Converting Oak Woodland Or Savanna to Vineyards May Stress Groundwater Supply in Summer

Research Article ▼ Converting oak woodland or savanna to vineyards may stress groundwater supply in summer

by Mark Grismer and Caitlin Asato

Water resources are important to land-

use planning, especially in regions where YinYang/iStockphoto converting native oak woodlands or savannas to wine grape vineyards may affect the amount of water available for restoring salmon runs. Research has shown that woodland conversion to grasslands (for possible rangeland graz- ing) leads to greater and more sustained stream flow and groundwater recharge; however, little information is available about woodland conversion to vineyards. To inform resource managers and planners, we developed a water balance model for soil and applied it to vineyards, native oak woodlands and annual grasslands to evaluate their relative use In areas where oak woodlands are converted to vineyards, the recharge of groundwater to streams may change. The authors developed a water balance model for soil, which they applied to vineyards, of groundwater. We applied the model oaks and grasslands. Water use, groundwater recharge to Sonoma County, using climate data savanna conversions have been raised in from 1999 to 2011, and determined that terms of losses in landscape ecological Hydrologists have long been interested oak tree canopy coverage of 40% to 60% diversity (Heaton and Merenlender 2000), in the impacts on soil water conditions results in annual groundwater extrac- adverse impacts on soils (Jackson et al. and groundwater recharge when convert- 1990), soil erosion, water quality (Hinck- ing native grasslands or woodland to tion equivalent to that of an established ley and Matson 2011) and water quantity. agriculture in the world’s semiarid and irrigated vineyard. However, vineyard One community concerned about vine- arid regions. Hydrologic analyses con- groundwater use far exceeded that of yard expansions referred to them as the ducted decades ago at the UC Hopland “tentacles of the wine-grape octopus” Research and Extension Center in coastal oak woodlands in late summer to early (Parrish 2011), and local scientists in So- Mendocino County considered land con- fall, which could further stress already noma County have raised concerns about version in terms of watershed water yields affected groundwater resources. We also biodiversity (Community Foundation (see page 145); Burgy and Adams (1977) evaluated the prediction sensitivity of 2009). These concerns are amplified by cli- characterized the focus at that time: mate change, which increases plant water the model to key parameters associated demands, possibly resulting in decreased Quantitative studies of the hy- with rain levels, soil water–holding ca- groundwater availability. And yet, wine drologic responses of watersheds pacity and irrigation management. grape (and associated wine) production where dense vegetative cover has is the leading agricultural commodity in been replaced with range and for- California in terms of net dollar value. age grasses have consistently shown ine grape production in Califor- In Sonoma County, wine grape pro- increases up to 50% or more (equiva- Wnia coastal counties has increased duction increased from less than 50,000 lent to 3- to 5 acre-inches per acre) in steadily during the past few decades, acres in 1998 (Merenlender 2000) to a peak annual runoff over long periods of often resulting in the conversion of native of 70,000 acres in 2002 and has leveled off measurement. These runoff studies oak woodlands or savannas into vine- at around 60,000 acres during the past cover the variety of conditions found yards. Generally, oak woodlands have decade (fig. 1). Concerns about the impacts tree canopy covers greater than about of vineyard water use on salmon runs are 40% by area, while oak savannas are pre- focused here (Lohse et al. 2008) because Online: http://californiaagriculture.ucanr.edu/ dominantly grasslands interspersed with the county includes a large part of the landingpage.cfm?article=ca.v066n04p144&fulltext=yes oaks. Concerns about oak woodland and Russian River watershed. DOI: 10.3733/ca.v066n04p144

144 CALIFORNIA AGRICULTURE • VOLUME 66, NUMBER 4 in Northern and Central California. 70 About half of the yield increase 65 occurs in the latter portion of the season, giving usable flow in dry pe- 60 riods. The balance of the increase is 55 produced as increased outflow dur- 50 ing the post-storm periods. 45

After oak woodlands were converted 40 to annual grassland, the researchers 35 found increased storm runoff volumes and the establishment of perennial 30 summer stream flows. Base flows had Grape acreage (1,000s) 25 increased and overall groundwater demand dropped when oaks were re- 20 moved. Currently, watershed manage- 15 ment encompasses a broader perspective, White grape 10 Red grape considering the land conversion im- Total wine grape pacts on ecological diversity, soils and 5 water quality. 0 Similar observations have been made 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 of increased groundwater recharge fol- Fig. 1. Annual wine grape production in Sonoma County, CA. Source: County annual agricultural lowing the conversion of native land crop reports. use (rangeland) to agriculture in the U.S. Southwest, Central Plains and High Plains (Gurdak et al. 2007; McMahon et al. Glossary 2006; Scanlon et al. 2005), and the Murray- Darling basin of Australia following con- Crop coefficient c(K ): Ratio of actual below; such recharge typically occurs version of native woodland (mallees) to crop water use to reference ETo; used to later in the rainy (winter) season. dryland farming and pasture (Thorburn determine irrigation water demand in Runoff: That part of the precipitation et al. 1991). vineyards and other crops. that appears in surface streams; may Scanlon et al. (2005) noted that ground- Effective rainfall: Fraction of total come either from the surface, or from water recharge under rangeland use is rain that infiltrates the soil after losses shallow groundwater (the latter is usu- typically nonexistent in arid regions and to leaf interception, surface runoff, de- ally referred to as interflow). quite small in some semiarid regions pression storage and evaporation. Soil water balance: Relatively simple (~ 1 millimeter per year); it increases to Evapotranspiration, plant: Sum of model that accounts for daily changes roughly tens of millimeters annually in soil water and canopy (leaf) evapora- in water storage in the root zone, as- dryland agriculture and hundreds of mil- tion, plus plant transpiration, a pro- sociated with such processes as root limeters in irrigated agriculture. The an- cess in which water moves through a water extraction (ETc), infiltration by nual recharge rate in semiarid regions is plant or tree and is subsequently lost rain or irrigation, and seepage losses to likely rainfall dependent; for example, in through stomata in the leaves. groundwater. a drip-irrigated coastal orchard, Grismer Evapotranspiration, consumptive use Water inputs to the root zone: et al. (2000) measured rainfall-driven (ETc): Root extraction of available soil Include effective rainfall (the fraction groundwater recharge rates of about 180 water used in plant transpiration. of total rain that infiltrates the soil after millimeters per year in both irrigated and Evapotranspiration, reference (ETo): losses to leaf interception, surface run- fallow areas from 1996 to 1998. Evapotranspiration possible from a tall off, depression storage, and evapora- Clearly, when regional water supply fescue grass crop when there is no limi- tion) and irrigation (in vineyards only allocations are based on water balances tation on available soil water. in this study). that include groundwater resources, rates Groundwater: Soil water stored in Water yield: Runoff from the drain- of groundwater recharge and lag times the root zone, or the combination of age basin, including groundwater to the depths associated with differ- this and deep groundwater below the outflow that appears in the stream ent land uses are critical (Grismer 2012; root zone, stored in water tables or plus groundwater outflow that leaves Sophocleous 2005). Scanlon et al. (2005) aquifers. the basin underground. Roughly, underscored that developing “sustainable Groundwater recharge: In the model, at the basin scale, water yield is the land-uses requires quantitative knowl- infiltration water that exceeds the ca- net precipitation minus the total edge of the linkages between ecosystem pacity of the root zone compartment evapotranspiration. change, recharge and groundwater qual- and percolates to deep groundwater ity.” An effort to understand water use in

http://californiaagriculture.ucanr.edu • October–December 2012 145 oak woodlands or savannas in terms of (2008) and Chen, Rubin et al. (2008) for while ETc for grasslands was about 300 these linkages has been under way during California oak savannas. millimeters. the past several years. Baldocchi et al. (2004) quantified the Miller et al. (2010) recorded ground- rates of canopy evaporation to soil mois- water uptake by California oak trees at Water use in oak woodland, grassland ture in California oak woodland and rates ranging from 4 to 25 millimeters While an understanding of vineyard grasslands and found that ETc rates for per month during June, July and August, water use has developed during the past the grasslands declined when volumet- representing about 80% of total ETc. They few decades, only in the past several ric water content dropped below 15%, suggested that “blue oaks should be con- years has there been research directed at corresponding to a soil water potential sidered obligate phreatophytes,” that is, measuring rates of water use in native oak of −1.5 mega-Pascals (MPa). (Soil water water-loving plants similar to, for exam- woodland and grassland systems, and potential is the measure of the relative ple, riverbank willows. They noted that their possible impacts on groundwater strength of water attachment to soil; the available groundwater, including deep resources (see box; fig. 2). more negative the “potential,” the more soil storage, provides a buffer to changes Teuling et al. (2006) summarized the tightly bound the soil water.) They found in the native oak tree hydroclimate. relationships between soil moisture and that at soil water potentials below −2.0 Following up on earlier studies of oak plant evapotranspiration (ETc) for a broad MPa, transpiration in grasslands effec- trees, David et al. (2004) and Baldocchi range of grasslands, native brush, tree tively ceased and the grass senesced. On et al. (2010) noted that the adaptation and savanna landscapes across the world, the other hand, the oak trees continued to potential of savanna vegetation dif- in an effort to establish the parameters for transpire, albeit at low rates, under very fered to accommodate low soil moisture regional and global atmospheric model- dry soil conditions (soil water potentials conditions during the summer and fall ing. Generally, they found that evapo- less than −4.0 MPa), probably due to deep in Mediterranean climates. In sparse transpiration rates declined exponentially root access to groundwater. Overall, an- evergreen oak woodlands of southern with decreasing soil moisture, which was nual woodland ETc was about 380 mil- Portugal, David et al. (2004) monitored later confirmed by Chen, Baldocchi et al. limeters at about 40% canopy coverage, sap flows (water use) in eight points of a

A closer look: The hydrologic cycle

While Earth’s total water content is virtually constant, Water vapor storage water itself continually changes state between liquid, in atmosphere solid and gas (fig. 2). It also moves and resides in different places, or compartments, shifting at different rates between them. Evapotranspiration Hydrologists view soil water balances partly in terms of water residence times. Water resides for different lengths of time in the atmosphere, on the soil surface, and within the soil root zone, a stream or Rain, snow, condensation groundwater. It may take a few minutes for a drop of Evaporation water to evaporate from a leaf, or it may exist frozen in a glacier for 10,000 years. Transpiration When large amounts of water shift suddenly due to human use or natural causes, the species and ecosys- Interception storage tem services supported in each compartment shift. If on plants municipalities or growers withdraw a large amount of Throughfall well or stream water in a drought, it could affect species dependent upon the stream, and groundwater levels may not recover for months, or even years. Scientists develop models to better understand soil Surface storage water balances based on conditions in a specific water (soil surface) basin. This study marshals data from a 12-year period Stream ow Storage in plants in Sonoma County, a major part of the Russian River In ltration storage basin, and estimates water movements. — Janet White Soil water storage

Output Groundwater storage (water yield) Fig. 2. The hydrologic cycle. Adapted from Encyclopedia of the Earth (Hubbart JA, et al. 2010, www.eoearth.org/article/Hydrologic_cycle).

146 CALIFORNIA AGRICULTURE • VOLUME 66, NUMBER 4 Holm oak (Quercus rotundifolia) stem over woodland continued through the night 2 years (May 1996 to August 1998). Leaf during summer and fall, underscoring

water potentials, canopy conductance, that groundwater satisfied this continu- Clark Jack Kelly whole-plant hydraulic conductance and ous water demand. meteorological processes were also mea- Vineyard water use sured, and all evidence indicated that the trees did not suffer water stress and Vineyard water use once vines are remained adequately watered throughout established has been extensively studied the 2 years. The greatest evapotranspira- to develop appropriate irrigation man- tion rates occurred during the summer agement and scheduling and to achieve when little, if any, rain or available soil particular grape quality and yields. Allen moisture was accessible, suggesting et al. (1998) summarized some of the that the root system’s direct access to a original evapotranspiration (ETc) crop co- 42-foot-deep water table met the tree’s wa- efficients (referred to as cK , the ratio of ac- ter demand. tual crop water use to reference ETo) used Likewise, Paço et al. (2009) noted in to determine irrigation water demand in Portugal that grass transpiration stopped vineyards. Pritchard (2010) refined them during the summer as surface soils dried, for California wine grape production, as but trees continued transpiring because did Caprile (2007) for the North Coast, in- of deep root access to groundwater. Tree cluding Sonoma County. transpiration represented more than half Grapevines are fairly drought tolerant of the ecosystem’s transpiration, in spite and can be managed with deficit irriga- of the low tree density (crown cover of tion to obtain the levels of sugars, tannins 21%). Transpiration accounted for 76% and and acids desired by the winemaker. For Grapevines are largely dormant after harvest, tree-canopy leaf rainfall interception loss premium wine grape production, the allowing winter rains to leach soils and replenish moisture in the root zone. Above, an Alexander for only 24% of overall tree evaporation. plants are often water stressed in late Valley vineyard. Baldocchi et al. (2010) examined mul- summer or fall to reach desirable grape tiple years of data from five comparable sugar and tannin levels, with less consid- Based on our observations of tree densi- evergreen oak woodlands in France, Italy, eration of yield. Following grape harvest, ties, our corollary hypothesis was that a Portugal and California. Confirming the vines are largely dormant, and winter certain percentage of oak tree canopy cov- a previous study of California oaks rains effectively leach soils and replenish erage (i.e., areal tree density) alternatively (Baldocchi and Xu 2007), they found that water available in the root zone for bud results in annual water use or ground- Mediterranean oaks survived in the sea- break in the late spring. water recharge equivalent to that from a sonally hot/dry, wet/cool conditions by vineyard in the same landscape. Study objectives ensuring that their ETc was less than the Our objectives were to determine daily available water supply. We hypothesized that following vine- and annual soil water use and deep per- Lastly, Fisher et al. (2007) found that yard establishment, annual groundwater colation (recharge) rates from a vineyard water use (sap flow) in Sierra Nevada oak recharge exceeds that from oak savannas. and oak savanna or woodland under average climate conditions of the past decade in Sonoma County, and to determine the amount of tree canopy coverage that would result in net groundwater use or net recharge equivalent to that of a vineyard.

UC Regents/ANR RepositoryUC Regents/ANR Soil water balance methodology We based our soil water process modeling on daily water balances in the root zones of a hypothetical vineyard, grassland and oak tree. Soil water balance is a relatively simple model that accounts for daily changes in soil water storage of the root zone associated with such processes as root water extraction (ETc), infiltration by rain or irrigation and deep percolation losses to groundwater. We then combined the grassland and oak tree balances to Water use by vineyards has been extensively studied to inform irrigation management and ensure create one balance for an oak woodland grape quality and yields. Above, wine grapes in Sonoma County’s Valley of the Moon region. savanna of variable tree density.

http://californiaagriculture.ucanr.edu • October–December 2012 147 Water inputs to the root zone include Effective rain (plus irrigation for effective rainfall (the fraction of total rain grapevines) minus plant consumptive that infiltrates the soil after losses to leaf use that was in excess of available water M. Wisniewski interception, surface runoff and depres- storage capacity in soil in the root zone sion storage, and evaporation) and irriga- became deep percolation or groundwater tion (vineyard only). Plant consumptive recharge. Grismer et al. (2000) observed water use (ETc), that is, water taken up by that rain-driven groundwater recharge the grass or trees, removes available soil reached depths of nearly 18 feet within moisture from the root zone. 4 to 6 months in drip-irrigated coastal In the model (fig. 3), oak trees were orchards; therefore, we assumed that lag allowed to extract groundwater to meet times for groundwater recharge (Grismer ETc demand when available soil moisture 2012) were likely less than 1 year. In other was zero. No additional water source was words, the previous year’s recharge was available to the grasslands beyond root available for irrigation pumping in the zone soil moisture, and in the vineyard next year. In comparing oak savannas ETc demand in excess of available soil with vineyards, we focused on annual moisture was met through irrigation groundwater use for each system. In this study, groundwater use by oak woodlands (groundwater pumping). Irrigation events To run the model, we used reference was equivalent to that of vineyards when tree were triggered when water was depleted evapotranspiration (ET ) and rainfall canopy cover was about 50%. Groundwater o demand by trees was greatest in the late in the root zone to a specified fraction depths averaged daily from four me- summer and early fall, and averaged 270 (maximum allowable depletion, MAD) of teorological stations across Sonoma milliliters annually. its available water capacity. County (Bennett Valley, East Petaluma, Santa Rosa grape vineyards (table 1). The original Kc and Windsor) values for oak trees were derived from to develop a those for olives (Allen et al. 1998) and continuous, modified based on studies by Baldocchi approximately et al. (2004), which indicated that oak tree E ective rain = Re 13-year record as ETc was 40% to 50% of annual ETo in the input data. We Sierra foothills. used seasonally In this model, plant consumptive use Evapotranspiration dependent Kc (as annual fractions of ETo) was consistent (ETc ) values for plant with published studies: 23% for the vine- consumptive use yards and 52% for oak trees (the results from Allen et al. were slightly higher in our study’s cooler Daily soil water balance calculations (1998) for grasses and wetter environment than in that of

AWCf = AWCi + Re + IW (vineyard only) – ETc and Caprile Baldocchi et al. [2004]). Estimated values (2007) for North for maximum available water capacity If AWCf > AWCm, then DP = AWCf − AWCm Irrigation water (IW) Coast wine (AWCm), effective rain fraction, irrigation If AWCf < 0, then GW = ETc − AWCi (trees only), otherwise AWCf = 0

If AWCf < AWCm × MAD, then apply IW (vines only) and recompute DP TABLE 1. Crop coefficients used in daily modeling of soil water processes in vineyards, oak trees and grasslands

Vineyards Oak trees Grasslands Root zone moisture (daily calculation) Period Kc Period Kc Period Kc 3/1–4/15 0.10 3/1–3/31 0.5 3/1–3/15 0.90 If AWCf < AWCm, then 4/16–4/30 0.20 4/1–10/1 0.6 3/16–4/30 0.95 AWC = AWC + (I + R − ET ) f i e c 5/1–5/15 0.25 10/2–11/25 0.5 5/1–5/15 0.25 If AWCf > AWCm, then AWCf = AWCm and 5/16–5/31 0.30 11/26–2/28 0.4 5/16–6/15* 0.10 DP = (I + Re − ETc ) − (AWCm − AWC i ) Deep 6/1–6/15 0.35 6/16*–10/13 0.00 percolation (Trees only) If AWCf < 0, then AWCf = 0 and 6/16–6/30 0.40 10/14–10/31 0.25 (DP) GW = (R − ET ) − AWC demand e c i 7/1–9/30 0.50 11/1–2/28 0.75 10/1–10/15 0.30 10/16–10/31 0.20 AWC = available water capacity MAD = maximum allowable depletion 11/1–11/15 0.15 ETc = K c × ETo (reference ET) Groundwater 11/16–11/30 0.05 i = initial, f = nal, m= maximum (GW) 12/1–2/28 0.01 Sources: Allen et al. 1998 (grasses and trees); Caprile 2007 (vineyards). * Variable date depending on available soil moisture. Fig. 3. Water balance model for soil in the root zone.

148 CALIFORNIA AGRICULTURE • VOLUME 66, NUMBER 4 and maximum allowable depletion were the maximum soil moisture–holding Tree canopy cover used, assuming loam soils and a water- capacity was never exceeded. Not surpris- Daily soil water balance calcula- table aquifer at 20 to 25 feet below the ingly, groundwater demand by the trees tions for the combined oak trees and ground surface. was greatest in late summer and fall. grasslands depended on the area of tree After running the model with the ini- Extraction rates averaged about 270 mil- canopy cover, and deep percolation and tial assumed parameter values for avail- limeters per year, similar to those mea- groundwater use rates fell. In grassland, able water capacity (AWC), maximum sured by Baldocchi et al. (2004). deep percolation was available to meet allowable depletion, effective rain fraction and groundwater depth, we evaluated the sensitivity of calculated groundwater TABLE 2. Major water balance factors for wine grape production in Sonoma County, 1999–2011 use values to these estimated parameters. This sensitivity analysis provided guid- Applied Deep Groundwater Year ETo* Rain* IWR† water Irrigations percolation use‡ ance on optimal parameter selection as ...... mm ...... no...... mm ...... well as a sense of the robustness of model 1999 1,169 598 400 500 4 384 116 predictions (that is, how much error in 2000 1,145 581 300 375 3 268 85 groundwater use calculations might be 2001 1,207 1,344 300 375 3 472 28 expected if assumed parameter values 2002 1,157 848 200 250 2 436 −61 were incorrect). 2003 1,109 890 200 250 2 471 −96 Estimating water needs 2004 1,169 676 200 250 2 331 44 In presenting the results, we set each 2005 1,100 961 200 250 2 528 −153 year beginning on March 1, when water 2006 1,074 820 200 250 2 441 −66 conditions in soil are most likely to be at 2007 1,166 636 200 250 2 313 62 or near capacity (table 2). Though net pos- 2008 1,185 396 300 375 3 259 241 sible annual water demand (ETo minus 2009 1,119 657 200 250 2 332 43 rain) ranged widely from 140 millimeters 2010 1,062 804 200 250 2 444 −69 in 2005 to 790 millimeters in 2008, the 2011 1,011 445 200 250 2 181 69 applied water needed for vineyard pro- Average 1,139 719 325 406 3.25 392 14.6 duction remained roughly the same from Standard 45 158 45 57 0.45 85 108.8 year to year, at about 300 millimeters. In deviation the model, applied water was typically * From accumulated daily CIMIS data, March 1 to Feb. 28 each year. given in three large irrigations, but the † IWR = irrigation water requirement, assuming 80% application efficiency. ‡ IWR minus deep percolation. depth of irrigation required was indepen- dent of irrigation frequency so that equivalent results would have been obtained if, TABLE 3. Major water balance factors for oak trees and grasslands, 1999–2011 for example, six 50-millimeter irrigations had been applied. Oak trees Grasslands The wide variability in net possible GW demand GW demand AWC depletion water demand by vineyards was reflected Year DP* GW demand period Net GW use rate date DP in the net groundwater use (applied wa- ...... mm ...... mm mm/day mm ter minus deep percolation) associated 1999 5.0 313.1 6/27–11/6 308.2 2.35 4/16 157.3 with vineyard production; groundwa- 2000 32.9 257.5 7/4–10/24 224.6 2.28 4/20 105.3 ter use ranged from a net recharge of 2001 98.9 389.9 6/4–10/28 291.0 2.65 4/22 253.7 153 millimeters in 2005 to a net use of 2002 149.3 295.7 7/2–11/5 146.3 2.33 4/16 283.0 241 millimeters in 2008. During the 13 2003 121.0 219.1 7/28–11/5 98.1 2.17 7/13 234.7 years, there was on average a very small 2004 50.1 308.9 6/23–10/15 258.7 2.69 4/21 194.3 groundwater demand of roughly 15 millimeters per year; this small demand was 2005 112.2 156.2 8/10–10/27 43.9 1.96 4/25 271.2 largely due to the deep percolation of 2006 175.8 235.8 7/14–11/1 60.0 2.12 6/1 222.5 winter rain. 2007 0.0 325.8 6/15–12/1 325.8 1.92 5/5 148.9 In oaks and grasses, deep percolation 2008 0.0 350.2 6/16–10/29 350.2 2.57 3/28 52.6 occurred only when maximum AWCm 2009 0.0 315.9 6/20–10/12 315.9 2.75 4/10 173.8 was exceeded (table 3). Grasses could only 2010 49.5 176.5 7/31–10/21 127.0 2.13 5/13 187.2 access soil moisture in the root zone, and 2011 129.6 194.9 7/26–11/4 65.3 1.85 4/20 52.3 plant water used (ETc) ceased when AWC Average 71.1 272.3 201.2 2.29 179.8 was zero; a condition noted in several Standard studies outlined above. For oak trees, rela- 62.8 71.4 114.2 0.30 75.9 deviation tively dry conditions from 2007 to 2009 * DP = deep percolation; GW = groundwater; AWC = available water capacity. resulted in no deep percolation, because

http://californiaagriculture.ucanr.edu • October–December 2012 149 Model prediction sensitivity groundwater demand by the trees. In TABLE 4. Density and canopy cover of oak all the years studied, deep percolation woodlands with equivalent groundwater usage to This modeling is subject to interpreta- decreased and net groundwater use vineyards, 1999–2010 tion because of the uncertainties associ- increased as tree cover increased. This ated with selecting various parameter linear relationship enabled us to deter- Year Cover Density* values. While precise values can be elu- mine the amount of cover that provided % trees/ha sive, their ranges are generally limited. the same net annual groundwater use as 1999 59 18.9 Aside from Kc values for daily water use that from the vineyard. On average, dur- 2000 58 18.4 by plants, the primary parameters affect- ing the 13-year period, oak tree cover of 2001 52 16.5 ing groundwater use are effective rainfall, about 50% resulted in groundwater use available water capacity and maximum 2002 52 16.4 equivalent to that for average vineyard allowable depletion. production (table 4). While vineyards re- 2003 42 13.3 Figure 5 shows the effects of different ceived irrigations during hot, dry months, 2004 53 16.8 effective daily rainfall values on annual oaks extracted needed water from deep 2005 38 12.0 groundwater recharge for the equivalent soil storage. vineyard and savanna shown in figures 2006 55 17.6 Completing a soil water balance us- 4 and 5, in which the assumed value was 2007 44 14.2 ing just the 13-year averaged daily ETo 0.6. Increasing effective rainfall values and rain data provided a similar area of 2008 73 23.2 implies that a greater proportion of daily trees (~ 45%) and allowed us to illustrate 2009 44 14.1 rainfall infiltrates and thereby more average conditions and determine the 2010 38 12.0 rapidly replenishes soil moisture while sensitivity of the calculated groundwater increasing possible deep percolation. In Average 51 16.1 use to variations in model parameters reality, effective rainfall values vary with (fig. 4). With the smallest available water Standard deviation 10 3.3 storm intensity and prior soil moisture capacity in the root zone, soil moisture in * Approximate, and assumes trees are 13 yards tall with a canopy conditions — high storm intensities and grasslands is depleted early in the season diameter of 20 yards and remaining area is grassland. near-saturated soil result in small effec- and gradually replenishes after plant se- tive rainfall values because rainfall runs nescence. The demand for additional wa- soil moisture. Similarly, vineyard ground- off rather than infiltrates. ter in vineyards (provided by irrigations) water use showed the three-step increase Rainfall interception by oak tree cano- begins after about 120 days, or early July, associated with groundwater withdraw- pies ranges from 20% to 30% of the total while deep groundwater demand in oaks als for irrigations. As demand leveled off, rain when the canopy is relatively dry, begins approximately 2 weeks later. winter rains replenished groundwater via possibly further reducing effective rain- In oak woodlands with 45% to 50% deep percolation and groundwater use fall. Similarly, increasingly steep ground tree canopy coverage, groundwater use rapidly declined. slopes increase runoff fractions of rainfall, was initially less than available soil mois- In both vineyards and oak woodlands reducing effective rainfall. Assuming ture, resulting in net recharge of about 10 (~ 45% canopy coverage), calculated losses to surface depressions of about millimeters per day for about 140 days. groundwater use was small at 5.5 mil- 10%, canopy interception and some soil At that point, cumulative groundwater limeters per year under these climate moisture, 0.7 may be a practical average demand steadily increased to 92 mil- conditions, although it is important to upper limit for the effective rainfall frac- limeters and leveled off. Groundwater note that vineyard irrigation results in tion, and, with the exception of very low- demand dropped as tree ETc decreased more groundwater use than oaks in intensity storms, 0.5 is the lower limit. The and early winter rains replenished the late summer. effect of uncertainty in this parameter

100 300 95 275 90 250 Vineyard 85 Savanna 80 225 75 200 y = 723.25x − 439.5 R2 = 1 70 175 65 150 60 125 55 y = 649.81x − 396.12 50 100 R2 = 0.9997 45 75 40 50 35 30 25 25 0 20 Vineyard −25 15

Oak trees groundwater Annual recharge (mm) −50 10

Fraction of soil water capacity lled (%) 5 Grassland −75 0 −100 0 30 60 90 120 150 180 210 240 270 300 330 360 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 Days after March 1 Eective rainfall fraction Fig. 4. Daily variation in root zone soil-water storage during an average Fig. 5. Effect of effective rainfall fraction used in modeling on net annual year for vineyards, oak trees (only) and grasslands. groundwater recharge for vineyard and equivalent oak woodland.

150 CALIFORNIA AGRICULTURE • VOLUME 66, NUMBER 4 is nearly identical for vineyards and oak water to become entirely depleted to 0% TABLE 5. Sensitivity of annual vineyard woodlands, suggesting that whatever the irrigations and groundwater use on available of capacity (dry) resulted in a greater net exact effective rainfall value, there would water capacity (AWC) groundwater recharge (19.5 millimeters be little or no change to estimated tree per year). The MAD factor therefore had a areas (as determined above). AWC range Irrigations* Groundwater use minor effect on estimated tree areas based Uncertainty in estimated available mm no. mm/year on annual groundwater use. water capacity stems from variation in 142–213 2 [19.5]† Overall, uncertainty in the model pa- soil types, soil layer thickness and the 106–141 3 5.5 rameter values results in an equivalent depth to relatively impermeable layers 85–105 4 30.5 variability in tree area that is similar in or groundwater. Setting available water range to that associated with hydrologic 71–84 5 55.5 capacity (storage volume) is the same variability (ETo and rain) (tables 2, 3 and * Maximum allowable depletion (MAD) = 30%. as setting a threshold, because changes † Brackets indicate net recharge. 4). For Sonoma County, conditions in in calculated groundwater use are not oak woodland with a tree canopy of 40% smooth but incremental (tables 5 and 6). to 60% are likely equivalent in terms of For typical established vineyards, capacity values for grapevines, trees and net annual groundwater use or recharge available water capacities likely range grasslands will have little effect on the to that of an established irrigated wine from 100 to 150 millimeters, within which equivalent tree area fraction as estimated grape vineyard. In both the vineyard and there is little effect on groundwater use from annual groundwater use. woodland, groundwater demand is great- (table 5). This suggests that the assumed Finally, the allowed maximum allow- est in late summer, though the rates from AWC value of 122 millimeters should be able depletion for irrigated agriculture de- groundwater pumping for vineyard irri- sufficient for comparisons. Available soil water capacity for oak trees is largely unknown, though expected to be large. Canopy coverage of 40% to 60% in oak woodlands results in average From the continuous water balance mod- annual groundwater use equivalent to irrigated mature wine grape eling, the assumed value of 250 millime- vineyards in Sonoma County. ters was exceeded in 9 of the 12 years, and in 2009 maximum soil storage was only 238 millimeters. Because oak savan- termines the managed soil water storage gation likely exceed those from extraction nas and woodlands are not irrigated, de- and frequency, and the possible depths by oak tree roots (fig. 6). creased available water capacity can offset of applied water. Wine grape vineyards Further study increased deep percolation and ground- are typically managed with deficit irriga- water demand. Thus, changing the water tion, allowing soil water to be substan- Overall, canopy coverage of 40% to available to trees from 200 to 250 millime- tially depleted to between 20% and 30% 60% in oak woodlands results in average ters results in relatively small values of capacity. In this range, net groundwater annual groundwater use equivalent to net groundwater use or recharge (table 6). recharge remained unchanged at 5.5 mature, irrigated, wine grape vineyards Overall, the range of likely available water millimeters per year. Allowing the soil in Sonoma County. However, though the

280 TABLE 6. Sensitivity of annual groundwater demand in oak woodland on soil available water 260 Vineyard capacity (AWC) 240 Savanna 220 Oak AWC Grass AWC Groundwater use 200 ...... mm ...... mm/year 180 160 300 95 19.3 140 290 90 16.5 120 280 85 13.8 100 270 80 11.0 80 60 260 75 8.3 40 250 70 5.5 20 Cumulative groundwater usage (mm) 240 65 2.8 0 230 60 0.02 -20 -40 220 55 [2.7]* -60 210 50 [5.5] 0 30 60 90 120 150 180 210 240 270 300 330 360 Days after March 1 200 45 [8.2] * Brackets indicate net recharge. Fig. 6. Daily variation in cumulative groundwater use during an average year for vineyards and equivalent oak woodlands.

http://californiaagriculture.ucanr.edu • October–December 2012 151 annual net use of groundwater was equivalent, the seasonal timing of this use is important to consider. Daily patterns indicate a relatively steady rate of groundwater extraction by oak tree roots beginning in late summer, when extraction increases RepositoryUC Regents/ANR and regional groundwater supplies are likely stressed. Water resource managers may need to consider this distinction between net use and timing. Our analysis also provides some in- sight into the concept of integrated oak woodland and vineyard landscapes, in which the oak tree/savanna grasslands are replaced with vines, leaving a patch- work of trees. These landscapes would provide greater ecological diversity than occurs in vineyards alone. However, oak trees in combination with irrigated vineyards could potentially result in greater net groundwater use than either landscape individually.

M. Grismer is Professor of Hydrology and Biologi- cal and Agricultural Engineering, UC Davis; and C. Asato is Undergraduate Student, Depart- Landscapes that integrate oak woodlands and vineyards (shown, in Glen Ellen) provide more ment of Biological and Agricultural Engineering, ecological diversity, but water resource managers must consider seasonal variations in water usage UC Davis. by both land uses, together and individually.

References Fisher JB, Baldocchi DD, Misson L, et al. 2007. What the Merenlender AM. 2000. Mapping vineyard expansion towers don’t see at night: Nocturnal sap flow in trees and provides information on agriculture and the environ- Allen RG, Pereira LS, Raes D, Smith M. 1998. Crop Evapo- shrubs at two AmeriFlux sites in California. Tree Physiol ment. Calif Agr 54(3):7–11. transpiration – Guidelines for Computing Crop Water Re- 27:597–610. quirements. FAO Irrigation and Drainage Paper 56. Food Miller GR, Chen X, Rubin Y. 2010. Groundwater uptake by and Agriculture Organization, United Nations. Rome, Italy. Grismer ME. 2012. Estimating agricultural deep drainage woody vegetation in a semiarid oak savanna. Water Re- lag times to groundwater: Application to Antelope Val- sour Res 46, W10503. doi:10.1029/2009WR008902. Baldocchi DD, Ma S, Rambal S, et al. 2010. On the differ- ley, California, USA. Hydrol Process (March). doi: 10.1002/ Paço TA, David TS, Henriques MO. 2009. Evapotranspira- ential advantages of evergreenness and deciduousness hyp.9249. in Mediterranean oak woodlands: A flux perspective. Ecol tion from a Mediterranean evergreen oak savanna: The Appl 20(6):1583–97. Grismer ME, Bachman S, Powers T. 2000. A comparison role of trees and pasture. J Hydrol 369(1-2):98–106. of groundwater recharge estimation methods in a semi- Baldocchi DD, Xu L. 2007. What limits evaporation from Parrish W. 2011. Anderson Valley, tentacle of the wine- arid, coastal avocado/citrus orchard, Ventura County, CA. grape octopus. Friends of the Eel River, Feb 4. http:// Mediterranean oak woodlands – The supply of moisture Hydrol Process 14(14):2527–43. in the soil, physiological control by plants or the demand eelriver.org/2011/02/will-parrish-anderson-valley-tenta- by the atmosphere? Adv Water Resour 30:2113–22. Gurdak JJ, Hanson RT, McMahon PB, et al. 2007. Climate cle-of-the-winegrape-octopus. variability controls on unsaturated water and chemi- Baldocchi DD, Xu L, Kiang N. 2004. How plant functional- Pritchard TL. 2010. Irrigation management of winegrapes cal movement, High Plains aquifer, USA. Vadose Zone J with a limited water supply. UC Cooperative Extension. type, weather, seasonal drought, and soil physical 6(3):533–47. properties alter water and energy fluxes of an oak-grass http://ucmanagedrought.ucdavis.edu/PDF/WINE- savanna and an annual grassland. Agr Forest Meteorol Heaton E, Merenlender AM. 2000. Modeling vineyard GRAPE%20DROUGHT%20LONG-1.pdf. 123:13–39. expansion, potential habitat fragmentation. Calif Agr Scanlon BR, Reedy RC, Stonestrom DA. 2005. Impact of 54(3):12–7. Burgy RH, Adams TE. 1977. Wildlands and watershed land-use and land cover change on groundwater re- management. Calif Agr 31(5):9–10. Hinckley E-LS, Matson PA. 2011. Transformations, charge and quality in the southwestern U.S. Glob Change transport and potential unintended consequences Biol 11:1–17. Caprile J. 2007. Improving Winegrape Quality with Irriga- of high sulfur inputs to Napa Valley vineyards. PNAS Sophocleous M. 2005. Groundwater recharge and tion Management. UC Cooperative Extension. http:// 108(34):14005–10. ucanr.org/sites/cccoopext/files/80970.pdf. sustainability in the High Plains aquifer in Kansas, USA. Jackson LE, Strauss RB, Firestone MK, Bartolome JW. 1990. Hydrogeol J 13:351–65. Chen Q, Baldocchi DD, Gong P, Dawson T. 2008. Model- Influence of tree canopies on grassland productivity and ing radiation and photosynthesis of a heterogeneous Teuling AJ, Seneviratne SI, Williams C, et al. 2006. nitrogen dynamics in deciduous oak savanna. Ag Ecosys Observed time scales of evapotranspiration re- savanna woodland landscape with a hierarchy of model Environ 32:89-105. complexities. Agr Forest Meteorol 148:1005–20. sponse to soil moisture. Geophys Res Lett 33, L23403. Lohse K, Newburn D, Opperman J, Merenlender AM. doi:10.1029/2006GL028178. Chen Q, Rubin Y, Ma S, Baldocchi DD. 2008. Observations 2008. Forecasting the relative impacts of land use on fine and stochastic modeling of soil moisture control on Thorburn PJ, Cowie BA, Lawrence PA. 1991. Effect of land sediment in anadromous fish habitat to guide develop- development on groundwater recharge determined evapotranspiration in a Californian oak savanna. Water ment and conservation programs. Ecol Appl 18:467–82. Resour Res 44:W08409. doi:10.1029/2007WR006646. from non-steady chloride profiles. J Hydrol 124:43–58. McMahon P, Dennehy KF, Bruce BW, et al. 2006. Stor- Community Foundation. 2009. Biodiversity Action Plan age and transit time of chemicals in thick unsaturated for Sonoma County. www.sonomacf.org. zones under rangeland and irrigated cropland, High David TS, Ferreira MI, Cohen S, et al. 2004. Constraints Plains, United States. Water Resour Res 42, W03413. on transpiration from an evergreen oak tree in southern doi:10.1029/2005WR004417. Portugal. Agr Forest Meteorol 122:193–205.

152 CALIFORNIA AGRICULTURE • VOLUME 66, NUMBER 4