from forest thin- ment Station, Ft. Collins, CO. ning or wildfires. Stream gauging stations should Burton, T. A. 1997. Effects of basin -scale timber harvest be installed in subwatersheds smaller than 100 on water yield and peak streamflow. Journal of American Water Resources Association. 33(6): 1187- acres and initially set aside from forest thinning. 1196. When paired watersheds are calibratedto one Chang, M. 2003. Forest Hydrology: An Introductionto another (minimum 3 yr), thinning should be al- Water and Forests. CRC Press, Boca Raton, Florida, lowed in one of the subwatersheds while the other 373 pp. Covington, W. W., P. Z. Fulé, M. M. Moore, S. C. Hart, T. is a control. In addition, soil infiltration tests E. Kolb, J. M. Mast, S. S. Sackett, and M. R. Wagner. should be conducted in control, thinned, and wild- 1997. Restoring ecosystem health in onderosapine forests of the Southwest. Journal of Forestry 95(4): fire affected areas to give scientistsa better under- 23-29. standing of impacts to soil physical characteristics. DeBano, L. F., D. G. Neary, and P. F. Ffolliott. 1998. Forest thinning should continue throughout the Fire's Effects on Ecosystems. John Wiley & Sons, urban interface and open forest in order toprevent New York. Department of Energy (DOE). 2000. Special environmen- wildfire. A few reasons to implement thinningas a tal analysis for the Department of Energy, National preventative measure are that it (1) protects Nuclear Security Administration, actions taken in human life and property from fire, (2) improves response to the Cerro Grande fire at Los Alamos National Laboratory, Los Alamos, New Mexico. U.S. forest health, (3) is less damaging to theecosystem Department of Energy, Los Alamos Area Office, Los than a severe stand -replacing fire, and (4) allows Alamos, NM. managers to easily incorporate natural and pre- Environmental Systems Research Institute (ESRI). 2002. ArcView 3.x. Computer Software. ESRI, Redlands, scribed fire into the ecosystem. After thinning is CA. accomplished and ladder fuels and stand densities Ffolliott, P. F., and M. M. Fogel. 2003. Connectingup- are reduced, prescribed fire should be imple- land watersheds to large river basins. Journal of the mented as a long -term management practiceto Arizona- Nevada Academy of Science 35(1): 71 -75. Fisher, R. F., and D. Binidey. 2000. Ecology and Manage- achieve "complete" forest restoration (Covington ment of Forest Soils. John Wiley & Sons, New York. et al. 1997). In the event of a wildfire, pinpointing Fort Valley Ecosystem Restoration Project. 2000. Envi- the area affected and its characteristicsmay im- ronmental Assessment for the Fort Valley Restoration prove the precision of estimating peak discharges Project. Available online at http/ /www.fs.fed.us/ r3/ coconino /nepa /ft_valley_01.html (accessed De- using HEC -HMS or other hydrologic models. cember 2004). 38 Leao and Tecle

Gifford, R. O., G. L. Ashcroft, and M. D. Magnuson. Robichaud, P. R., and T. A. Waldrop. 1994. A compari- 1967. Probability of selected precipitation amounts in son of surface runoff from low and high severity site the western region of the United States. Western Re- preparation burns. Journal of the American Water gional Research Publication T -8, Agriculture Experi- Resources Association 30(1): 27 -34. ment Station, University of Nevada. 26 p. Gottfried, G. J. 1991. Moderate timber harvesting in- Robichaud, P. R., J. L. Beyers, and D. G. Neary. 2000. creases water yields from an Arizona mixed conifer Evaluating the effectiveness of postfire rehabilitation treatments. General Technical ReportRMRS- GTR -63. watershed. Journal of the American Water Resources Association 27(3): 537 -547. USDA Forest Service, Rocky Mountain Forest and Hewlett, J. D. 1982. Principles of Forest Hydrology. Uni- Range Experiment Station, Fort Collins, CO. versity of Georgia Press, Athens. 183 pp. Scharffenburg, W. A. 2001. Hydrologic Modeling System Hill, G. W., T. A. Hales, and B. N. Aldridge. 1988. Flood HEC -HMS Users Manual, Version 2.1. U.S. Army hydrology near Flagstaff, Arizona. U.S. Geological Corp of Engineers Hydrologic Engineering Center, Survey, Water- Resources Investigations Report 87- Davis, CA. pp. 186. 4210.31 pp. Tecle, A. 1991. Hydrology and watershed management Johnson, R. R. 1998. An investigation of curve number in southwestern ponderosa pine forests. In Multire- applicability to watersheds in excess of 25000 hec- source Management of Southwestern Ponderosa Pine tares (250 sq km). Journal of Environmental Hydrol- Forests: The Status of Knowledge. U.S. Department ogy 6. of Agriculture, Forest Service, Southwestern Region. Maidment, D., and D. Djokic. 2000. Hydrologic and hy- 410 pp. draulic modeling support with geographic informa- U.S. Army Corp of Engineers. 2000. Rio de Flag, Flag- tion systems. Environmental Systems Research Insti- staff, Arizona Feasibility Report and Final Environ- tute, Redlands, CA. 216 pp. mental Impact Statement. U.S. Army Corp of Engin- McCuen, R. H. 1982. A Guide to Hydrologic Analysis eers, Los Angeles. Using SCS Methods. Prentice -Hall, Englewood Cliffs, URS Consultants. 1990. City of Flagstaff hydrology re- NJ. 145 pp. port for alternative flood control study for the Rio de Murray, M. R. 1974. Extreme Runoff Events- Rio de Flag. COF No. 03 -88023 (503 -68). Phoenix, Arizona. Flag: As Researched from Local Newspaper Head- lines. Flagstaff, Arizona. Van Mullem J. A., D. E. Woodward, R. H. Hawkins, A. Natural Resource Conservation Service (NRCS). 1986. T. Hjelmfelt, and Q. D. Quan. 2002. Runoff curve National Engineering Manual Section 4, Hydrology. number method: Beyond the handbook. Second Fed- U.S. Department of Agriculture, Beltsville, Maryland. eral Interagency Hydrologic Modeling Conference, Newry, D. G., G. J. Gottfried, L. F. DeBano, and A. Tecle. Las Vegas, NV. 2003. Impacts of fire on water resources. Journal of Viessman, W., and G. L. Lewis. 1996. Introduction to the Arizona- Nevada Academy of Science 35(1): 23- Hydrology, 4th ed. Harper Collins College Publish- 41. ers, New York. 760 pp. Nyland, R. D. 1996. Silviculture Concepts and Applica- Ward, A. D., and W. J. Elliot. 1995. Environmental Hy- tions. McGraw Hill, New York. 633 pp. drology. CRC Press, Boca Raton, Florida. 462 pp. Ponce, V. M. 1989. Engineering Hydrology- Principles Williams, J. E., C. A. Wood, and M. P. Dombeck, Editors. and Practices. Prentice Hall, Englewood Cliffs, NJ. 1997. Watershed Restoration: Principles and Prac- Riley, A. L. 1998. Restoring Streams in Cities: A Guide tices. American Fisheries Society, Bethesda, Mary- forPlanners, Policymakers, and Citizens. Island land. 561 pp. Press, Washington D.C. 423 pp. Robichaud, P. R. 2000. Fire effects on infiltration rates Woodward D. E., R. H. Hawkins, A. T. Hjelmfelt, J. A. after prescribed fire in northern Rocky Mountain Van Mullein, and Q. D. Quan. 2002. Curve number forests, USA. Journal of Hydrology 231 /232: 220 -229. method: Origins, applications and limitations. Robichaud, P. R., and R. D. Hungerford. 2000. Water Second Federal Interagency Hydrologic Modeling repellency by laboratory burning of four northern Conference, Las Vegas, NV. Rocky Mountain forest soils. Journal of Hydrology 231/232: 207 -219. USE OF THE ANALYTIC HIERARCHY PROCESS INFOREST BUDGET ALLOCATION IN DURANGO, MEXICO

Gustavo Perez- Verdin1 and Aregai Teclea

In the past, Mexican forest management deci-improve water quality or provide other environ- sion makers and consultants focused primarilyon mental services. During the year 2003,a combina- allocating funds only to manage timber- related tion of federal and state funds equivalent to US resources. Timber growth, cattle grazing, and $4.1 million supported 1200 projects in the state of agriculture have been the main activities in forest Durango alone. The aims of the projectscan be lands representing the primary sources ofa farm- classified into nine categories: timber management, ing community's livelihood. Recently, stakehold- environmental evaluation, skill and manpower ers have shown a growing interest in other attri- development, providing technical assistance, butes of the land, such as environmental services,reducing operational costs, enhancing plantations, water production, conservation of natural re-promoting recreation and ecotourism, managing sources, biodiversity, and cultural and recreationalnon -timber products, improving water quality and values. More attention has been paid to the inte-quantity, and providing other environmental grative management of natural resourcesas services. alternative sources of livelihood. In this context, According to Kangas (1994), there is limiteduse the concept of multiple -use forestry is being of proper analytical methods to multiobjective revised by federal and local governments tomore planning and budget allocation -especially meth- efficiently allocate funds for multiresource forestods that involve public opinion. The decision- management and to incorporate the views andmaking processes should involve all stakeholders' interests of all stakeholders in the management of wishes and aspirations in developing management natural resources. Representative stakeholders objectives and should exhaustively analyze all include common and private landowners, therelevant alternatives (Tecle 1992). For example, logging industry, NGOs, the general public, and landowners can use decision -making techniques to federal and local institutions. handle complex multiple resource management A considerable portion of the necessary funds problems to better serve societal needs. However, for managing Mexican natural resourcescomes there is a challenge to incorporate andmeasure all from the National Forest Commission (CONAFOR stakeholders' wishes and aspirations to determine is its Spanish acronym). CONAFOR is anew the most efficient alternative forest management Mexican federal agency created by a Presidential scheme. In this context, the theory of value has Decree on April 4, 2001. Its main goals are topro- been well studied because it prescribes the best mote efficient forest management and restoration decision alternative while involving middle- activities, as well as to enforce and monitorsus- ground terms such as "maximizing" valueor tainable forest development policy. To achieve "minimizing" costs (Kangas 1994). these goals, CONAFOR is currently using various In this paper we use the analytic hierarchy approaches such as funding projects that increase process (AHP) to evaluate stakeholder wishes and forest stock levels through the development ofmanagement objectives concerning projects plantations and management of native forests, funded by CONAFOR. The AHP (Saaty 1980)can fostering landowners' skills, promoting theman- evaluate such multiple- objective decision problems agement of non -timber products such as resins,by incorporating public values into the decision- oregano, and mushrooms, and recently, compen- making process (Schmoldt et al. 2001). Due to its sating landowners whose primary objective is to relative simplicity, effectiveness, and ability to deal with both quantitative and qualitative criteria, the 1School of Forestry, Northern Arizona University, Flagstaff AHP is used to determine the most efficient bud- 40 Perez - Verdin and Tecle

get allocation for forest multiobjective manage- The alternative that produces the highest utility ment planning. In the process, we first describe the value is the most preferred option (Schmoldt et al. essentials of AHP, and then we apply it to CONA- 2001; Tecle 1992). FOR's budget for managing forest resources ina Pairwise comparisons help the decision maker multiobjective decision -making framework. to see his or her preferences by comparing two elements at a time; the weights represent the METHODOLOGY decision maker's preference structure on the Kangas (1992, 1994, 1999), Saaty (2001), and objectives. Saaty (1988) used weights of 1, 3, 5, 7, Duke and Aull -Hyde (2002) have described the and 9 to respectively represent equal, weak, essen- various steps involved in the application of the tial, demonstrated, and absolute importance of one AHP technique to a multiobjective forest manage- objective over the other in a pairwise comparison. ment problem. The basic principle for structuring a The weights of 2, 4, 6, and 8 represent intermediate hierarchical problem involves describing the values of preferences. This methodology is based elements in a lower level in terms of some or all of on extensive research on psychological behavior, the elements in the next higher level. Theprocess which demonstrated that an individual cannot begins by (1) formulating a problem starting from simultaneously compare more than five to nine a primary objective (e.g. maximizing the overall objects without being confused ( Kangas 1994). utility); (2) identifying the interested parties, stake- Using the pairwise comparisons, the relative holders, or decision makers involved; (3) defining weights (importance /preference) of elements at objectives that represent the wishes and aspira- each hierarchical level are determined using the tions of all interested parties (e.g. performance eigenvector method (Saaty 2001). To compute the objectives, improving economic benefits); and (4) eigenvector, the analyst first needs to create the articulating feasible alternatives that must be matrix of pairwise comparisons A, as shown in applied to achieve the objectives. Criteria and equation [1] subcriteria are used to specify the objectives in step (3) when they are complex and are measured in 1 w1 / w2.wl / wn more detailed and ordinal forms. A=aj =w2 / wl 1... w2/ wn [1] The theoretical foundation behind the AHP is the utility theory of value. Zahedi (1987) showed ion / w1 wn / w2 1 that the process of selecting alternatives is consistent with maximizing a decision maker's either where ail represents the pairwise comparisons of single or multiple attribute utility functions. element i and j; and when i = j, aij =1; wi /wi is the Results implied that the AHP and the utility relative importance of element i over element j; maximization criterion can be combined to solve and n is thenumber of elements compared. decision problems. Multiobjective forest manage- Because the AHP involves subjective assignment ment involves multi- attribute utility functions that of values, some inconsistency may be expected; can be evaluated using AHP by incorporating hence Saaty (1980) proposed an eigenvector weights and relative rankings of alternatives method to test for inconsistencies.If matrix A (Duke and Aull -Hyde 2002). contains inconsistencies, they can be estimated The most general objective of a multiobjectiveusing a Consistency Index (CI) described in the form of forest management planning problem is to maximize the overall utility U of the system. Maximi- CI =Amax n)/ (n-1) [2] zation of utility thus becomes the highest level in the hierarchy of the multiobjective problem formu- where Xmax is the largest eigenvector of the matrix lation. A utility model is a mathematical tool that A. The eigenvector value ( ?max ) is obtained by multiplying the matrix A by the vector of relative describes a problem in terms of features, suchas goals or objectives that express the wishes and importance weights, e.g., aspirations of individuals and /or groups. A very Xmax = yyaijwi simple utility model can be described in terms ofa i utility value U that is the sum of the individual ob- The vector of relative weights is obtained by nor- jective weights (at) multiplied by the decision vari- malizing each column, and then normalizing by ables Xi that represent a particular alternative i, row (Saaty 1988). Matrix A has to be estimated for U= EiaiXi. each decision variable at all levels of the system; in Analytic Hierarchy Process in Durango, Mexico 41

each, the eigenvector is scaled 'to add up to 1 toproducts development (NTP), and water harvest- obtain level priorities. We developed a basic ing /environmental services (WH). Table 1 shows program using a spreadsheet to calculate both the the funds allocated to these programs in 2003. relative weights and the eigenvectors. If the pair- The main goals of the forest development pro- wise comparisons for an n x n matrix include no gram are to promote the sustainability of primarily inconsistencies, then A = n ; otherwise A is timber- related products, strengthen landowners' simply the reciprocal of the matrix (Saaty 2001). resource management skills, and reduce opera- The more consistent the comparisons are, the tional costs. It is the most important program with closer the value of the computed Xmaxis to n. A the largest budget allocation and the largest num- consistency ratio (CR) measures the coherence of ber of projects funded. The plantation develop- the pairwise comparisons: ment program is aimed at increasing forest stocks by establishing new commercial plantations, CR = 100(CI / ACI) [3] producing seedlings, and reforesting burned areas where ACI is the average consistency index forand former agricultural or pasture areas. The non- randomly generated comparisons, which varies timber products development program is not as with the size of the matrix (Saaty 1988). Kangas large as the first two programs, but it is focused on (1994) recommended that a CR of 0.10 or less is the development of oregano, mushrooms, fisher- acceptable. ies, and ecotourism projects. Though it was designed to operate in only six states of Mexico, Problem Description the state of Durango was included because of its In the past, Mexican planners and resourceenormous potential to generate these products in managers rarely focused on developing strategies its semi -arid, temperate forest areas. The fourth to strengthen the multiple use and sustainable program, water harvesting /environmental ser- development practices of landowners. Hence, fed-vices development, is the most recent program eral programs resulted in poor resource manage- created by the federal government to protect and ment scenarios that jeopardized sustainable de-enhance amenity values, improve water quality velopment and eventually led to the degradationand quantity, increase carbon sequestration, of the environment. promote recreation, and reduce the pressure on The lack of a consistent policy to strengthen ejidos timber -based products. It was not funded in 2003, [common properties] has grave social implications but both federal and state authorities expect it to that result in the degradation of natural resources and consequently, this prevents rural communities develop quickly as the public becomes more from making s ustainable use of such resources and familiar with it. therefore increasing their quality of life. This creates The mandate to develop these programs a vicious circle of degradation and poverty. involves giving monetary resources directly to the (Semarnat 2000, p. 43) landowners to achieve sustainable forest develop- Recent changes have improved the manage-ment. The fundamentals of these national pro- ment of natural resources by developing new grams are fully described in the document "2001- forest policies that promote the sustainable 2006 National Forest Program," which includes management of environmental services and thatseven national strategies and 14 objectives to encourage policy enforcement, research, and a promote sustainable forest development in Mexico more efficient decision -making process. These changes were necessary to alleviate the disparities between farmers and forest landowners. While farmers received a government subsidy to culti- Table 1. Budget allocation of federal programs in 2003. vate their lands, forest landowners never received Budget funds to properly manage their natural resources, Program (mill$) % but instead were blamed for the deterioration of land and the poor quality of water. Therefore, the Forest development 2.16 53 current government started four national pro- Development of plantations 1.45 35 grams to help forest landowners become more Non -timber products development 0.49 12 competitive and to diminish their differences from Water harvesting & environmental farmers. These programs are forest development services 0 0 (FD), development of plantations (PD), non -timber Total 4.10 100 42 Perez -Verdin and Tecle

(http:/ /www.conafor.gob.mx/documentos_conaf preferences for each group. We calculated the or /ENG /pdf /19.pdf, and http: / /www.conafor. eigenvectors and consistency indexes and checked gob. mx /documentos_conafor /ENG /pdf /01.pdf, for potential inconsistencies. A similar process was visited June 7, 2004). To satisfy the goals and also used for the management objectives. In this objectives identified above, we propose using the case, we made the pairwise comparisons for each four national programs (FD, PD, NTP, and WH) as management objective and for each national the feasible projects competing for the limited program. The consistency indexes revealed no funds. We also reduced the nine categories of proj- inconsistencies in the pairwise comparisons (see ects into six management objectives: (1) improving Appendix A). economic benefits, (2) increasing water yield, (3) In formulating the research problem, we first increasing recreation or other environmental identified the most general objective in the hier- benefits, (4) increasing forest stocks, (5) reducing archical process, which is to maximize the utility of operational costs, and (6) increasing the yield of forest budget allocation in the state of Durango non -timber products. We have not included some (first level). We estimated the relative importance biological objectives such as protecting biodiver- of each interested party and structured their re- sity or wildlife habitat because they are not within spective weights as level 2. Kangas (1994) indi- CONAFOR's areas of responsibilities. However, cated that the pairwise comparisons of interested we included all of the interested parties in Du- groups may be made by the office administering rango involved in the multiobjective forestthe area, whose members may have a good management decision -making process. Theseunderstanding of the importance of each group. parties are grouped into public or government Hence, pairwise comparisons for each party were sector (PUB), landowners (LND), private sector based on personal knowledge of the area and the (PRV), and non -governmental organizations role of the interested parties in the management of (NGO), which includes environmentalist groups natural resources. We also considered the scope of and non -profit forest organizations. Landowners, forest law, which gives more importance to forest the most dominant group in Durango, includeslandowners than to any other group in the man- common and private properties. Common prop- agement of forest resources. erties, or ejidos, are collectively owned expanses of Level 3 consisted of the management objectives land (Alcorn and Toledo 1998), which occupy up and their relative weights, which came from the to 70 percent of forestlands. Thus, the essence of responses to the survey questionnaires. Finally, this study is to determine the most efficient budgetlevel 4 consisted of the alternatives or national allocation to the four national programs, taking development programs. Note that not every objec- into account the views of the four interested tive has links to all four of the national develop- groups and the six management objectives. ment programs (Figure 1). Formulation of the problem in an analytic hierarchy process frame- Data Acquisition and Problem Formulation work links each level to the immediate levels Twenty representative persons were identified above and below it, thereby tying the entire and asked to answer an online questionnaire to scheme together mathematically (Saaty 1988). reveal their individual preferences for budget allocation in the management of natural resources RESULTS in Durango. The individuals represented each of Equation [1] describes quantitatively the rela- the four interested groups; however, seven refused tive importance or weights of each interested to take part in the study. Thus, we used only 13 party, management objective, and national pro- individual responses to construct the matrix for gram separately (see Appendix A). The elements the pairwise comparisons. The 13 people consisted of the matrix were determined by normalizing of four individuals from the public sector, three each column and then averaging across each row landowners, three non -government organizations, (Saaty 1988). The relative weights for the inter- and three from the private sector. The question- ested parties are 0.18 for the public sector (PUB), naire started with the identification of the respon- 0.51 for landowners (LND), 0.17 for the private dents, followed by a question concerning theirsector (PRV), and 0.14 for non- government knowledge of the four national programs, and organizations (NGO). The relative importance of questions that asked them to compare and rank the landowners is slightly higher than the other stake- management objectives. We then averaged the holders (see Appendix AI). In this case, Xm _ pairwise comparisons and constructed a matrix of 4.25, ACI = 0.90, CI is 0.08, and CR is 0.09. The CR Analytic Hierarchy Process in Durango, Mexico 43

Allocating budget for multiobjective forest management

Public sector Landowners Private sector Non -governmental organizations

1 1 Improve Reduce Increase m° Increase forest Increase economic operatwnal water recreation or other stocks mn timber benefits yield environmental costs products yield Illtter. , Aliallr44111,44~11:* ----°°...VP;fr AP- Aikk. Forest Development of Non -timber Water development plantations products harvesting and program pmt development environmental program services program

Figure 1. Hierarchies of budget allocation for the multiobjective forestmanagement problem in Durango, Mexico.

is within the standards recommended by Kangas of management alternatives or national programs). (1994), i.e., CR 5 0.1. Thus, the global priority (utility) ofa management The next step involves prioritization of the alternative can be calculated as decision objectives for each interested party (Table 2). The decision objective that received the highest 4 6 overall weight from among the four interest GP = : LPIGf y LPOik x LPMSik}] groups was increasing forest stocks (0.97), whereas i =1 k =1 reducing operational costs received the lowest value where GP, is the global priority (overall utility) of (0.44; see Appendix All for individual objective management alternative i, LPIGf is the local prior- pairwise values). ity of interest group j, LPOfk is the local priority of To optimally allocate the available budget, the objective k from the point of view of interestgroup four national programs (i.e., the desiredmanage- j, and LPMSik is the local priority of management ment types) are evaluated with respect to theiralternative i (national program) with respect to performance in achieving the desired management objective k. Special attention should be paid to de- objectives. This evaluation assigns a preference termining GP, because all management objectives structure or weight to each objective; an expert do not have links to all management alternatives; consultant is expected to indicate the respective therefore no LPMSik values are reported forsome weights. In this study, we assumed the role of the of the cells in Table 3. In addition, thesum of all expert consultant and calculated the weights based GPimustbe equal to one. on Kangas's methodology (Kangas 1994). As We determined the global priorities for the shown in Figure 1, the AHP includes four levels of management alternatives using equation [4]. For hierarchy (an overall objective, four interestedexample, the calculation of the global priority for parties, six management objectives, and four types the forest development program (FD) is shown 44 Perez - Verdin and Tecle

below: financial resources to develop water harvesting GPFD = [0.18 x and environmental service programs while reducing the proportional budget allocation to the forest {(0.10 x 0.44) + (025 x 020) + (0.23 x 0.17) + development program. Figure 2 compares the (0.24 x 0.16) + (0.06 x 0.67) + (0.12 x 0.30)} + 0.51 x desired budget allocation with the actual budget distribution for fiscal year 2003. The desired {(0.26 x 0.44) + (0.12 x 0.20) + (0.07 x 0.17) + budget was obtained by multiplying the global (0.28 x 0.16) + (0.18 x 0.67) + (0.09 x 0.30)} + priorities and the total budget assigned in 2003 0.17 x (i.e., US $4.1 million). {(0.18 x 0.44) + (0.12 x 0.20) + (0.19 x 0.17) + We performed a sensitivity analysis to evaluate (0.24 x 0.16) + (0.13 x 0.67) + (0.14 x 0.30)} + the effect of changes in local proprieties on the 0.14 x decision outcomes. The analyst can manipulate a {(0.09 x 0.44) + (0.24 x 020) + (0.22 x 0.17) + decision maker's preference structure to test its (0.21 x 0.16) + (0.07 x 0.67) + (0.17 x 0.30)}] robustness in selecting the preferred management = 0.307. alternative. To demonstrate the sensitivity of the model in this study, we evaluate any changes to The global priorities for the rest of the national the global priority values by varying the landown- programs are PD 0.214, NTP 0.177, and WH 0.303. ers' preference structures while keeping constant The management alternative with the highest glo- the other interested parties' weights. When the bal priority (i.e., utility) is forest development, which utility values of the landowners are less than 0.50, is followed closely by water harvesting. Since these water harvesting obtains the highest priority values. results are based on the wishes and aspirations of However, forest development would become the representative groups of the society, the govern- most preferred alternative if the landowners' ment may allocate its budget for forest resources utility values were greater than 0.5 (Figure 3). The management among the four national programs in basis for accepting the desired budget allocation is accordance with the resulting global utility values related to the land tenure system existing in Du- to best achieve the overall management objectives. rango. As expressed earlier, the new forest policies In the past, budget allocations followed nosys- are structured to help the landowners, so it is tematic approach. In 2003, for example, 53 percent unlikely that the preference of any other group of the forest budget went to forest development, 35 could prevail. Hence, we can conclude that the percent to plantations development, 12 percent to global priority values of the national programs are non -timber products, and nothing to water har-not sensitive to changes in group preferences, as vesting. However, if current policies were de-the two top management alternatives differ little signed to satisfy the overall social benefits, the from each other with such changes. government would need to redistribute the budget To corroborate the differences from the sensi- among all national programs to meet the needs of tivity analysis, we performed a t -test, along with a all interested parties. According to these results, test for homogeneity of variances, by dividing the government planners would need substantial landowners' utility values into two classes: those

Table 2. Objective weights according to each interest group involved in the multiobjectiveforest management process.

Overall Objectives PUB LND PRV NGO Weight

Improving economic benefits 0.10 0.26 0.18 0.09 0.63 Increasing water yield 0.25 0.12 0.12 0.24 0.73 Increasing recreation or other environmental benefits 0.23 0.07 0.19 0.22 0.71 Increasing forest stocks 0.24 0.28 0.24 0.21 0.97 Reducing operational costs 0.06 0.18 0.13 0.07 0.44 Increasing non -timber products yield 0.12 0.09 0.14 0.17 0.52 X 6.50 6.37 6.60 6.16 CR 0.08 0.06 0.10 0.03 Analytic Hierarchy Process in Durango, Mexico 45

Table 3. Local priorities of management alternatives (national programs) with respect to decision objectives (levels 3 to 4).

Management Alternatives (National Programs)*

Objectives FD PD NTP WH CR

Improving economic benefits 0.44 0.17 0.39 3.03 0.02 Increasing water yield 0.20 0.20 0.60 3.00 0.00 Increasing recreation or other 0.17 0.39 0.44 3.03 0.02 environmental benefits Increasing forest stocks 0.16 0.42 0.14 0.27 4.07 0.03 Reducing operational costs 0.67 0.33 2.00 0.00 Increasing non -timber products yield 0.30 0.16 0.54 3.01 0.01 * Blank cells represent the management objectives that do not have costs associated with a particular nationalprogram.

having utility values less than 0.50 are in class 1 tion is based on multiple use of resources, dis- and those with utility values greater than or equal regarding past misinterpretations of the single use to 0.50 in class 2. We identified the four manage-of forest resources. For example, water harvesting ment alternatives (national programs) as depen- and environmental services should be rated similar dent variables and their utility values classes as the to forest development because both would pro- factors or independent variables. The t -test showed duce benefits not only to the landowners but to no significant differences among the two classes; society as a whole. We recommend use of the AHP this means that changing the utility values ofin a more comprehensive manner, including public landowners has no effect on the global priorities ofopinions as much as possible, as well as other each management alternative. objectives such as wildlife habitat and biodiversity. The AHP is suitable for handling qualitative and CONCLUSIONS AND RECOMMENDATIONS quantitative data coming from public opinions or This study used the AHP to optimally distribute any other measured social or biological entity. a forest management budget among different However, we also recommend caution in using the programs in the state of Durango, Mexico. We AHP because it requires normalization of all data analyzed the perceptions of four representativeto a common scale; it needs to have diverse and groups, four national programs, and six manage- representative value functions in the decision- ment objectives to allocate the budget for multi - making process. Studies like the one by Duke and objective forest management. Two programs- Aull -Hyde (2002) can be implemented to gathera forest development and water harvesting /environ-large number of public opinions by using survey mental services -are preferred and should thereforeinstruments and can facilitate a more reliable and receive the highest budget allocation. Since water fair decision -making process. harvesting /environment servicesisa new In addition, the AHP by no means should be program, authorities should redirect future budget used as the only tool suitable for decision -making distributions by reducing the budget designated to analysis (Tecle 1992), nor should it be used to the other programs, especially to forest develop- replace public participation; on the contrary, both ment. This should lead to a fair and optimal approaches should be considered in a complemen- allocation of the entire budget among all the tary sense. The final decision should incorporate national programs. an integrative assessment and sensitivity analysis. Three major reasons favor this distribution ofSensitivity analysis reveals model reliability and the desired budget allocation. First, the money gives valuable information concerning the effects comes from taxes and a proper distribution ofof varying utility values and weights on the choice funds should involve public participation. Second, of management alternatives (Kangas 1994). Like- the desired distribution of the budget tends to be wise, integrative assessment helps decision makers optimal because resource managers know where it consider the interactions and feedback of the vari- is most needed and where it can benefit the most. ous facets of a decision problem as well as identi- Third, the method used to determine this distribu-fying and exploring different sources of risk and 46 Perez - Verdin and Tecle

2.50

2.00

1.50

1.00

0.50

FD PD NTP WH Alternatives (National Pro ennisl

Figure 2. Budget allocation to national programs in the state of Durango, Mexico. In2003, no funds were allocated to the water harvesting program.

0.4 FD- - pD- - - NTP WH 0.4 0.3 0.3 0.2 0.2 0.1 - 0.1 - 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Landowners' utility values

Figure 3. Changes in global priority values with varying landowners' utility values.In this case, the utility values of the other interested parties are assumed to be equal.

uncertainty. The use of AHP can clarify public ACKNOWLEDGMENTS preferences in a schematic manner and delineate The authors wish to express their gratitude to the initial links among management objectives, Dr. Ciro Hernandez -Diaz of Silviculture and Wood alternatives, and public preferences. In thisResearch Institute and Ramon Silva- Flores, a manner, the AHP helps to understand and clarify a resource planner with CONAFOR- Durango, for complex problem with many interacting compo- their invaluable help in getting the data used in the nents and to formulate it in a multiobjective frame- pairwise comparison matrices and making the work. The multiobjective forest managementcase CONAFOR budget available. This project was study in Durango is a good example of a successful partially funded by the state of Arizona's Prop. 301 application of the AHP to solve a budget allocation funding and CONACYT. problem. Analytic Hierarchy Process in Durango, Mexico 47

LITERATURE CITED Alcorn, J. B., and V. M. Toledo. 1998. Resilient resource management in Mexico's forest ecosystems: The con- Semamat. 2000. Strategic Forest Program 2025. Secretari- tribution of property rights. In Linking Social and a de Medio Ambiente y Recursos Naturales. Available Ecological Systems: Management Practices and Social http:/ /www.conafor .gob.mx /documentos_conafor/ Mechanisms for Building Resilience, edited by F. ENG /pdf /19.pdf (accessed June 7, 2004). Berkes and C. Folke, pp. 216 -249. Cambridge Univer- Saaty, T. 1980. The Analytic Hierarchy Process. Planning sity Press, Cambridge UK. Priority Setting, Resource Allocation. McGraw -Hill, Duke, J. M., and R. Aull -Hyde. 2002. Identifying public New York. preferences for land preservation using tae analytic hierarchy process. Ecological Economics 42: 131 -145. Saaty, T. 1988. Multicriteria Decision Making: The Ana- Kangas, J. 1992. Multiple -use planning of forestre- lytic Hierarchy Process. RWS Publications, Pittsburgh, sources by using the analytic hierarchy process. Scan- A. dinavian Journal of Forest Research 7: 259 -268. Saaty, T. 2001. Fundamentals of the analytic hierarchy Kangas, J. 1994. An approach to public participation in process. In The Analytic Hierarchy Process in Natural strategic forest management planning. Forest Ecology Resource and Environmental Decision Making, edited and Management 70: 75-88. by D. L Schmoldt, J. Kangas, G. A. Mendoza, and M. Kangas, j. 1999. The Analytic Hierarchy Process (AHP): Pesonen, pp. 15-35. Kluwer Academic Publishers, Standard version, forestry applications and advances. Norwell, MA. In Multiple Use of Forests and Other Natural Re- Tecle, A. 1992. Selecting a multicriterion decision mak- sources, edited by F. Helles, P. Holten- Andersen, and ing technique for watershed resources management. L. Wichmann, pp. 96 -105. Kluwer Academic Publish- Water Resources Bulletin 28(1): 129-140. ers, Norwell, MA. Zahedi, F. 1987. A utility approach to the analytic hier- Schmoldt, D. L., J. Kangas, and G. A. Mendoza. 2001. Ba- archy process. Mathematical Modeling 9: 387 -395. sic principles of decision making in natural resources and the environment. In The Analytic Hierarchy Pro- cess in Natural Resource and Environmental Decision Making, edited by D. L Schmoldt, J. Kangas, G. A. Mendoza, and M. Pesonen, pp. 1-13. Kluwer Academ- ic Publishers, Norwell, MA. 48 Perez -Verdin and Tecle

Appendix A

AI. Pairwise comparisons (A) ofstakeholders' utilities and vector of relativeimportance (w)

A w PULDPRNG 0.18 PU 1 1/4 2 1 0.51 X ; =4.25; CR=3.09 Aw = a=LD 4 1 3 3 0.17 PR1/21/3 1 2 0.14 NG 1 1/31/2 1

PU = Public Sector, LD = Landowners; PR= Private Sector, and NG = Non -government organizations

All. Pairwise comparisons ofstakeholders' utilities of managementobjectives and vector of weights

Public sector preferences EE WY REC FS OPC NTP EE 1.00 0.33 033 0.33 3.00 1.00 '0.10 WY 3.00 1.00 2.00 0.50 3.00 3.00 x 0.25 ; max =6.50; CR=0.08 REC 3.00 0.50 1.00 2.00 4.00 1.00 0.23 FS 3.00 2.00 0.50 1.00 3.00 2.00 0.24 OPC 0.33 0.33 0.25 0.33 1.00 0.50 0.06 NTP LOO 0.33 1.00 0.50 2.00 1.00 0.12

Landowners preferences EE WY REC FS OPC NTP EE 1.00 3.00 3.00 1.00 2.00 2.00 0.2C WY 0.33 1.00 2.00 0.50 0.33 2.00 0.12 X ; max =6.37; CR=106 REC 0.33 0.50 1.00 0.25 0.33 1.00 0.07 FS 1.00 2.00 4.00 1.00 3.00 2.00 0.28 OPC 0.50 3.00 3.00 0.33 1.00 2.00 0.18 NIP 0.50 0.50 1.00 0.50 0.50 1.00 0.09_

Private Sector preferences EE WY REC FS OPC NTP EE /.00 3.00 1.00 LOO 0.50 1.00 0.18 WY 0.33 1.00 1.00 0.50 0.12 1.00 1.00 X ;2 =6.60; CR=3.10 REC 1.00 1.00 1.00 0.50 3.00 2.00 0.19 FS 1.00 2.00 2.00 1.00 3.00 1.00 0.24 OPC 2.00 1.00 033 0.33 1.00 1.00 0.13 NTP 1.00 1.00 0.50 1.00 1.00 1.00 0.14 Analytic Hierarchy Process in Durango,Mexico 49

Non -governmental organizations preferences EE WY REC FS OPC NTP 0.09 EE 1.00 0.33 0.33 0.33 0.24 1.00 1.00 X ;Amax=6.16; CR=0.03 WY 3.00 1.00 1.00 1.00 3.00 2.00 0.22 REC 3.00 1.00 1.00 1.00 4.00 1.00 0.21 FS 3.00 1.00 1.00 1.00 3.00 1.00 0.07 OPC 1.00 0.33 0.25 0.33 1.00 0.33 0.17 NTP LOO 0.50 1.00 1.00 3.00 1.00

EE = Improve economic benefits; WY = Improve water yield; REC = improve recreation/environmentalservices; FS = Increase forest stocks; OPC = Reduce operationalcosts; and NTP = increase non- timber products yield.

AIII. Pairwise comparisons ofmanagement objectives' utilities and vector of weights(w)

Improve Economic Benefits Improve Water Yield Improve Rotration and Environmental Services FD NTP w WH FD PD WH w FD 1.00 3.00 0.44 FD PD WH w 1.00 FD 1.00 1.00 0.33 0.20 NTP 0.33 1.00 0.17 FD 1.00 0.50 0.33 0.17 0S0 PD 1.00 1.00 0.33 0.20 WH 1.00 2.00 1.00 0.39 PD 2.00 1.00 1.00 0.39 WH 3.00 3.00 1.00 0.60 A=3.03; CR=0.02 WH 3.00 1.00 1.00 0.44 .Ì,max=3.00; CR=-0.00 A =3.03; CR-41.02

Increase forest stocks Reduce operational costs Increase non- timber products

FD PD NT? WH w FD FD NTP w 1.00 0.50 1.00 0.50 0.16 FD PD NTP w FD PD 2.00 1.00 2.00 0.67 1.00 3.00 2.00 0.42 FD 1.00 2.00 0.50 0.30 NTP 1.00 NI? 0.50 1.00 0.33 033 1.00 0.50 0.14 PD 0.50 1.00 0.33 0.16 WH 2.00 0.50 2.00 1.00 0.27 NTP 2.00 3.00 1.00 0.54 Amax=2.00; CR=0.00 A max=4.07; CR=0.02 A=3.01; CR-4.01

FD = Forest Development; PD = Plantations Development;NTP = Non -timber products; WH = Water Harvesting SOIL LOSS FOLLOWING THE RODEO -CHEDISKI WILDFIRE: AN INITIAL ASSESSMENT

Pablo A. Garcia- Chevesich,1 Peter F. Ffolliott,1 and Daniel G. Neary2

The Rodeo -Chediski wildfire, the largest fire in had been obtained on volcanic soils, with only Arizona's history, damaged, destroyed, or dis- limited work on sedimentary soils. rupted the hydrologic and ecological functionings The watersheds exhibit the relatively flat topog- of all the ecosystems impacted. The extent to raphy (most of the slopes are less than 10 %) that is which these functionings will return to pre -firecommon on the Colorado Plateau. Elevations conditions, if they ever will, is open to conjuncture. range from 6800 to 7000 ft (2073 -2134 m). Creta- It is therefore very important that the impacts ofceous undivided material that is similar to the this wildfire be documented to establish a refer-Coconino sandstone formation lies beneath the ence for the time -trend responses of impacts from watersheds. McVickers soils in the Soldier -Hogg- future wildfires of comparable magnitudes. To McVickers Association characterize the two water- that end an initial assessment was made of soil loss sheds. Hendricks (1985) described these moder- on two watersheds in the ponderosa pine forests at ately well to well drained soils as having a fine, the headwaters of the Little Colorado River after sandy loam texture. Sixty -five percent of the the Rodeo -Chediski wildfire. One of the water- annual precipitation of 20-25 inches (508 -635 mm) sheds experienced a high -severity burn and the falls from October to April, much of it as snow, other a low- to medium -severity burn. Soil loss is and most of the remainder falls in rainstorms from continuing to be monitored to obtain a longer,July to early September. More than 90 percent of more comprehensive picture of the impacts of this the intermittent pre -fire streamflow from the catastrophic wildfire event. watersheds occurred as a result of snowmelt- runoff or winter rains prior to the Rodeo -Chediski WATERSHEDS STUDIED wildfire. Before the wildfire, summer storms, Two nearly homogeneous watersheds, 60 acres while often intense, rarely produced significant (24 hectares) each, were established in 1972 by the stormflows. University of Arizona and the USDA Forest The watersheds had been "moth- balled" in 1977 Service, Rocky Mountain Research Station, along after completion of the baseline studies. However, Stermer Ridge, about 9 miles (15 km) south ofthe 3 ft (0.9 m) H -flumes were left in place in Overgaard, Arizona, at the headwaters of the Little anticipation of future monitoring needs. After the Colorado River. The watersheds were located in Rodeo -Chediski wildfire, these control sections the previously cut -over, mostly uneven -aged were refurbished and reinstrumented with water - ponderosa pine forests of the Colorado Plateau level recorders, and a weather station on the site physiographic province. The most recent harvest was reestablished. The set of 30 permanently of the pre -fire timber removed 45 percent of the located sample points (plots) that had originally sawtimber by group selection in the early 1960s. been placed on each watershed to sample on -site The objective of establishing these watersheds was hydrologic and ecological parameters were re- to obtain baseline information on the hydrologic located to provide a basis to study the impacts of and ecological functioning of ponderosa pine varying fire severities on hydrologic and ecological forests situated on sedimentary soils (Ffolliott and processes (Ffolliott and Neary 2003). Baker 1977). Most of the information about the A fire severity classification system that related ponderosa pine forests in the region before 1972 fire severity to the soil- resource response (Hunger- ford 1996; DeBano et al. 1998) on the sample plots 1School of Natural Resources, University of Arizona, Tucson was extended to the area of each watershed to 2Rocky Mtn. Research Station, USDA Forest Service, Flagstaff determine the relative portions of the respective 52 Garcia -Chevesich, Ffolliott, and Neary

watersheds that were burned at low, moderate, RESULTS AND DISCUSSION and high severity (Wells et al. 1979). This extrapo- Watershed Differences lation of the classification indicated that one of the Soil loss estimated by the measurements of soil Stermer Ridge watersheds had experienced thepedestals after the first summer monsoonal rains impacts of a high- severity, stand -replacing fire, following the Rodeo -Chediski wildlife is shown in and the other watershed had been exposed to aFigure 1. More soil was lost on the two Stermer low- to medium- severity, stand -modifying fire. Ridge watersheds in this period than that reported by Rich (1962) for a comparable time period (that METHODS is, following the first post -fire summer monsoonal Estimates of soil loss were initially obtained rains) after a "hot wildfire" burned 60 acres (24 from measurements of 23 soil pedestals on thehectares) of ponderosa pine- Douglas -fir forest severely burned watershed and 18 soil pedestals vegetation on the upper part of the 381 ac (154 ha) on the lightly to moderately burned watershed at South Fork of Workman Creek in 1957. In compar- the end of the summer monsoonal rains 2 months ison to Stermer Ridge, the soils on the Workman after the wildfire. These soil pedestals wereran- Creek watersheds are underlain by Dripping domly located on the watersheds but they were Springs quartzite rock that has been intruded by not distributed in a manner that provided an un- diabase and basalt plugs and sills. Troy sandstone biased sample of the watershed conditions; most of outcrops on the upper portion of these watersheds, the pedestals were found on the upper and middle including the area burned. slopes of the watersheds. Three erosion pinswere Figure 1 shows the average soil loss following then installed around each of the sample pointson the first winter precipitation and snowmelt- runoff the two watersheds to obtain unbiased estimates of period and second summer monsoonal rains after post -fire soil loss. Two pins were placed 6 ft (1.8 the Rodeo -Chediski wildfire based on measure- m) upslope and one pin the same distance down- ments of the erosion pins. There was greater soil slope of the point. Measurements of soil losswere loss during the former than the latter, a difference made 10 and 16 months after the wildfire. The attributed largely to a "breaking down" of the erosion pins were reset after measurement to esti- extensive post -fire hydrophobic conditions of the mate soil loss in the intervening period. Soil loss soils on the severely burned watershed. Soil loss for the 10 -month period represented thatasso- on the lightly to moderately burned watersheds ciated with the first post -fire winter precipitation was similar in these two post -fire periods, presum- and snowmelt- runoff events, whereas soil loss for ably because of the comparatively small change in the 16 -month period, that is the intervening 6- the less extensive hydrophobicity of the soils on month period, represented loss from thesummer this watershed. monsoonal rains occurring in the second year after The measurements of soil pedestals suggest that the wildfire. A bulk density value obtained from 25-30 tons per acre (56-67 metric tons /hectare) of soil samples collected from unburned siteson soil were eroded on the severely burned watershed Stermer Ridge was used as a basis to convert meas- in response to the first summer monsoonal rains, urements of average soil loss to corresponding and 15 -20 tons per acre (34-45 metric tons /hec- erosion rates in terms of tons per acre (metric tare) of soil were lost on the lightly to moderately tons /hectare). burned watershed (Figure 2). Soil erosion rates on Analyses of variance were used to compare the the severely burned and lightly to moderately averages of soil loss relative to watershed condi-burned watersheds after the first winter precipitation and hillslope position, slope percent, andtion and snowmelt- runoff period following the aspect within a watershed. Tukey -Kramer multiple wildfire were 22 and 12 tons per acre (49 to 27 comparison tests were made to evaluate the differ- metric tons /hectare), respectively. Erosion rates ences among respective means. When heterocedas- after the second summer monsoonal rains were ticity was detected by the Bartlett's test for ho- statistically similar, averaging 11 tons per acre (25 mogeneity of variances, nonparametric Tukey -type metric tons /hectare) for both watersheds. multiple comparison tests were used toassess Sediment transport was used as a proxy for soil differences among means (Zar 1999). All statistical loss by Campbell et al. (1977) in their evaluation of inferences were made at the 0.10 alpha level. the effects of a wildfire (the Rattle Burn) on a pon- Soil Loss After Rodeo- Chediski Wildfire 53

® Severely burned watershed o Lightly -to- moderately burned watershed

0.35 9 -1 0.30 8 .-. E 0.25 7 E o 6 0.20 5-° 0.15 a 4 c°n 0.10 3 2a)

Figure 1. Averages and 90% confidence intervals forsoil loss on the Stermer Ridge watersheds for three post -fire precipitation periods after the Rodeo -Chediski wildfire. (Confidenceintervals were not calculated for soil loss after the first post -fire summer monsoonal rains because the soilpedestals measured were not distributed in amanner that provided an unbiased sample of the watershed conditions.)

® Severely burned watershed o Lightly -to- moderately burned watershed

35 80 30 70 CD 25 60 cn a) ó 50 ócn ° 20 40 ° a o 15 (7) ó`" a 10 > a 5

0 First monsoon Winter precipitation and Second monsoon snowmelt runoff Post -fire precipitation periods

Figure 2. Averages and 90% confidence intervals for erosion rates on the Stermer Ridge watersheds for three postfire precipitation periods after the Rodeo -Chediski wildfire.(Confidence intervals were not calculated for the erosion rates after the first post -fire summer monsoonal rains because the soil pedestals measured were not distributed ina manner that provided an unbiased sample of the watershed conditions.) 54 Garcia- Chevesich, Ffolliott, and Neary

derosa pine ecosystem in north -central Arizona in A few differences were observed in post -fire 1972. These investigators reported that about 1.7 soil loss relative to slope percent, but they were tons per acre (3.8 metric tons /hectare) of sediment inconsistent and the causes for their occurrence are were carried off a "severely burned" watershed of unknown. Furthermore, analyzing the effects of 20 ac (8 ha) in the first year after the wildfire. Only slope percent on soil loss after the Rodeo -Chediski a few pounds of sediment were measured on a 10 wildfire was limited by the flat topography of the ac (4 ha) "moderately burned" and a 45 ac (18 ha) Stermer Ridge watersheds. unburned watershed in the same period. The There were no significant differences in soil loss sedimentary soils characterizing these watershedsby aspect on either watershed following the first are derived from the Kaibab limestone formation. winter precipitation and snowmelt- runoff period Nearly 40 percent of the soils are classified as after the wildfire, and the observed differences in either the Soldier or McVickers series, and about soil loss after the second season of summer 50 percent are designated as unnamed, extremely monsoonal rains were few and inconsistent, and stony limestone outcrops. Parenthetically, Gott-reasons for the differences are unknown. Analysis fried and Neary (2003) are monitoring sediment of soil loss over the full range of aspects was not transport on the Workman Creek watersheds after possible because of the limited (mostly northwest the Coon Creek wildfire of 2000. Information to be to northeast) orientation of the watersheds. gained from these measurements will further expand the knowledge of wildfire effects on soil SUMMARY loss and sedimentation on watersheds in Arizona's Watersheds denuded by devastating wildfire montane forests. are vulnerable to a high rate of soil loss and, as a consequence, can yield large amounts of post -fire Hillslope Position, Slope Percent, Aspect sediment. While the magnitude of soil loss on the Soilloss on upper, middle, and lower hillslope Stermer Ridge watersheds following the Rodeo - positions of the Stermer Ridge watersheds is Chediski wildfire is site -specific, it reflects the shown in Figure 3 for the first winter precipitation response that might be expected from catastrophic and snowmelt- runoff period and the second sum- wildfire of varying severities on watersheds with mer monsoonal rains after the wildfire. The only similar conditions to those on the Stermer Ridge difference occurred on the lower slopes after thewatersheds. Soil loss will generally be highest in winter precipitation and snowmelt- runoff period, the first year after a wildfire, especially if the where soil loss on the severely burned watershed burned watersheds experience high- intensity was greater than that on the lightly to moderately rainfall events immediately after the fire has ex- burned watershed. A possible explanation for this posed the soil surface; this was the case following difference is the combined effect of the greater ex- the Rodeo -Chediski wildfire. Researchers are tent of post -fire hydrophobic soils on the severely continuing to monitor soil loss on the Stermer burned watershed and the higher erosive power of Ridge watersheds to determine the rate at which the larger -volume post -fire overland flows on the soil loss declines in subsequent years as protective lower than on the upper and middle slopes of the vegetation becomes established. watershed (DeBano et al. 1998). Soil loss after the second summer monsoonal rains was not related to hilislope position on either of the watersheds. Soil Loss After Rodeo -Chediski Wildfire 55

® High severely burned watershed o Lightly -to- moderately burned watershed

0.35 9

0.30 8

--.. 0.25

co óu) 0.20

t0.15

0.10

0.05 1

0.00 0 Upper Middle Lower Hillslope position

0.35 - 8 0.30 - B - 7 0.25 - E - 6 E u) U) ó U) 0.20 - - 5ó .o u)

á, 0.15 - -4m c co a) 5 > á> - 3> < 0.10 - Q 2 0.05 - - 1

0.00 i 0 Upper Middle Lower Hillslope position

Figure 3. Averages and 90% confidence intervals for soil loss on the Stermer Ridge watersheds fordifferent hillslope positions after (A) the first winter precipitation and snowmelt- runoff period and (B) the secondsummer monsoonal rains after the Rodeo -Chediski wildfire. 56 Garcia- Chevesich, Ffolliott, and Neary

REFERENCES Campbell. R. E., M. B. Baker, jr., P. F. Ffolliott, F. R. Hendricks, D. M. 1985. Arizona Soils.University of Ari- Larson, and C. C. Avery. 1977. Wildfireeffects on a zona Press, Tucson. 244 pp. ponderosa pine ecosystem: An Arizonacase study. USDA Forest Service, Research Paper RM-191. 12 pp. Hungerford, R. D. 1996. Soils: Fire inEcosystem Man - DeBano, L. F., D. G. Neary, and P.F. Ffolliott. 1998. a6ement Notes: Unit II -I. USDA ForestService, Fire's Effects on Ecosystems. John Wiley& Sons, New National Advanced Resources TechnologyCenter, York. 333 pp. Marana, Arizona. Ffolliott, P. F., and M. B. Baker, Jr.1977. Characteristics Rich, L. R. 1969. Erosion and sedimentmovement fol- of Arizona ponderosa pine standson sandstone soils. lowing a wildfire in a ponderosa pine forestof central USDA Forest Service, General TechnicalReport RM- Arizona. USDA Forest Service, RockyMountain For- 44. 6 pp. est and Range Experiment Station, ResearchNote 76. Ffolliott, P. F., and D. G. Neary. 2003.Initial assessment 12 pp. of the Rodeo -Chediski fireimpacts on hydrologic Wells, C. G., R. E. Campbell, L. F.DeBano, C. E. Lewis, processes. Hydrology and Water Resources in Arizona R. L. Fredricksen, E. C. Franklin, R. C.Froelich, and P. and the Southwest 33: 93 -102. H. Dunn. 1979. Effects of fireon soil: State-of- knowl- Gottfried, G. J., and D. G. Neary. 2003.Preliminary as- edge review. USDA Forest Service, GeneralTechnical sessment of sediment measurements at theweir Report WO-7. 34 pp. basins at Workman Creek, centralArizona. Hydrol- Zar, J. H. 1999. Biostatistical Analysis. ogy and Water Resources in Arizona and the South- Prentice Hall, west 33: 103 -116. Englewood Cliffs, New Jersey. 663pp. WATER CONSUMPTION OFCOMMON PLANTS IN THESOUTHWEST U.S.

Aregai Teclel

Farmers and plant biologists makeextensive STUDY AREA use of evapotranspiration estimates to determine Because evapotranspiration is highly the amount of water needed for plantgrowth and influenced by both the climate and topographiccharacteristics survival. Hydrologists and waterresource mana- of an area, it is appropriate gers use such knowledge to estimate the amount of to provide a description of the different aspects of climate water lost from the soil, bodies of water, andvege- and other tated areas through plant uptake. In factors that affect ET in the Southwest.The most the latter case, important of these climatic characteristics evapotranspiration (ET) is a measure of how much are precipitation, temperature, relative humidity, water is used by the plants for transpirationand wind tissue building. It is expressed speed, cloud cover, and solar radiation(Beschta as volume per unit 1976; Tecle 1990). Annual precipitation area, or depth over a particular area. The total ET in the for a crop or plant is usually the total Southwest ranges from less than 15cm in the amount of deserts to more than 90 water lost from the plant from the period cm in the high mountains, of itsand temperatures vary from lessthan 10° C planting to its harvesting time. Because suchwaterduring some winter nights is considered permanently lost by the on the mountains to as plants, the high as 50° C onsummer days in the deserts. The process is also known as consumptiveuse. A good Southwest is also characterized by lowrelative understanding of this process is importantto determine the amount of water needed humidity, mostly clear skies, ample windspeed, to grow a and a good dose of solar radiation, particular plant or to developa large -scale agri- all of which cultural project in a certain directly contribute to the highevapotranspiration area. It also helps to rate in the area (Beschta 1976). ensure optimal water use -where it is most pro- The information gathered for this study ductive and needed-especially in theface of is from scarcity. Arizona and neighboringareas in the Southwest. This paper provides a compendium of The climate of the study site is arid withan aver- estimates age annual rainfall of about 20 cm. The regional of water consumed by variouscrops, grasses, fruit trees, shrubs, and some forest trees of the topography is highly variable, with plainsin the U.S. lower elevations and along the Southwest, as determined by various researchers. coasts and rugged mountains along the Mogollon Rim, the The most important factors contributingto the western amount of consumptive use are identified, fol- part of the Colorado Plateau, the SierraNevada Mountains, and the Coastal Ranges. These lowed by a discussion ofcommon wayw to esti- physical features have some influenceon moisture availa- mate water consumption by plants. Althoughthe bility and the rates of data presented are for short periods oftime, the evapotranspiration. The information may be valuable for providing consumptive use values for the differentcrops and farm- other vegetation types presented in ers, foresters, gardeners, and other waterusers and this paper represent these features and the climatic conditionsin managers with a good estimate of the ET values of the region. various plants. In the absence of suchinformation, the different groups may alsouse the Blaney- PURPOSE OF THE STUDY Criddle method, which is discussed insome detail, to estimate the average ET value of Estimates of daily evapotranspiration valuesare a plant, or the used extensively to determine plant consumptive use of water by plants. water requirements, and knowledge of the consumptiveuse of different plants is useful indetermining plant water needs as well as designing themost appro- 1School of Forestry, NorthernArizona University, Flagstaff priate irrigation- scheduling scheme.The latter, in 58 Tecle

turn, is important for establishing the right irriga- FACTORS THAT INFLUENCE tion application rates for variouscrops or other EVAPOTRANSPIRATION vegetation types. Projection of the expectedirri- Many factors or variables influence the gation scheduling scheme for amount a plantation during a and rate of ET in an area. The mostimportant of season requires estimation of periodic "expected these can be categorized evapotranspiration" for the different plants as climatic, soil, plant, grown cultural, or a combination of these factors.The in the area. But most importantly, informationon ET would help in adopting climatic factors that may affect the level of ET inan proper water resource area include wind speed, land cover albedo, cloud management, especially where water isscarce such cover, solar radiation, relative humidity, air and as in the U.S. Southwest. dew point temperatures, and precipitation, The purpose of this paper is therefore mainly and to these may be affected byan area's aspect, altitude, present the estimated water consumptive use of and latitudinal and longitudinal position(Dooren- various crops, vegetables,grasses, shrubs, fruit bos and Pruitt 1977). trees, and the most common deciduous and ever- Also the ways these factorsoccur have a direct green trees in the Southwest. This information also bearing on the ET values. In the demonstrates that ET values for the case of precipita- same plants tion, for example, amount, duration,rate, frequen- can differ both spatially and temporally. The varia-cy of occurrence, and the inter -arrival time be- tion is a function of an area's specificclimatic and tween occurrences can affect the ET value of physiographic characteristics. a plant in a particular place. The effects ofsuch variables on ET can be through their influence CONSUMPTIVE USE OF SOME on water availability and changes in temperature. COMMON PLANTS Soil conditions mainly affect plant ET values Evapotranspiration (ET) or plant watercon- indirectly. The effects are usuallya function of soil sumptive use is a measure of the quantity ofwater water availability, restriction in heat used in the processes of plant transpiration movement, and the and reduction in energy absorption. Theeffects of building of plant tissues and the waterevaporated plants on ET, on the other hand, from plant -intercepted precipitation are more direct; during a the most influential factorsare plant cover density, specific period of time. The ET valuesgathered type of plant population, spacing and orientation here are from differentsources and provide results in planting, plant height, rooting depth, of studies made at different time periods. stage of Table 1 growth, shading, and the rate of water uptakeby shows the seasonal evapotranspiration valuesfor plants (Jensen et al. 1990). These factors various crops grown in Arizona and other may have similar either negative or positive effectson the water places in neighboring states, and Table 2 gives the consumption of plants. For example, theremay be specific ET values for certain plantsgrown in the more ET from a plant growing densely than from Owens Valley, CA (Duell 1990) a as well as for a few sparsely spaced one, and shadingmay decrease grasses grown elsewhere in the Southwest (Todd the ET value froma plant while plant size in- 1970). The ET values in Table 1are given for spe- creases it. cific seasons in specific years. Cultural factors may also affect the amount and An interesting point in the data of Table 2 is the rate of ET. These factors are directly relatedto large variation in the saltgrass ET valuesfrom one management activities or can be controlled easily type of cover to another. For example, valuesvary from a low of 385 mm to suit management needs. They include land per year for desert sink preparation schemes, irrigation practices, tillage, scrub type vegetation to 989 mm per year for rushweeding and harvesting methods, surface mulch- and sedge meadow cover types (Duell1990). This ing and shading levels. Other influential means the ET value for any particular factors range plant may arise from the interaction or combination of depends on the type and severity of theprevailing the above factors. For example, ET reaches climatic conditions in an its area; the ET values ofmaximum when uniformly dense vegetationis these range plants or any other plantsin the U.S. growing in water -saturated soils. Southwest should reflect such circumstances. Table 3 provides the annualaverage ET values METHODS OF MEASURING ET for certain hardwood plants and shrubs. However, There are numerous methods of measuringand it is important to note that thesevalues were estimating ET (Shuttleworth 1993). Some require measured during just oneyear, 1960 (Todd 1970; simple measurement suchas using pan evapora- Raymond and Rezin 1989). tion, while others are more complex andinvolve Plant Water Consumption in U.S.Southwest 59

Table 1. Seasonal evapotranspirationamounts (in mm) for well - watered, the Southwest. common crops in Arizona and other areas in

Crop Location Period Years ET Values References Field Crops Barley Mesa, AZ 16 Dec -15 May 1952-53,56 643 Erie et al. 1968 Cotton Mesa & Tempe 1 Apr -15 Nov 1954 -62 1046 Erie et al. 1968 Sugarbeet Mesa, AZ 1 Oct -17 July 1965 -66 Wheat 1054 Erie and French 1965 Mesa, AZ 15 Nov -15 May 1969 -70 582 Safflower Erie et al. 1968 Mesa, AZ 1 Jan -15 July 1958 -60 1153 Soybean Erie et al. 1968 Mesa, AZ 16 June -31 Oct 1944 564 Sorghum Erie et al. 1968 Mesa, AZ 1 July -30 Oct 1955 -58, 60 Flax 645 Erie et al. 1968 Mesa, AZ 1 Jan-30 June 1943 -44 795 Corn Erie et al. 1968 Davis, CA 15 May -20 Sept 1970 -71 640 Corn, sweet Erie et al. 1968 Mesa, AZ 16 Mar -15 June 1959, 61 -62 498 Rice Lourence & Pruitt 1971 Davis, CA 1 May -30 Sept 1968 -69 920 Beans Lourence & Pruitt 1971 Various areas 250 -500 5 Grain Palo Verde, CA 12 months 1981 -1982 579.12 Owen -Joyce & Kimsey 1987 Fruit Crops Grapefruit Phoenix, AZ 12 months 1931 -34 1217 Erie et al. 1968 Oranges Phoenix, AZ 12 months 1931 -34 993 Erie et al. 1968 Plumes Arvin, CA 12 months 1962 -63 1072 State of Calif. 1967,1974 Vineyards Different places Seasonal - 450 -900 State of Calif. 1967, 1974 Vegetable Crops Broccolli Mesa, AZ 1 Sep-14 Feb 1960 -62 500 Erie et al. 1968 Cabbage Mesa, AZ 1 Sept -15 Mar 1960 -62 622 Erie et al. 1968 Cantaloupe Mesa, AZ 1 Apr -15 July 1959 -60 485 Erie et al. 1968 Carrots Mesa, AZ 16 Sept -31 Mar 1960-62 422 Erie et al. 1968 Cauliflower Mesa, AZ 16 Sept -31 Mar 1960 -62 472 Lettuce Erie et al. 1968 Mesa, AZ 16 Sept -31 Dec 1960-62 216 Onion, dry Erie et al. 1968 Mesa, AZ 1 Nov -15 May 1961-62,64 592 Onion green Erie et al. 1968 Mesa, AZ 16 Sept -31 Jan 1960-62 445 Potatoes Erie et al. 1968 Phoenix, AZ 15 Feb -15 June 1959 -63 617 Tomatoes Erie et al. 1968 Seasonal 300 -600 Doorenbos & Pruitt 1977 Forage Crops Alfalfa Mesa & Tempe 12 months 1946,50,62-63 1887 Alfalfa Erie et al. 1968 Palo Verde 12 months 1981 -1982 1615.44 Valley, CA Owen -Joyce & Kimsey 1987 Bermuda Mesa, AZ 16 Apr -15 Oct 1959 -60,63 -64 1105 Erie et al. 1968 Turf Reno, NV 112 days 1%5-67 554 Tovey et al. 1969 Grass Arvin, CA 12 months 1%0-65 1308 State of Calif. 1967, 1974

various activities. The latter type includesthe soil 30-40 cm from the firstcore to minimize sampling depletion method, the use of lysimeters,and a error. In this way, the amount of moisture in the method involving the Bowen ratio. soil samples is measured. The soil water depletion methodinvolves Another method uses lysimeters, measuring the change in soil water which are to indicate the tanks filled with soil in whichcrops or other plants amount of water lost due to ETover a period ofgrow under natural conditions time. In an irrigation system, for example, to measure the soil is amount of water lost by evaporation andtranspi- sampled 2-4 days after irrigationor rain and again ration. This method provides 7 -15 days later or just before the a direct measure- next irrigation ment of ET and is frequently usedto study climatic period. The holes from which the soilcores are effects on ET. This method removed are usually filled with assumes conditions soil, and are inside the lysimeters to be essentiallythe same as marked to enable taking thenext samples some those outside. 60 Tecle

Table 2. Annual evapotranspiration values for range plants in Jan -Dec, 1984 -1985 in theOwens Valley, CA (Duell 1990) and for other grassesin the Southwest (Todd 1970).

Type of Cover Plant Type ET Values (mm) Sources Alkali meadow Alkali sacaton 828 Duell 1990 Rabbitbrush meadow Saltgrass orrubber rabbitbrush 471 Duell 1990 Desert sink scrub Rubber rabbitbrush, alkali sacatonor Mormon tea 606 Duell 1990 Desert sink scrub Saltgrass greasewood 385 Duell 1990 Alkali meadow Saltgrass, alkali sacaton, and rubber rabbitbrush 626 Duell 1990 Alkali meadow Nevada saltbrush, alkali sacaton & rubber rabbitbrush 820 Duell 1990 Rush and sedge meadow Saltgrass, alkali sacaton & Baltic rush 989 Duell 1990 Chaparral Southern California chaparral 508 Todd 1970 Grass type California woodland -grass 457 Todd 1970 Chaparral Arizona chaparral 445 Todd 1970 Grass and shrub Semiarid grass and shrub 269 Todd 1970

Table 3. Annual ET values in 1960 forsome common trees and shrubs in the U.S. Southwest. Plant Type ET Values (mm) Source Lodgepole pine 483 Todd 1970 Engelmann spruce -fir 381 Todd 1970 White pine -larch -fir 559 Todd 1970 Mixed conifer 559 Todd 1970 True fir 610 Todd 1970 Aspen 584 Todd 1970 Pacific Douglas- fir - hemlock and redwood 762 Todd 1970 Interior ponderosa pine 432 Todd 1970 Interior Douglas -fir 533 Todd 1970 Pinion - juniper 368 Todd 1970 Saltcedar 1812.1 Raymond and Rezin 1989 Arrowweed 1628.2 Raymond and Rezin 1989 Mesquite 2585.3 Raymond and Rezin 1989 Cottonwood -willow 712.5 Raymond and Rezin 1989 Mixed stand 1396.8 Raymond and Rezin 1989

A third method is the energy balance method, ET value of an area is touse an evaporation part. which involves the Bowen ratio, which is the ratio The value so determined is then adjustedusing an of upward energy flux as sensible heat to the latent adjustment factor to estimate the actual ETvalues energy flux in the same direction. This method is of specific plants. Other used to determine hourly more complex methods or shorter values of ET used to estimate ET values include the radiation- (Jensen et al. 1990.) based estimating method of Priestley and Taylor (1972), the temperature and radiation methodof ESTIMATING EVAPOTRANSPIRATION Jensen and Haise (1963), the Penman model, which There are a number of methods forestimating is based on combinedenergy and mass transfer reference evapotranspiration (ET). Themethods principles (Penman 1948), the temperature -based differ from one another in terms of the main cli- methods of Hargreaves and Samani (1982), thesoil matic characteristic(s) or variables theyuse. As moisture depletion method (Bell et al. 1987), the noted above, the simplest method to determine the lysimeter method (Aboukhaled et al. 1982),and Plant Water Consumption in U.S. Southwest 61

numerous other techniques (Shuttleworth 1993). obtained from tabular or graphical relationships One method that is commonly used aroundthe like those developed in Doorenbosand Pruitt world to estimate ET that isa little more involved (1977). than the pan evaporation method is the Blaney- Once ETr is determined, the actual ET fora par- Criddle method, described in Doorenbosand ticular plant can be estimated usingan appropriate Pruitt (1977). This method is presented here be- adjustment value, k. Accordingly, theactual evapo- cause it is relatively simple and requires only a few transpiration (ETa) for a particular plantcan be variables to calculate the referencecrop ET values. estimated using: The variables, whichare easy to measure, are mean daily temperature, daily minimum relative ETa = k *ETr [5] humidity, average daily sunshine hours,daytime The value of k in this equationrepresents the wind speed, daily solar radiation, andmonthly ratio of actual plant ET to reference ET.As such it extraterrestrial radiation of anarea. The primary is important that accurate k and referenceET val- equation for this methodmay be expressed as ues be eventually developed for the different plants grown in order to estimate theirconsump- ETr = a + b *[p* (0.46 *T + 8.0)] [1] tive use. For the time being, general referenceET where ET,. = referencecrop ET in mm /day for a rates are developed for the study site byextrapo- particular month; a and b= adjustment factors; p = lating information provided in Yitayew (1990). mean daily percentage of total daylight hours fora The average daily reference ET values(in mm/ given month and latitude; and T= mean dailyday) for the 12 months of theyear are given in temperature in °C. Figure 1. Using the monthlyaverage daily values, The mean daily temperature used in the above the annual total reference ET for the studysite is equation for reference ET can be determinedusing estimated to be 1100 mm (or 43.3 inches;Blee the equation 1988). Then, assuming a k value of 0.75 forthe area, the annual average ET for the site is about825 mm T(° C) =[Tmax ( °C) +Turin( °C)]/ 2 [2] (or 32.5 inches). where Tmax and Taro are, respectively, themaxi- CONCLUSIONS AND RECOMMENDATIONS mum and minimum daily temperatures inan area. The adjustment factor a, used in the Blaney- Knowledge of the rate and amount ofevapo- Criddle method for estimating ET,., is determined transpiration from crops and other agricultural using minimum relative humidity and dailysun- plants is important for proper planning andman- shine hours as shown in the following relationship, agement of water use by plants. Such knowledge is developed for Arizona conditions (Yitayew 1990): especially more useful in arid and semi -aridareas, which face the dual problems of water scarcityand a = 0.0043* [(RHn in) - (n / N)1 -1.41 [3] high plant consumption of water. This information enables farmers and plantation managers where RHmi=, = daily minimum relative to decide humiditywhen and how much water to at the place of interest and n/N use in their projects. = the ratio of actual Also knowledge of the water consumptive to maximum daily sunshine hours. use of plants helps farmers decide the kindsof crops to In the absence of measured data, the valueof plant under uncertain water conditions.For ex- n/N can be estimated using the equation(Yitayew ample, farmers would plant 1990) crops that have lower water consumption values during drought periods n/N = 2 *(Rs /Ra) -0.5 [4] or in arid environments compared to farmers in wetter climates. where RS = solar radiation in millimeters per day A couple of cautionary notesare in order in us- and Ra = extraterrestrial radiation in millimeters ing the ET values of the differentcrops, fruit trees, for the month and latitudinal location ofthe site grasses, shrubs, and trees provided in this study. under consideration. Because most of the ET values presentedare based The value for the other adjustment factor,b, in on single year or single season measurements, the the Blaney -Criddle method also dependsupon amounts may depend on the particular climatic estimates of minimum relative humidity, average conditions prevailing at the time ofmeasurement. daily sunshine hours, and daytime wind speed, as Hence, long -term measurement of ET wouldpro- described in Yitayew (1990) and Doorenbosand vide more reliable data for the variouscrops and Pruitt (1977). The values for these variablescan be other plants listed in this study. 62 Tecle

1 0: - Jan Feb Mar Apr May JunJul Aug SepOctNov Dec Months

Figure 1. Monthly average reference evapotranspirationvalues for northern Arizona (the bars represent average daily values in mm).

REFERENCES CITED

Aboukhaled, A., A. Alfaro, and M. Smith. 1982. Lysime- Erie, L. J., O. F. French, and K. Harris. 1968. Consump- ters. FAO Irrigation and Drainage Paper No. 39. Food tive use of water by crops in Arizona. Arizona Agri- and Agriculture Organization of the United Nations, cultural Experimental Station Technical Bulletin 169. Rome.. Rome. 44 pp. Bell, J. P., T. J. Dean, and M. G. Hodnett. 1987. Soilmois- ture measurement by an improved capacitance tech- Hargreaves, G. H., and Z. A. Samani. 1982. Estimating nique, Part II: Field techniques, evaluation and cali- potential evapotranspiration. Technical Note. Journal bration. Journal of Hydrology 93: 79-90. of Irrigation and Drainage Engineering 108(3): 225- Beschta, R. L. 1976. Climatology of the ponderosa pine 230. type in Central Arizona. Technical Bulletin 228. Col- Jensen, M. E., and H. R. Haise. 1963. Estimatingevapo- lege of Agriculture, Agricultural Experiment Station. transpiration from solar energy. Proceedings of the University of Arizona, Tucson. 24 pp. American Society of Civil Engineers. Journal of Irri- Blee, J. W. H. 1988. Determination of- vaporation and gation and Drainage 89: 15-41. seepage losses, Upper Lake Mary near Flagstaff, Jensen, M. E., R. D. Burman, and R. G. Allen (Editors). Arizona. U.S. Geological Survey, Water Resources 1990. Evapotranspiration and irrigation waterre- Investigation Report 87 -4250. quirements. ASCE Manuals` and Reports on Engi- State of California. 1967. Vegetative wateruse. Depart- neering Practice 70. American Society of Civil ment of Water Resources Bulletin 113 -2.82pp. Engineers, New York. 332 pp. State of California. 1974. Vegetative wateruse. Depart- Lourence, F. J., and W. O. Pruitt. 1971. Energy balance ment of Water Resources Bulletin 113-3. and water use of riceown in the Central Valley of Doorenbos, J., and W. O. Pruitt. 1977. Guidelines For California. Agronomy Journal 63: 827-832. Predicting Crop Water Requirements. FAO Irrigation Owen- Joyce, S. J., and S. L. Kimsey. 1987. Estimates of and Drainage Paper No. 24. Food and Agriculture consumptive use and groundwater return flow using Organization of the United Nations, Rome. 143pp. water budgets in Palo Verde Valley, California. U.S. Duell, L. F. W. 1990. Estimates of evapotranspirationin alkaline scrub and meadow communities of Owens Geological Survey, WRI Report 87- 4070.50 pp. Valley, California, using the Bowen -ratio, Eddy Penman, H. L. 1948. Natural evapotranspirationn from correlation, and Penman combinaton methods. USGS bare soil and grass. Royal Society of London Proceed- Water Supply Paper 2370 -E. U.S. Government Print- ings Series A. 193: 120 -114. ing Office, Washington, D.C. 39 pp. Priestley, C. H., and R. J. Taylor. 1972. On the assessment Erie, L. J., and O. F. French. 1965. Water management of of surface flux and evapotranspiration using large- fall -planted sugarbeets in the Salt River Valley of Ari- scale parameters. Monthly Weather Review 100: 81- zona. Transactions ASAE 11: 792 -795. 92. Plant Water Consumption in U.S. Southwest 63

Raymond, L. H., and K. V. Rezin. 1989. Evapotranspira- tion estimates using remote sensing data, Parker and Palo Verde Valleys, Arizona and California. USGS Water Supply Paper 2334. U.S. Government Printing Office,ashm on, D.C. 18 pp. Shuttleworth., W.gJ.. 1993. Evaporation. In Handbook of Hydrology, edited by D.R. Madment,pp. 4.1 -4.53. McGraw-Hill, New York. Tecle, A. 1990. Hydrology and watershed management in southwestern ponderosa pine forests. In Multi- resource Management of Southwestern Ponderosa Pine Forests: The Status of Knowledge, technical edit by A. Tecle and W. W. Covington,pp. 207 -350. USDA Forest Service, Albuquerque, NM. Todd, D. K. (Editor). 1970. The Water Encyclopedia. Water Information Center, Port Washington, NY. 101

Tovey,R., J. S. Spencer, and D. C. Muckel. 1969. Turf- grass evapotranspiration. Agronomy Journal 61: 863- 867. Yitayew, M. 1990. Reference evapotranspiration estimates for Arizona. Technical Bulletin 266. Avicul- tural Experiment Station, University of Arizona, Tucson. 30 pp. GEOMORPHOLOGY OF SMALL WATERSHEDS IN AN OAK ENCINAL IN THE PELONCILLO MOUNTAINS

Daniel G. Neary1 and Gerald J. Gottfried1

Oak savannas cover large areas in mountains responses of these watersheds before and after fire and high valleys of the southwestern United States is reintroduced into the landscape. and northern Mexico. However, relatively little information is available about these lands to aid in THE SOUTHWESTERN ENCINAL their management (Gottfried et al. 2000a). Twelve The encinal or oak savannas and woodlands, small watersheds on the east slope of the Peloncillo also referred to as the Madrean evergreen forma- Mountains in southwestern New Mexico were tion, are concentrated in the Sierra Madre Occi- selected for study of the hydrology and ecology of dental of Mexico and extend northward into oak savannas, and for evaluating the impacts of southeastern Arizona, southern New Mexico, and cool season and warm season fires on a number of Texas. This formation covers approximately 80,290 ecosystem components. km2 (31,000 mi2; Ffolliott 1999). The differentiation Fire was the most important natural disturb- between woodlands and savannas is based on the ance in these ecosystems prior to European settle- amount of tree canopy closure; the term savanna is, ment. During the twentieth century fires became commonly used for areas with an open tree cover less frequent because of the impacts of past over- and a large, often continuous grass component grazing on the native herbaceous vegetation and (Van Devender 1995). The oak savannas and fire suppression activities (McPherson 1992). Some woodlands are a major vegetation type within the areas currently have high fuel accumulations that Coronado National Forest of southern Arizona and could contribute to detrimental stand -replacingsouthwestern New Mexico, where they occupy wildfires. Prescribed burning is a technique to more than 3430 km2 (1324 mi2; R. E. Lefevre, restore the natural processes within the savannas, personal correspondence, July 1999). Oak and to reduce densities of woody species, to increasejuniper mapping units cover approximately 401 herbaceous plant cover and production, and to km2 (155 mi2) in the Borderlands Project Area create mosaics of vegetation on the landscape. (Muldavin et al. 1998). However, questions remain about the effects of The woodlands and savannas provide a number burning season and fire intensities on this eco- of natural resources and amenities (Ffolliott 1992). system. They supply tree and other vegetative products, A considerable amount of ecological and related such as firewood, fence posts, acorns, and bear - information for the Peloncillo Mountains and the grass (Nolina microcarpa) for local consumption, Southwestern Borderlands Region has been and are important for livestock grazing and wild- collected in recent years through the efforts of the life habitats, especially for threatened, endangered, USDA Forest Service, Rocky Mountain Research or sensitive species. Forty -one species of neotropi- Station, the Malpai Borderlands Group, the cal birds were tallied in the oak woodlands of Animas Foundation, Arid Lands Project, and their southern Arizona, which are important breeding associates (Gottfried et al. 1999a, 2000b). The chan- and migration habitats (Block et al. 1992). Water is nel information presented here was gathered toa critical resource in these arid areas, and water- quantify easily visible differences in the stream shed management activities are designed to pro- channels of the Peloncillo watersheds. This infor- tect soil stability and water quality and to sustain mation is crucial to understanding the hydrologic stream flows. Recreational activities, such as hiking, camping, hunting, and bird watching, are increasing as the population of the region grows, 1USDA Forest Service Rocky Mtn. Research Station, Flagstaff 66 Neary and Gottfried

and they represent a large income producer forstreamflow and surface runoff characteristics of local communities. the oak woodlands. Surface runoff is the result of Annual precipitation in the Borderlands encinal both rainfall and snowmelt; however, high - exceeds 406 mm (16 in). More than half falls in the intensity summer rains produce most of the May through August growing season. The latesurface runoff and can accelerate erosion and sedi- spring and autumn are dry. The proportion ofmentation. Encinal watersheds also are important annual precipitation attributed to summer rains because they often contain channels that transmit decreases from southeastern Arizona to the north- water from the higher elevation conifer forests. west. Extremes in annual precipitation range from Many of the problems on these watersheds are 305 to 1016 mm (12-40 in; Gottfried et al. 1995). linked to past overgrazing of livestock. Channels Encinal landscapes generally occur between 1189 accumulate soil and rocks from the side slopes that and 2195 m (3900 -7200 ft) in elevation. Precipita- are moved by periodic runoff events. Low flows tion and stand densities are observed to increase result in a redistribution of sediments in the chan- with elevation in the Peloncillo Mountains, as has nels whereas larger flows tend to flush sediments also been reported in the Santa Catalina Moun- through the system. Good encinal watershed tains, north of Tucson (Whitaker and Niering condition, based on erosion rates, gully presence, 1964). Stands commonly are in a variety of sites, and soil compaction criteria, is necessary so that including along drainages, on rocky slopes, on accelerated erosion and sedimentation do not alluvial basin fill and fans, and on residual soils ofimpair water quality. This is best accomplished by rhyolitic pediments and elevated plains (USDA maintaining healthy, well- stocked stands of trees Forest Service 1997). Soil type and depth influences and herbaceous vegetation (Lopes and Ffolliott stand structural development. 1992; Baker et al. 1995). Watershed management A large variety of oaks are found in the encinal objectives are to minimize any adverse effects of woodlands. Emory oak (Quercus emoryi) is com- current land use activities to the soil and water mon in the mountains of the southwestern Border- resources, to increase water yields, and to rehabili- lands and can be associated with Arizona white tate degraded watersheds. However, it is unlikely oak (Q. arizonica), gray oak (Q. grisae), and Mexican that management practices will increase stream- blue oak (Q. oblongifolia). These oaks are small, flow from the oak woodlands because of the high often multiple- stemmed, irregularly formed trees. evapotranspiration demands relative to the low Natural oak regeneration from seed is episodic; precipitation. sprouting from stumps and roots is a more com- Research on other aspects of the hydrological mon regenerative mechanism (Borelli et al. 1994). cycle include a report by Haworth and McPherson Redberry juniper (Juniperuserythrocarpa) or (1991) that up to 70 percent of the late summer- alligator juniper (J. deppeana) and border pinyon early fall precipitation is intercepted directly under (Pinus discolor) are intermixed with the oaks on the canopy of Emory oak. Throughfall varied with many sites. Chihuahua pine (P. leiophylla var. the amount of storm precipitation; tree size was chihuahuana) is often found along major drainages. important for smaller storms. Ffolliott and Tree densities are variable, from scattered indi- Gottfried (1999) and Gottfried et al. (1999b) have viduals to several hundred stems per hectare. studied transpiration by Emory oak in the San Stand density has been related to soil properties, Rafael Valley of southeastern Arizona. elevation, and other topographic characteristics, and to fire and human land use histories (Gottfried DeBano et al. (1996) discussed the impacts of et al. 1995). fire on water resources in the Madrean Province ecosystems. The most obvious impact occurs from ENCINAL HYDROLOGY large, high- severity fires that destroy the vegeta- Little information is available about the hydrol- tion and cause a decline in interception by both ogy of the encinal woodlands and savannas (Lopes vegetation and litter, resulting in soil water repel- and Ffolliott 1992; Baker et al. 1995). Much of the lency. The latter can result in increased overland hydrological research in the region has been con-flows and erosion, particularly on steep slopes. A ducted in the Chihuahuan Desert vegetation at reduction in the plant cover could result in a Walnut Gulch, near Tombstone, Arizona (Oster- decrease in evapotranspiration on the watershed kamp 1999). One of the main gaps concerns and an increase in soil water storage. Geomorphology of Watershed in Encinal 67

THE PELONCILLO WATERSHEDS composition of granite. Vincent (1998)reported Twelve small watersheds were selectedon the that the surface geology in the AnimasValley was east side of the Peloncillo Mountains ofsouthwest- formed by ash flow tuffs and lavaflows that ern New Mexico for evaluating the impacts of cool erupted from calderas and smallervents during a season and warm season prescribed burningon period of the Oligocene (36-26 millionyears before oak savanna ecosystemscommon to the region present). Five ancient calderasare in or border the (Gottfried et al. 2000a). Hydrology isa major part Animas Valley. The tectonic tilt isto the east. of the fire experiments planned forthese water-Rhyolite results from chemicaldifferentiation sheds. Climatic and hydrological recordsfor the (separation or isolation of compounds)as mantle area are scarce, making basic hydrological analyses material migrates and melts itsway through the important in the design of streamflowmeasure- earth's crust to vent at the surface. Therocks are ment structures. This region is characterized rich in potassium and silica, and relativelylow in hydrologically by a very widerange of flows. The calcium, iron, and magnesium (Vincent andKrider main instrumentation problemwas to design 1998). Biggs et al. (1999) surveyed thegeology and installations that are not so small that theywould geomorphology in adjacentareas of Arizona and be unable to measureuncommon, but significant, indicated that the formationsare the result of naturally high streamflows andany high flows many separate rhyolite flows that exhibit contorted related to the prescribed burning, butnot so large flow banding. Phenoclasts in the flowsare usually that the majority of morecommon low flows could less than 1 mm (0.04 in) wide andcomprise 1 -5 not be measured accurately. percent of individual flows. Many flows havea The 12 watersheds selected forinstrumention in vitric base that stands outas a dark band and may 2001 are located north of WhitmireCanyon in the have interbeds of lithic tuffs thatare several meters Coronado National Forest, approximately50 km thick. Tuffs are rocks consolidated withvolcanic (31 mi) south of Animas, New Mexico,and north ash. of the Geronimo Trail road. The site iswest of the The hillslopes adjacent to the Animas Valleyare Cascabel portion of the Diamond A Ranch,which eroded and generally consist of colluvium-covered holds the permit for the grazing allotmenton the bedrock or exposed bedrock (Vincent andKrider watersheds. An east -west ridge betweenWhitmire 1998). Slopes are commonly between5 and 30 and Walnut Canyons is the maintopographicpercent. The resulting colluvial soilsare about 0.4 divide. Elevations range from about 1640to 1704 m (1.3 ft) thick and consist of a 5 -10cm (2-4 in) m (5380 -5590 ft). Six watersheds were selected for surface horizon of gravelly silt loamto gravelly study on the north and south sides of theridge clay loam and a 10-25cm (4-10 in) B horizon of between Whitmire and Walnut Canyons. The40-70 percent gravel -sized clasts and intersticesof watersheds have been divided intogroups of clay loam to silty clay and a subsoil of interlocking three. stones. Each set of three watersheds has twoburning The 1991 General Ecosystem Surveyconducted treatments and an untreated control treatment thatby the Southwestern Region of the ForestService will be assigned randomly. The watershedsrange classified the soils in thisarea of the Peloncillo in size from 8 to 36 ha (20-83ac; Gottfried et al. Mountains as Typic Haplustalfs, mesic, deep, 2000a). Watersheds A through Gare on the south gravelly, loam compactedor deep, very cobbly, side of the ridge and H through Nare on the north sandy loam, gullied. Slopesvary from 0 to 40 side. Channel lengths range from 226to 769 mpercent. Another Forest Service soilsurvey in the (743 -2522 ft), and slopesare between 2.5 and 7 area classified 45 percent of the soilsas Typic percent. Areas were determined by GeographicHaplustolls, coarse -loamy, mixed, mesic,25 per- Information System (GIS) procedures, andthe cent as the Typic Haplustalfs, and 15percent rock other characteristics were determined fromU.S. outcrops. One site was classified primarilyas Geological Survey maps andon- the -ground Lithic Ustorthents, loamy, mixed, nonacid,mesic, measurements. typical. Both sites supported Emory oak,Tourney oak (Q. toumeyi), alligator juniper, andredberry GEOLOGY AND SOILS juniper. The basic parent material in the study area is The USDA Soil Conservation Service (Cox1973) rhyolite, a fine- grained igneous rock that hasthe soil survey of Hidalgo County classifiedall of the 68 Neary and Gottfried

soils in the area as rock land with bedrock at 0-30mire Creek drains into Animas Creek. Maximum cm (0-1 ft). The well -drained surface soils, classi- water year discharges ranged from 1.00 m3 /s (35.3 fied in the Lehmans Series, havea permeability of ft3 /s) in 1988 to 95.94 m3 /s (3390 ft3 /s) in 1974. 50-160 mm /hr (2.0 -6.3 in /hr). Rock land slopes Specific discharges of 0.0130 m3 /s /km2 (1.19 ft3/ are between 10 and 25 percent, and in this land s/mi2) had a return period of 1.03years, and those type 30-85 percent of the soil surface consists of of 1272 m3 /s /km2 (115.24 ft3 /s /mi2) hada return exposed bedrock, stones, and cobbles. period of 36 years. Osterkamp (1999) developed graphs of the relationship between unit runoff and HYDROLOGY area. However, the flow that was determined from The baseline data of streamflow and water qual- the graph for a 16.2 ha (40 ac) watershed is much ity generated by this study will increase the knowl- lower than has been observed on the smaller oak edge about the hydrology of small oak watersheds experimental watersheds. This is probably because in this region that can be used for landmanage- small watersheds often are flashier than larger ment activities. The initial task was to review areas where runoff is distributed over a large area. existing information (Gottfried et al. 2000a). CoxOsterkamp's data are mostly from the Walnut (1973) indicated that annual precipitation in theGulch Watersheds, which are on well -drained rock land units in Hidalgo County ranged from sandy loam soils derived from fan deposits and 305 to 457 mm (12 -18 in). The highest annual alluvium. precipitation recorded prior to 1973 was 743mm Field estimates of peak discharge were deter- (29.27 in) at the old Cloverdale Ranger Station, mined for the Peloncillo experimental watersheds south of the study area, and the highest monthly using channel slope, cross -sectional area measure- total was 257 mm (10.12 in) at the Dunagan Ranch, ments, and high -water marks along the channel north of the area. Runoff from theseareas is rapid, (Gottfried et al. 2000a). Discharge was calculated and the hazard of water erosion is moderate. The using the Chezy -Manning Equation (Linsley et al. water supplying capacity of the soil is between 127 1958). A roughness value (n) of 0.052 wasas- and 203 mm (5-8 in). sumed, indicating a winding channel with pools, Osterkamp (1999) determined runoff and sedi- shallow stages, and large rocks. Peak flow esti- mentation for the Upper Animas Creek Basin mates ranged from 0.06 to 0.79 m3 /s (2 -28 ft3 /s). using records from the Walnut Gulch Experi- The low -flow channels contain rocks and logs, and mental Watersheds, the San Simon Wash Valley in occasional pools and adjacent banks are covered Arizona, and the Jornada Experimental Range in with relatively dense herbaceous vegetation and south -central New Mexico. The nearest long -term oak trees. Roughness values would vary along the weather records from Animas, New Mexico, whichcourse of the stream and by stage, and some sec- is on the eastern base of the Peloncillo Mountains tions would have lower n values of 0.035 or 0.042. at an elevation of 1355 m (4445 ft), indicatean Peak stream discharges were estimated using average annual precipitation of 288 mm (11.32 in). alternative techniques. The first technique was to Ben Brown, Diamond A Ranch, provided precipi- utilize discharge equations developed by the U.S. tation data from recently established weatherGeological Survey for the southwestern states stations located south and east of the studyarea at (Thomas et al. 1994). The 2, 5, and 10yr recurrence an elevation of 1554 m (5100 ft). These records intervals were calculated and compared with the were only for 2 years (1998 and 1999). They dem- discharges calculated from the field data. The onstrate the variability of precipitation in the area. field- estimated values averaged higher than the 2 Approximately 44 percent of the 417 mm (16.42 in) yr USGS return interval for five of the watersheds, of the annual precipitation in 1998 fell in the April but all were less than the 5 yr and 10 yr estimated through September growingseason, whereas discharges (Gottfried et al. 2000a). The Natural almost all of the 495 mm (19.50 in) recorded in Resources Conservation Service (NRCS) developed 1999 fell during this period. Several recording a similar procedure to simulate a unit hydrograph weather stations will be established within the for streams where rainfall and streamflow dataare experimental area to measure local conditions. unavailable (USDA Soil Conservation Service Osterkamp (1999) presented data from a U.S. 1984). This model assumes that some of thepre- Geological Survey gauging station thatwas active cipitation remains on the watershed and does not in Upper Animas Creek from 1959 through 1994. contribute to runoff. Site -specific information, plus The station, which measured flows from the 76.4 tables and charts in the NRCS surface runoff km2 (29.5 mil) watershed,was near where Whit- manual, were used to calculate peak discharges for Geomorphology of Watershed in Encinal 69

2, 5, and 10 yr recurrence intervals (Gottfried et al. Table 1 presents the resultsof the Peloncillo 2000a, Table 2). Values calculatedby the NRCS channel survey. It includes method for all three return intervals a subdivision of Water- were substan- shed E into Ea and Eb becauseof the differences tially higher than the fieldestimates. observed in the two forks. Channeldistances ranged from 720 to 1020 CHANNEL CONDITION SURVEY m (2362 -5784 ft) with a mean of 1020 m (3346 ft). Therewas indeed a great Because of easily observed differencesin channel conditions and the wide disparity in percentage of individualwatershed range of values in lengths with rock channels (6.4%in Watershed A peak discharged produced by thefield estimation, to 41.0% in Watershed K). This USGS, and NRCS methods, could make a a decision was made to significant difference in totalwater yields as well determine the character of all thechannels in the as storm peakflow response. The watersheds 12 Peloncillo watersheds. A with consensus opinion was the higher percentages of bedrock reached among the principal channels are apt investigators that the to be much more flashy in nature thanthe lower physical condition of the channelmight strongly influence hydrologic ones. Sub -watershed Eb has almost twice thechan- response, especially where nel length in bedrockas does its pair, Watershed long reaches of channelwere only bedrock outc- Ea. These two sub -watersheds also rops. vary consider- ably in fine and coarse alluvium, Line transect surveys with Watershed were conducted on all Eb having more fine alluvium(12.5 %) than its pair channels, including side channels,up to the point Ea (1.1 %), as well as much less where a distinct channel could coarse alluvium. not be determinedThere was also a largerange between all the and it blended in with theupper slope topogra- Peloncillo watersheds in fine alluvium phy. A 100 m tape was extended (0.2- 35.7 %) from the main and coarse alluvium (18.3- 74.0 %). The watersheds Parshall flume upslope along thechannels inwith the higher percentages of alluvium sequential stages until each channel in their was indeci- channels might prove to be less flashyand have pherable. Lengths of channel in thefollowinglower, but sustained flows since conditions were measured: bedrock, their channels coarse would have a larger in- channelwater storage alluvium, fine alluvium, vegetation,woody debris, capacity. There was far less variation and other. Accumulated lengths in in the re- these condition maining three categories (vegetation,woody classes were summed for each watershed. debris, and other) between the watersheds.

Table 1. Channel characteristics of thePeloncillo Mountains experimental watersheds, New Mexico. Coronado National Forest,

Percent of Channels Fine Coarse Woody Watershed Distance (m) Rock Alluvium Alluvium Vegetation Debris Other A 420 6.4 24.7 62.4 4.6 0.7 1.2 B 750 8.6 4.9 73.6 10.7 1.4 0.8 C 685 17.8 0.2 63.3 13.3 1.1 4.3 E 1763 33.5 8.8 38.1 15.9 1.3 2.4 Ea 571 21.6 1.1 56.8 15.1 2.8 2.6 Eb 1192 39.2 12.5 28.6 15.7 1.6 2.4 F 881 21.4 5.4 42.0 27.4 3.0 0.8 G 542 12.0 20.1 34.0 31.7 1.4 0.8 H 1378 11.2 16.2 68.2 3.7 0.7 0.0 I 1622 15.9 3.0 74.0 6.4 0.7 0.0 J2 962 29.2 25.8 31.8 11.7 0.6 0.9 K 875 41.0 27.3 18.3 9.4 0.6 3.4 M 1466 14.0 5.0 67.0 8.0 2.5 3.5 N 1175 26.6 35.7 21.1 10.8 1.5 43 Mean 1020 21.3 11.5 47.1 13.2 1.4 2.0 Minimum 420 6.4 0.2 18.3 3.7 0.6 0.0 Maximum 1763 41.0 35.7 74.0 27.4 2.8 4.3 70 Neary and Gottfried

CONCLUSIONS Ffolliott, P. F. 1992. Multiple values of woodlands in the southwestern United States and northern Mexico. In There is sufficient variation in the geomorphol- Ecology and Management of Oak and Associated ogy the Peloncillo experimental watersheds to Woodlands: Perspectives in the Southwestern United States and Northern Mexico, technical coordinators, P. expect pre -fire and post -fire differences in total F. Ffolliot, G. J. Gottfried, D. A. Bennett, V. M. Her- water yield and peakflow discharges that are nandez C., A. Ortega- Rubio, and R. H. Hamre, pp. 17- related to channel condition. The percentage of 23. April 27-30, 1992, Sierra Vista, AZ. General Tech- nical Report RM -218. USDA Forest Service, Rocky channel lengths that are bedrock range from 6.4 to Mountain Forest and Range Experiment Station, Fort 41.0 percent. The watershed hydrologic data will Collins, CO. need to be analyzed with these differences in Ffolliott, P. F. 1999. Encinal woodlands in the southwestern United States. In Ecology and Management of mind. The data will supplement research on the Forests, Woodlands, and Shrublands in the Dryland hydrology of the larger Upper Animas Valley Regions of the United States and Mexico: Perspectives Basin (Osterkamp 1999) by providing information for the 21st Century, edited by P. F. Ffolliott and A. Ortega- Rubio, pp. 69-82. University of Arizona- on the dynamics of small upstream watersheds in Centro de Investigaciones Biologicas del Noroeste, S. encinal woodlands of southeast Arizona and C.-USDA Forest Service. southwest new Mexico. Ffolliott, P. F., and G. J. Gottfried. 1999. Water use by Emory oak in southeastern Arizona. Hydrology and Water Resources in Arizona and the Southwest 29: 43- LITERATURE CITED 48. Baker, M. B. Jr., L. F. DeBano, and P. F. Ffolliott. 1995.Gottfried, G. J., P. F. Ffolliott, and L. F. DeBano. 1995. Hydrology and watershed management in the Forests and woodlands of the sky islands: Stand Madrean Archipelago. In Biodiversity and Manage- characteristics and silvicultural prescriptions. In ment of the Madrean Archipelago: The Sky Islands of Biodiversity and Management of the Madrean Southwestern United States and Northwestern Archipelago: The Sky Islands of Southwestern United Mexico, technical coordinators L. F. DeBano, P. F. States and Northwestern Mexico, technical coordi- Ffolliott, A. Ortega- Rubio, G. J. Gottfried, R. H. nators L. F. DeBano, P. F. Ffolliott, A. Ortega- Rubio, Hamre, and C. B. Edminster, pp. 329 -337. September G. J. Gottfried, R. H. Hamre, and C. B. Edminster, pp. 19- 23,1994, Tucson, AZ. USDA Forest Service, Rocky 152 -164. September 19-23, 1994, Tucson, AZ. USDA Mountain Forest and Range Experiment Station. Forest Service, Rocky Mountain Forest and Range General Technical Report RM -GTR -264. Fort Collins, Experiment Station. General Technical Report RM- CO. GTR -264. Fort Collins, CO. Biggs, T. H., R. S. Leighty, S. J. Skotnicki, and P. A. Pear- Gottfried, G. J., L. G. Eskew, C. G. Curtin, and C. B. three. 1999. Geology and Geomorphology of the San Edminster. 1999a. Toward Integrated Research, Land Bernardino Valley, Southeastern Arizona. Arizona Management, and Ecosystem Protection in the Malpai Geological Survey Open -File Report 99 -19. Arizona Borderlands: Conference S January 6-- 8,1999, Geological Survey, Tucson. 20 pp. Douglas, AZ. Proceedings R11P -10. USDA Forest Service, Rocky Mountain Research Station, Fort Col- Block, W. M., J. L. Gamy, K. Severson, and M. L. Morri- lins, CO. 136 pp. son. 1992. Use of oaks by neotro ical migratory birds Gottfried, G. J., P. F. Ffolliott, G. Schiller, and Y. Cohen. in the Southwest. In Ecology and Management of Oak 1999b. Water use by dryland oaks. In IALC Land and Associated Woodlands: Perspectives in the South- Management Workshop: A Tenth Anniversary western United States and Northern Mexico, technical Celebration of USFS, CSREES, Israeli, and IALC coordinators, P. F. Ffolliot, G. J. Gottfried, D. A. Ben- Partnerships, pp. 3-4. International Arid Lands nett, V. M. Hernandez C., A. Ortega- Rubio, and R. H. Consortium, Tucson, AZ. Hamre, p .65-70.April 27 -30,- 1992, Sierra Vista, AZ. General Technical Report RM -218. USDA Forest Ser- Gottfried, G. J., D. G. Neary, and R. J. Bemis. 2000a. vice, Rocky Mountain Forest and Range Experiment Watershed characteristics of oak savannas in the Station, Fort Collins, CO. southwestern borderlands. Hydrology and Water Resources in Arizona and the Southwest 30: 21 -28. Borelli, S., P. F. Ffolliott, and G. J. Gottfried. 1994. Natur- Gottfried, G. J., C. B. Edminster, R. J. Bemis, L. S. Allen, al regeneration in encinal woodlands of southeastern and C. G. Curtin. 2000b. Research support for land Arizona. The Southwestern Naturalist 39: 179-183. management in the southwestern Borderlands. In Cox, D. N. 1973. Soil Survey of Hidalgo County, New Land Stewardship in the 21st century: The Contribu- Mexico. U.S. Department of Agriculture, Soil Conser- tions of Watershed Management, technical coordi- vation Service and Forest Service, Washington D.C. 90 nators P. F. Ffolliott, M. B. Baker Jr., C. B. Edminster, PP. M. C. Dillon, and K. L. Mora, pp. 330 -334. March 13- DeBano, L. F., P. F. Ffolliott, and M. B. Baker Jr. 1996. 16, 2000, Tucson, AZ. Proceedings RMRS -P -13. USDA Fire severi ty effects on water resources. In Effects of Forest Service, Rocky Mountain Research Station, Fort Fire on Madrean Province Ecosystems: A Symposium Collins, CO. Proceedings, technical coordinators P. F. Ffolliott, L. F. Haworth, K., and G. R. McPherson. 1991. Effects of DeBano, M B. Baker Jr., G. J. Gottfried, G. Solis- Garza, Quercus emoryi on precipitation distribution. Journal of C. B. Edminster, D. G. Neary, L. S. Allen, and R. H. the Arizona -Nevada Academy of Science: Proceedings Hamre,p.77-84.March 11-15, 1996,Tucson, AZ. Supplement 21: 20. USDAForestService, Rocky Mountain Forest and Linsley, R. K. Jr., M. A. Kohler, and J. Paulhus. 1958. Range Experiment Station. General Technical Report Hydrology for Engineers. McGraw -Hill, New York. RM -GiR -289, Fort Collins, CO. 340 pp. Geomorphology of Watershed in Encinal 71

Lopes, V. L., and P. F. Ffolliott. 1992. Hydrology and Thomas, B. E., H. W. Hjalmarson, and S. D.Waltemeyer. watershed management of oak woodlandsin south- 1994. Methods for Estimating Magnitude eastern Arizona.In Ecology and Management of Oak and Fre- and Associated Woodlands: Perspectives quency of Floods in the Southwestern United States. in the South- U.S. Geological Survey Open -File Report93 -419. western United States and Northern Mexico, technical Tucson, AZ. 211 pp. coordinators, P. F. Ffolliot, G. J. Gottfried, D. A.Ben- USDA Forest Service. 1997. Plant associations of nett, V. M. Hernandez C., A. Ortega- Rubio, and R. H. Arizona and New Mexico. Volume 2: Woodlands.USDA Hamre, pp. 71 -77. April 27 -30, 1992, Sierra Vista,AZ. Forest Service Southwestern Region, Albuquerque, General Technical Report RM -218. USDA Forest Ser- NM. 196 p . vice, Rocky Mountain Forest and Range Experiment USDA Soil Conservation Service. 1984. Surface runoff. Station, Fort Collins, CO. Chapter 2 in Engineering Field Manual for Conserva- McPherson, G. R. 1992. Ecology of oak woodlandsin tion Practices. SCS AS-45. USDA Soil Conservation Arizona. In Ecology and Management of Oak and Service, Washington D.C. Associated Woodlands: Perspectives in the South- Van Devender, T. R. 1995. Desert grassland history: western United States and Northern Mexico, technical Changing climates, evolution, biogeography, and coordinators, P. F. Ffolliot, G. J. Gottfried, D. A. community dynamics. In The Desert Grassland, edited Bennett, V. M. Hernandez C., A. Ortega- Rubio, and R. by M. P. McClaran and T. R. Van Devender,pp. 68-99. H.Hamre,pp.24-33.April 27-30, 1992, Sierra Vista, University of Arizona Press, Tucson. AZ. General - Technical Report RM -218. USDA Forest Vincent, K. R. 1998. Tectonics and Earthquake Hazards Service, Rocky Mountain Forest and Range Experi- of the Southern Animas Creek Valley, Hidalgo Coun- ment Station, Fort Collins, CO. ty, New Mexico. New Mexico Bureau of Mines and Muldavin, E., V. Archer, and P. Neville. 1998. A Vegeta- Mineral Resources, Santa Fe. 22 pp. tion Map of the Borderlands Ecosystem Management Vincent, K. R., and P. R. 'Crider. 1998. Geomorphic Area. University of New Mexico, New Mexico Natural Surface Maps of the Southern Animas Creek Valley, Heritage Program, Albuquerque. 58 pp. Hidalgo County, New Mexico. New Mexico Bureau of Osterkamp, W. R. 1999. Runoff and Sediment Yield from Mines and Mineral Resources Open -File Report No. Proxy Records: Upper Animas Creek Basin, New OF -429, Santa Fe. 59 pp. Mexico. Research Paper RMRS- RP -18. USDA Forest Whitaker, R. H., and W. A. Niering. 1964. Vegetation of Service, Rocky Mountain Research Station, Fort Col- the Santa Catalina Mountains, Arizona: A gradient lins, CO. 50 pp. analysis of the south slope. Ecology 46: 429 -452.