Research Article Soil suitability index identifies potential areas for groundwater banking on agricultural lands by Toby A. O’Geen, Matthew Saal, Helen Dahlke, David Doll, Rachel Elkins, Allan Fulton, Graham Fogg, Thomas Harter, Jan W. Hopmans, Chuck Ingels, Franz Niederholzer, Samuel Sandoval Solis, Paul Verdegaal and Mike Walkinshaw

Groundwater pumping chronically exceeds natural recharge in many agricultural Two recent trends in California have regions in California. A common method of recharging groundwater — when surface tended to increase the rate of groundwa- ter overdraft in agricultural regions. water is available — is to deliberately flood an open area, allowing water to percolate First, over the past two decades, ir- into an aquifer. However, open land suitable for this type of recharge is scarce. Flooding rigation technologies have significantly agricultural land during fallow or dormant periods has the potential to increase improved water use efficiencies (Canessa groundwater recharge substantially, but this approach has not been well studied. Using et al. 2011; Howell 2001; Orang et al. 2008; Tindula et al. 2013; Ward and Pulido- data on soils, topography and crop type, we developed a spatially explicit index of the Velazquez 2008). Where surface water suitability for groundwater recharge of land in all agricultural regions in California. We is used for irrigation, a consequence of identified 3.6 million acres of agricultural land statewide as having Excellent or Good applying less water is that groundwa- potential for groundwater recharge. The index provides preliminary guidance about ter recharge is diminished because of a reduction in deep percolation of excess the locations where groundwater recharge on agricultural land is likely to be feasible. water. A variety of institutional, infrastructure and other issues must also be addressed before Second, expanding worldwide markets this practice can be implemented widely. have driven significant expansions of nut and wine grape acreage. For example, the almond acreage in California has dou- alifornia is experiencing its third the water applied to crops in California bled, to roughly 1 million acres, since 1994 major drought since the 1970s, (roughly 34 million acre-feet per year) is (NASS 2014). Much of this expansion has Cand projections suggest that such supplied by groundwater sources, but in occurred in the San Joaquin Valley where episodes will become longer and more times of drought the proportion can in- rates of rainfall and natural groundwater frequent in the second half of the 21st crease to as much as 60% (Megdal 2009). recharge are low. This shift in cropping century (Barnett et al. 2008; Cayan et al. As a result, groundwater levels fall during systems to high value perennial crops re- 2010). Droughts place more demand on droughts (Ruud et al. 2004). If groundwa- duces the flexibility of agricultural water groundwater resources to buffer surface ter is not replenished during wet years, demand because the economic costs of not water shortfalls. Ordinarily, about 30% of long-term overdraft occurs. From 2005 irrigating are severe. Inflexible demand through 2010, average annual overdraft has made agriculture even more reliant in the Central Valley was estimated to be on groundwater during dry periods when Online: http://californiaagriculture.ucanr.edu/ between 1.1 and 2.6 million acre-feet (De- surface water resources are curtailed. landingpage.cfm?article=ca.v069n02p75&fulltext=yes partment of Water Resources 2015). doi: 10.3733/ca.v069n02p75

During fallow or dormant periods, agricultural lands have the potential to serve as percolation basins for groundwater recharge.

http://californiaagriculture.ucanr.edu • APRIL–JUNE 2015 75 David Doll David Groundwater recharge flooding has not occurred in recent years Five factors that Natural groundwater recharge is the (Bachand et al. 2011). As the climate determine the feasibility predominant source of groundwater re- warms, flooding may become more fre- plenishment in almost all basins. It is typ- quent and extreme as a result of episodic of groundwater recharge ically unmanaged and can be slow. Water snowmelt events driven by warm winter on agricultural land percolates into aquifers from a variety of rains. Recycled water (highly treated surface water sources including precipita- wastewater) is another potential source. Deep percolation: Soils must tion, streams, rivers, lakes, surface water There are a variety of institutional and 1. be readily able to transmit water conveyance facilities — such as unlined other barriers to widespread agricultural beyond the root zone (1.5 m, 5 ft). canals — and applied irrigation water. groundwater banking in California. Root zone residence time: The Natural recharge also may occur from Water rights for operation of aquifers as 2. duration of saturated/near satu- horizontal subsurface inflow from one reservoirs are challenging to navigate; rated conditions after water ap- part of a groundwater basin to another. water conveyance infrastructure has plication must be acceptable for Natural recharge requires no dedicated limited capacity; regional planning to the crops grown on lands under infrastructure or land. capture river flood waters may be difficult consideration for groundwater Groundwater banking is a manage- to organize; fields with high percolation banking throughout the entire crop ment strategy that stores surface water in rates at the surface may be underlain by root zone. aquifers for future withdrawal. It expands low-percolation layers that slow or block managed water storage capacity, which the recharge of deeper aquifers; it can Topography: Slopes that nega- in California consists mainly of surface be difficult to assess how much capacity tively influence the even distribu- 3. water reservoirs. Groundwater banking is a given aquifer has to store banked tion of water will be more difficult achieved through the intentional applica- groundwater; certain crops and certain to manage. tion of surface water. During hydrologic stages of crop growth do not tolerate Chemical limitations: High soil cycles when surface water is abundant, flooded conditions; and the quality 4. salinity may result in saline leachate extra surface water can be “deposited” of water recharged to an aquifer via (poor water quality) that must be in a groundwater bank by application to agricultural land may be degraded due to avoided to protect groundwater constructed percolation basins, through excessive leaching of contaminants from quality. injection wells, or through joint manage- soil such as pesticides and nitrates. Soil surface condition: Certain ment of rivers and groundwater to ef- To date, few well-documented trials 5. soils may be susceptible to fect riverbed infiltration into underlying of groundwater banking have been con- compaction and erosion if large aquifers. ducted on agricultural land. Since nearly volumes of water are applied. A key limitation to groundwater re- all agricultural land is privately owned Surface horizons with high sodium charge is the lack of suitable percolation and operated, participation in ground- are prone to crusting that may basins available for deliberate flooding. water banking programs depends on contribute to decreased surface In this paper, we consider a new strategy cooperation from the landowner or land infiltration rates. for groundwater banking that involves manager. Therefore, a clear understand- applying water to agricultural lands out- ing of the risks and best practices associ- side of the usual irrigation season for the ated with this practice is paramount. specific purpose of recharging a ground- In this study, we take a first step to- water basin. Given the millions of acres ward better understanding opportunities of irrigated farmland in California, using to recharge groundwater using agricul- agricultural lands as percolation basins tural landscapes in California by identify- has the potential to increase groundwater ing and mapping the soil and topographic recharge during wet periods when sur- conditions most conducive to groundwa- face water is available. ter recharge. In California, one potential source of Groundwater banking index water for recharge on agricultural land is river floodwaters, because surface water This study developed a Soil Agricul­ rights may be easily re-negotiated (or may tural Groundwater Banking Index not apply) for the excess water. This flood- (SAGBI) that provides a composite evalu- water approach has the dual benefit of ation of soil suitability to accommodate withdrawing large amounts of water from groundwater recharge while maintaining a river that is at or near flood stage and re- healthy soils, crops and a clean ground- ducing downstream flood risks (Bachand water supply. The SAGBI is based on five et al. 2011). The frequency and intensity of major factors that are critical to successful Wine grapes are especially tolerant of river flooding is difficult to forecast. For agricultural groundwater banking: deep flooding. This field flooded following heavy instance, flood flows on the Kings River percolation, root zone residence time, rains in 2006; while the crop was reduced that year, the vines survived and are thriving today. from 1975 to 2006 had an average reoc- topography, chemical limitations and soil Paul Verdegaal Paul currence interval of 2 to 3 years, though surface condition (see sidebar, this page).

76 CALIFORNIA AGRICULTURE • VOLUME 69, NUMBER 2 We modeled each of the five factors was applied to a soil profile’s lowest Ksat outcome, we included in the model a satu- using U.S. Department of Agriculture to score the likelihood of deep percolation ration residence time factor for soils. The Natural Resources Conservation Service (fig. 1). root zone residence time factor estimates (USDA-NRCS) digital soil survey data. Root zone residence time factor. Pro­ the likelihood of maintaining good drain- The suitability of each factor was ex- longed duration of saturated or nearly age within the root zone shortly after wa- pressed through a scoring system based saturated conditions in the root zone can ter is applied. This rating is based on the on a combination of fuzzy logic functions cause damage to perennial crops, and in harmonic mean of the Ksat of all horizons and crisp ratings (see sidebar, this page). some cases, crop loss (table 1). About one- in the soil profile, soil drainage class and Deep percolation factor. Successful third of California’s irrigated cropland is shrink-swell properties. The harmonic groundwater banking depends on a high occupied by perennial crops and vines. mean is typically used when reporting rate of water transmission through the Table 1 provides estimates of tolerance the average value for rates and tends to soil profile and into the aquifer below. A to saturation for some common tree and be lower than a standard average. Poorly high percolation rate is especially impor- vine crops before and after budbreak drained soils and soils with high shrink- tant if floodwaters are the water source compiled through a survey of UC ANR swell received the lowest scores with used because floodwaters are available Cooperative Extension (UCCE) commod- a crisp rating of 1. All other soils were for diversion over a narrow time frame. ity experts. Annual crops were not in- The deep percolation factor is derived cluded in the survey because we assumed from the saturated hydraulic conductivity that these fields generally would be fallow Fuzzy logic and crisp scores (Ksat) of the limiting layer (the soil horizon during times of excess surface water avail- with the lowest Ksat). Saturated hydraulic ability. In general, crops become prone uzzy logic is a method by which conductivity is a measure of soil perme- to damage after budbreak and there is a Fmembership to a class or condition ability when soil is saturated. Many soils range in tolerance among crops and root- can be partial (maybe) rather than dis- in California have horizons (layers) with stocks (table 1). For example, wine grapes crete (true or false; or A or B). Thus, fuzzy exceptionally low Ksat values that severely and pears may be able to withstand more logic allows reasoning to be approximate limit downward percolation, such as than two weeks of saturated conditions rather than fixed and exact. Variables cemented layers (duripan, petrocalcic), before budbreak, while avocados and cit- are evaluated via fuzzy logic scores that claypans (abrupt increases in clay content) rus have no tolerance. range between 1 and 100, reflecting the and strongly contrasting particle size dis- Our survey identified that many crops degree of vagueness of a membership tributions. Soils with these horizons were are unable to withstand long periods of being completely false (1) or completely given crisp scores of 1. For other soils, a saturated conditions in the root zone. true (100). Fuzzy logic is appropriate for “more is better” fuzzy logic rating curve To account for this potential adverse this model analysis because in agricul- tural landscapes, the above five factors are relative as opposed to absolute, TABLE 1. Survey results of tree crop vulnerability to saturated conditions which poses challenges in quantifying Tolerance to saturation Tolerance to saturation Recommended N them using the raw data. Crop Rootstock before budbreak after budbreak fertilizer rate We used fuzzy logic statements such lbs N/ac/yr as (1) “more is better” where the score Almonds Peach; peach x 1 1 250 increases with higher factor values; (2) almond hybrid “less is better” where the score increases Almonds Plum; peach x plum 2–3 1 250 as factor values decrease; and, (3) “opti- hybrid mum range” where the score is highest Avocados — 0 0 150 across a certain range of factor values Cherries — 1 0 60 and decreases above and below that Citrus — 0 0 100 range. Using the suitability of root zone Wine grapes — 4 2 15–30 residence time as an example, the fuzzy Olives — ? ? <100 logic statement “less is better” enables Pears P. betulaefolia 4 4 100–150 the suitability of that factor to vary be- tween 1 and 100 (from unsuitable to Pears P. communis 4 3 100–150 optimally suitable) rather than having to Pears Cydonia oblonga 3–4 2–3 100–150 choose between absolutes, e.g., suitable Pistachios — ? ? 200 (true) or not suitable (false). Crisp ratings Plums/prunes Peach 1 1 150 are defined scores that apply to a well- Plums/prunes Plum; peach x plum 2–3 1 150 understood system, and hence do not hybrid require fuzzy scoring. For example, slope Pomegranate — ? ? 100 classes as reported in soil surveys reflect Walnuts — 2–3 1 200 limitations of common practices such as The following scores were used to estimate vulnerability: 0 - No tolerance for standing water; 1 - tolerant of standing water up to 48 hours; irrigation and cultivation practices and 2 - tolerant of standing water up to 1 week; 3 - tolerant of standing water up to 2 weeks; 4 - tolerant of standing water > 2 weeks; ? - tolerance unknown. are scored in our model with crisp ratings. Tolerance to saturated conditions is based on expert opinion and has not been supported by controlled experimentation.

http://californiaagriculture.ucanr.edu • APRIL–JUNE 2015 77 Soil Agricultural Modied rating to reect Groundwater deep tillage and removal Banking Index of restrictive horizons

Root Zone Topographic Deep Percolation Chemical Limitations Surface Condition Residence Time Limitations weight = 27.5% weight = 20% weight = 5% weight = 27.5% weight = 20%

Harmonic mean Erodibility factor (Kw)

Lowest Ksat in soil of Ksat (all horizons) Depth weighted and sodium prole presence of drainage class Slope class average of electrical adsorption ratio restrictive horizons and high conductivity (geometric mean of shrink‐swell soils scores for two values)

Fuzzy logic “more Crisp ratings: Fuzzy logic “less is Fuzzy logic “more is Fuzzy logic “less is is better” K ; Optimal = 100 better” Kw and SAR; sat better” K ; better” EC; Crisp ratings of sat Good = 75 Crisp ratings of Crisp ratings of 1 for Crisp ratings of 1 for restrictive horizons Moderate = 50 1 for SAR ≥ 13; shrink‐swell soils and 1 for EC ≥ 16 dsm and and 100 for Challenging = 25 1 for Kw > 0.45 poorly drained soils 100 for EC ≤ 4 dSm Ksat> 0.42μsmm Extremely challenging = 1 100 for Kw ≤ 0.2

Fig. 1. Schematic of the Soil Agricultural Groundwater Banking Index. scored using a fuzzy logic rating curve of where sediments are derived from marine infiltration (Le Bissonais 1996). We used “more is better” for Ksat (fig. 1). sediments in the Coast Range. The chemi- two soil properties to diagnose surface Topographic limitations factor. cal limitations factor was quantified us- condition, the soil erosion factor and Agricultural groundwater banking will ing the electrical conductivity (EC) of the the sodium adsorption ratio (SAR). The likely be implemented by spreading water soil, which is a measure of soil salinity. A surface condition factor was calculated across fields. Level topography is better fuzzy logic rating curve “optimum and by the geometric mean of fuzzy logic suited for holding water on the landscape, less is better” was used to score chemical scores from these two properties. A thereby allowing for infiltration across limitations. The “less is better” statement geometric mean is a way to identify the large areas, reducing ponding and mini- implies that soils with low salinity score average value of two or more properties mizing erosion by runoff. Ranges in slope high and soils with high salinity values that have different ranges in value. SAR percent were used to categorize soils into score low. Soil profiles with EC < 4 dS/m values greater than 13 indicate that the four slope classes: Optimal (slope classes were considered optimal (score of 100). soil is prone to crusting. A “less is better” 0%–1% and 0%–3%), good (slope classes Beyond this threshold, scores decreased fuzzy logic curve was used to evaluate 0%–5% and 2%–5%), moderate (slope with increasing EC. Soils with EC values SAR, where values greater than 13 were classes 0%–8% and 3%–8%), challenging above 16 dSm−1 received a score of 1 (fig. assigned a crisp rating of 1, and values of (slope classes 5%–8%, 3%–10% and 5%– 1). A variety of other contaminants such 0 were assigned an optimal rating of 100. 15%), and extremely challenging (slope as pesticides and nitrate are also present Soil surface horizon Kw, the soil erod- classes 10%–30% and 15%–45%) (fig. 1). in agricultural soils. However, because ibility factor of the Revised Universal Topographic limitations were scored us- this type of contamination is dependent Loss Equation, was used to estimate the ing crisp ratings that generally reflect the on management history, the USDA-NRCS potential soil susceptibility to erosion, USDA-NRCS slope classes because these soil survey does not document it and we disaggregation and physical crust forma- classes were designed in consideration of were unable to evaluate it. tion (USDA-NRCS 2014). A fuzzy logic limitations for standard agricultural man- Surface condition factor. Groundwater rating curve, “optimum and less is better,” agement practices (Soil Survey Division banking by flood spreading can subject was used for scoring the surface condition Staff 1993). the soil surface to changes in its physical factor. Kw values < 0.20 were considered Chemical limitations factor. Salinity is a condition. Depending on the quality of ideal (score = 100); beyond this threshold, threat to the sustainability of agriculture the water and depth of water, standing factor scores decreased with increasing and groundwater in California, especially water can lead to the destruction of ag- Kw values. along the west side of the Central Valley gregates, the formation of physical soil SAGBI calculation. Each of the five (Kourakos et al. 2012; Schoups et al. 2005), crusts and compaction, all of which limit model factor scores was assigned a weight

78 CALIFORNIA AGRICULTURE • VOLUME 69, NUMBER 2 based on its significance to groundwater have been altered to the point that they in the weighted average. Data below the banking (fig. 1). The SAGBI score was cal- are now considered endangered in the restrictive horizon was included in the culated by the weighted geometric mean Central Valley (Amundson et al. 2003). depth-weighted average if populated in of the scaled factors. The factors were As a result, soil surveys of much of the the database. The depth-weighted average weighted as follows: Deep percolation region — many of which were conducted of Ksat was used in place of the harmonic (27.5%), root zone residence time (27.5%), decades ago — are outdated with respect mean to estimate hydraulic conductivity topographic limitations (20%), chemical to alterations by deep tillage. To address for the root zone residence time factor. limitations (20%) and surface condition this problem, we created an updated soil Map unit aggregation. SAGBI scores (5%). Factor weights were applied based disturbance map using geospatial analy- were calculated for most agricultural on expert opinion. Factors with greater sis. A map of orchard and vineyard crops soils populated in the USDA-NRCS Soil relevance to groundwater recharge were was created using California Department Survey Geographic Database (SSURGO). weighted more heavily, while factors of Water Resources land use maps (is- Soil survey delineations represent map that may be modified by management, sued between 2001 and 2011) and aerial units, which often contain more than one such as surface condition, were given a imagery from the National Agricultural soil type. The map units range in size lower weight. SAGBI scores were catego- Imagery Program (NAIP) and Google from 5 acres to roughly 500 acres. To cre- rized into six groups: Excellent, Good, Earth (2012 to 2014). This file was over- ate a regional map, each map unit was Moderately Good, Moderately Poor, lain in a geographic information system scored with the SAGBI value using the Poor and Very Poor based on the natural with a map of soils with water-restrictive soil component that comprised the largest groupings of the dataset. horizons. We assumed that all tree and percentage of the map unit area. If there Soils modified by deep tillage. In recent vine cropland with restrictive soil lay- was a tie (i.e., one map unit containing decades, high value orchard and vineyard ers (based on soil survey data) has been two components of equal area), the most crops have expanded onto soil landscapes modified by deep tillage, generating an limiting (lowest) SAGBI score was chosen that contain restrictive horizons. A stan- updated map of modified soils. for the map unit. dard practice for tree and vine establish- To reflect the mixing of soil horizons Spatial patterns of SAGBI ment on these soils is deep tillage up to a in the calculation of the deep percolation depth of 6 feet to destroy restrictive layers factor, the depth-weighted average of Our study area included over 17.5 mil- that impede root penetration. This prac- Ksat for the entire soil profile was used in lion acres of agricultural land (irrigated tice increases deep percolation rates and place of the lowest Ksat for each profile. and non-irrigated) as identified by the drainage conditions compared to natu- We reduced the deep percolation factor state Farmland Mapping and Monitoring rally occurring soils. Soils with root- and rating for soils with claypans by 20% to Program. Based on our initial modeling, water-restrictive horizons in California reflect the risk that modified claypans which did not initially consider the ef- will reform, which fects of deep tillage, soils in the Excellent, can occur in as Good and Moderately Good suitability Yreka Soil Agricultural little as four years groups comprised over 5 million acres, or Groundwater Banking in soils with weak 28% of the study area (fig. 2 and table 2). Index Redding structure (White et These highly rated soils were most abun- Excellent al. 1981). Cemented dant on broad alluvial fans on the east Good layers (not includ- side of the Central Valley stemming from Chico Moderately good ing bedrock) were the Mokelumne, Stanislaus, Merced, Kern Moderately poor assumed to have and Kings rivers (fig. 2). Excellent, Good Sacramento been removed by and Moderately Good ratings are also Santa Rosa Poor deep tillage and found throughout much of Napa, Salinas Very poor Stockton were not included and Santa Maria valleys and in patches San Francisco Modesto Merced TABLE 2. Summary of the areal extent of Soil Agricultural Groundwater Salinas Fresno Banking Index groups generated from soil survey data Visalia SSURGO modified by SAGBI group Original SSURGO data deep tillage Bakers eld acres %* acres %* Santa Maria Excellent 1,477,191 8 1,557,035 9 Good 1,747,712 10 2,020,921 11 Oxnard Los Angeles Moderately Good 1,786,972 10 1,984,414 11 Indio ± Moderately Poor 1,343,250 8 1,364,066 8 Poor 4,866,942 28 4,586,645 26 0 50 100 San Diego Very Poor 6,375,277 36 6,084,142 35 Miles Lab Soil Resource California Total† 17,597,345 17,597,222 Fig. 2. Spatial extent of Soil Agricultural Groundwater Banking Index * Percent of total study area. suitability groups when not accounting for modifications by deep tillage. † Modified SAGBI ratings had 123 fewer acres because two soils lacked sufficient data to adjust.

http://californiaagriculture.ucanr.edu • APRIL–JUNE 2015 79 along the Russian River in Mendocino table 2). Some areas of Good and Excellent potential for applied water to flow, by sub- and Sonoma counties and the northern ratings were found on sandy floodplains surface transport, into rivers and streams. parts of the Coachella Valley. The best of rivers and streams, especially along the Thus, these systems should not be priori- soils — the Excellent and Good groups Sacramento and Feather rivers. tized for groundwater banking unless it — occupied about 3.2 million acres, rep- Floodplains may not be ideal locations is known that the surface water bodies resenting 18% of the study area (fig. 2 and for groundwater banking because of the are losing streams — that is, surface water bodies that discharge to groundwater. Most major streams that traverse the San A. Yreka Deep percolation factor Joaquin Valley, for instance, are known to be losing streams. 80–100 Extensive Moderately Good areas 60–80 Redding were mapped on the western margins 40–60 of the San Joaquin and Sacramento val- Chico 20–40 leys where soils tend to be finer textured 0–20 and sometimes salt-affected (saline). Moderately Good groups were also Sacramento Santa Rosa mapped in basin alluvium where low Stockton energy flood events have deposited fine San Francisco sediments. Moderately Good groups oc- Modesto Merced cupied 1,786,972 acres or 10% of the study area. These areas may require careful con- Salinas Fresno Visalia sideration for groundwater banking. The majority of land in the study area (72% or ~12.6 million acres) was classified Bakers eld as Moderately Poor, Poor or Very Poor Santa Maria SAGBI groups (fig. 2 and table 2). Soils with low SAGBI scores were abundant Oxnard Los Angeles throughout the basin margins of the ± Indio Sacramento and San Joaquin valleys as well as across land interstratified 0 50 100 San Diego between recent alluvial fan deposits of Miles

B. C. Yreka Root zone residence Yreka Chemical time factor limitations factor

Redding 80–100 Redding 80–100 60–80 60–80

Chico 40–60 Chico 40–60 20–40 20–40 0–20 0–20 Sacramento Sacramento Santa Rosa Santa Rosa

Stockton Stockton San Francisco San Francisco Modesto Modesto Merced Merced Salinas Fresno Salinas Fresno Visalia Visalia

Bakers eld Bakers eld Santa Maria Santa Maria

Oxnard Oxnard Los Angeles Los Angeles ± Indio ± Indio

0 50 100 San Diego 0 50 100 San Diego Miles Miles Lab Soil Resource California

Fig. 3. Spatial extent of Soil Agricultural Groundwater Banking Index factors (A) deep percolation, (B) root zone residence time and (C) chemical limitations.

80 CALIFORNIA AGRICULTURE • VOLUME 69, NUMBER 2 the Mokelumne, Tuolumne, Stanislaus, profile and poor drainage. Poorly and of the salt-rich nature of the marine sedi- San Joaquin, Kings and Kern rivers. very poorly drained soils have properties ments within the Coast Range and poor Very Poor and Poor ratings are also or conditions that promote saturation in drainage conditions on the west side that found on the northern portions of the the upper parts of the soil profile, such as prevent salts from leaching out of soil. Salinas and Santa Maria Valleys and high clay content, water restrictive lay- There are other chemical limitations of throughout most agricultural regions in ers or regionally shallow water tables. soils we could not evaluate that would in- Sonoma County and southern parts of The least suitable soils in this factor were fluence groundwater banking, most nota- the Coachella Valley. those with poor drainage or high shrink- bly the concentration of residual nitrate in Of the SAGBI components, the deep swell properties. Low scores for root zone soil. Crops with high nitrogen demand or percolation factor was limiting over the residence factor were widespread along high residual nitrate in soil in the fall after greatest area (fig. 3A). These limiting con- the west side of the San Joaquin and harvest may not be suitable for ground- ditions arise from different characteristics Sacramento valleys in soils weathering water banking (table 1). of soils. For example, old, highly devel- from Coast Range alluvium (fig. 3B). Poor The surface condition factor was oped soils found along the margins of the drainage and fine textured soils were also weighted lowest among all other factors Central Valley contain water-restrictive found in the basin alluvium towards the because compaction from standing water horizons (either cemented hardpans or center of valleys. Low scores for this fac- can be fixed with tillage and amend- claypans). The center of the valley con- tor were also found on alluvial fans that ments. Low surface condition factor rat- tains young soils with fine (clay-rich) have drainages confined to the metamor- ings were abundant in soils with loamy texture throughout the soil profile. Both phic portions of the Sierra Foothills such surface textures or high SAR and were of these soil landscapes contain at least as the Calaveras River fan, which tend to located throughout the study area but one soil horizon with low permeabil- have fine textured sediments compared tended to be concentrated on the west ity. In contrast, high deep percolation to fans sourced in granitic terrain in the side of the Central Valley where sodium- scores were found on coarse-textured high Sierra Nevada. affected soils are common (fig. 4A). soils derived from recent (e.g., < 80,000 Chemical limitations had a localized Soil landscapes with low slope factor years) alluvial fans with drainages influence on the distribution of SAGBI ratings were limited to the margins of the sourced in granitic terrain of the Sierra ratings. Most of the salt-affected soils are valleys (fig. 4B). This sloping terrain is a Nevada and the Salinian block within the present along the west side of the San result of uplift by the Coast Range and Coast Range. Joaquin Valley and to a lesser extent along Sierra Nevada over geologic time scales, Areas limited by the root zone resi- the western margin of the Sacramento which increased slope gradients and ac- dence factor typically had soils with uni- Valley (fig. 3C). The distribution of salt- celerated erosion. The natural erosion of formly fine texture throughout the soil affected soils results from a combination the valley margins has created gentle to

A. Yreka B. Yreka Soil surface Topography factor condition factor 80–100 Redding 80–100 Redding 60–80 60–80 40–60 Chico 40–60 Chico 20–40 20–40 0–20 0–20 Sacramento Sacramento Santa Rosa Santa Rosa

Stockton Stockton San Francisco San Francisco Modesto Modesto Merced Merced Salinas Fresno Salinas Fresno Visalia Visalia

Bakers eld Bakers eld Santa Maria Santa Maria

Oxnard Oxnard Los Angeles Los Angeles ± Indio Indio

0 50 100 San Diego San Diego Miles Lab Soil Resource California

Fig. 4. Spatial extent of Soil Agricultural Groundwater Banking Index factors (A) surface condition and (B) topographic limitations.

http://californiaagriculture.ucanr.edu • APRIL–JUNE 2015 81 steeply undulating landforms (see photo, Moderately Good SAGBI suitability Implications below). groups increased from 28% to 31% of the There are approximately 5.6 million land area, adding 550,494 acres of suit- acres of land with soils in Excellent, Good Modified SAGBI scores to reflect able agricultural land for groundwater and Moderately Good SAGBI suitability deep tillage banking (table 2). A majority of improved groups, a significant amount of agricul- When deep tillage on orchard and SAGBI scores were located in the eastern tural land capable of accommodating vineyard croplands was incorporated San Joaquin Valley and Tulare Basin, deep percolation with low risk of crop into the model, the Excellent, Good and where soils with restrictive horizons are damage or contamination of groundwater common (fig. 5). It is possible that over by salts. Most suitable soils for agricul- time, more suitable land for groundwa- tural groundwater banking occur on or ter banking will become available as near alluvial fans created by rivers drain- marginal soils continue to be developed ing the Sierra Nevada. Perhaps not coinci- Yreka and modified for agricultural purposes dentally, these are also the areas that have (Charbonneau and Kondolf 1993). California’s most successful groundwater The final SABGI that accounts for deep banking programs (Water Association of Redding tillage represents the best estimate of Kern County 2014). soil suitability for groundwater banking. Our preliminary survey of UCCE pe- Over 12 soil survey areas are classified rennial crop experts suggests that pears, Chico as out-of-date in agricultural regions of wine grapes and some rootstocks of vari- California (USDA-NRCS 2014) and do ous Prunus species (i.e., almond, peaches not accurately document the extent of and plums) are best suited for ground- soil modification by deep tillage. These water banking if planted on suitable soils Sacramento Santa Rosa modified SAGBI ratings provide and managed appropriately, especially an updated assessment of the after budbreak. While extensive in acre- Stockton current state of soils in the age, almonds may be less ideal because of San Francisco study area. the trees’ sensitivity to saturated condi- Modesto tions and high nitrogen demand (table 1). Merced Walnuts may be an option given that Fresno budbreak typically occurs in late April. Salinas Wine grapes may be the best option be- Visalia Soil Agricultural cause of the extensive acreage planted, Groundwater low nitrogen demand and tolerance to Banking Index standing water (table 1). Almonds with Bakers eld Excellent plum rootstocks may also be suitable; Good Santa Maria however, currently almonds with water Moderately good tolerant rootstocks are generally planted Moderately poor in soils that are poorly drained and thus Oxnard Poor Los Angeles less likely to be suitable for groundwater Very poor Indio banking. Recharge potential 0 50 100 A preliminary calculation based only San Diego

± Miles Lab Soil Resource California on soil properties and crop type shows that landscapes rated Excellent or Good Fig. 5. Spatial extent of Soil Agricultural Groundwater Banking Index suitability groups accounting for could be used to bank as much as 1.2 modifications by deep tillage. million acre-feet of water per day. This Toby A. O’Geen Toby

The undulating agricultural land found along many valley margins in California is poorly suited to groundwater banking because application of floodwater or waste water would be difficult to apply at these sites, which are typically drip irritated.

82 CALIFORNIA AGRICULTURE • VOLUME 69, NUMBER 2 estimate assumes 1 foot per day of water infiltration on lands in the Excellent and Good categories that are planted with grapes (460,000 acres) or alfalfa (300,000 acres), or fallowed (440,000 acres). There are significant limitations to this esti- mate. Most importantly, California lacks the infrastructure to accommodate and route such large volumes of water to the fields in such a short time (presuming that floodwater is the source of the wa- ter). Plus, the heterogeneity in precipita- tion across the state makes this estimate improbable (that is, it is unlikely that floodwater availability would be geo- graphically close to the best lands for UC Merced Viers, Josh recharge). Offsetting these limitations to some degree are other crop types that would be suitable for recharge (i.e., an- nual crops) but were not included in this estimate. Agricultural groundwater banking must be approached with caution. The financial risk associated with crop loss may exceed the potential benefits of water savings. Perennial crops carry particular risks and uncertainties. For instance, while trees and vines are generally more tolerant of saturation before budbreak than after (table 1), determining a reliable cutoff date for this increased tolerance is difficult. Tree and vine roots gener- ally start to grow several weeks before budbreak, so damage from waterlog- ging can occur well before budbreak. Moreover, budbreak for a given species

varies by location across the state. In ad- Doll David dition, standing water on trunks can lead Orchards of walnuts (above) and almonds (below) may be viable sites for groundwater recharge, though to aerial Phytophthora or other diseases. the potential for water damage to such high-value crops adds risk. Investigating this opportunity in less valuable cropping systems, such as alfalfa, irrigated pasture and annual crops may irrigation districts for enrolling in a long- participation in groundwater banking be more promising until further research term program. Long-term commitments programs likely would be lower as well. on tree crop sensitivity to standing water from growers likely would be needed for Most annual cropping systems would has been conducted. basin-scale planning purposes. be suitable for groundwater banking if If groundwater banking on agricul- Although not included among the water is applied when land is fallow. The tural lands becomes a priority, coordina- crops listed in table 1, alfalfa may be an major risk in annual crops is leaching of tion at the policy, market and planning ideal crop for groundwater banking be- residual pesticides or fertilizer in the soil. levels would be needed to provide an cause it requires little or no nitrogen fer- Appropriate management practices for adequate land base ready to opportunisti- tilizer, reducing the risk that groundwater groundwater banking with specific an- cally capture floodwaters. Adoption of recharge would transport nitrates into nual crops would need to be developed. this practice would likely require some aquifers. Alfalfa is sensitive to flooding If agricultural groundwater banking form of support to mitigate or protect and saturated conditions; thus the timing becomes an important water security growers from the risks of crop failure. of flooding should coincide with older practice, the SAGBI may provide valuable For example, growers who make their fields (typically 4 to 5 years old) slated information to guide future changes in land available for floodwater capture and for replanting. Because the financial risk cropping systems. groundwater banking could receive cred- associated with crop damage is lower in SAGBI can be a powerful aid to deci- its from municipalities or irrigation dis- alfalfa than in tree and vine crops, the fi- sion makers and stakeholders when con- tricts. They could also receive credits from nancial incentive needed to drive grower sidering the tradeoffs associated with the

http://californiaagriculture.ucanr.edu • APRIL–JUNE 2015 83 implementation of groundwater banks during groundwater banking events. ratings of each SAGBI factor for any loca- utilizing agricultural land for direct re- Furthermore, deep sediment likely con- tion that has a SAGBI rating, illustrating charge. It was also developed with the tains hydraulically restrictive horizons the transparency of the model and allow- intention of informing growers of the po- that have not been documented, creating ing for further investigation of individual tential hazards associated with this prac- uncertainty as to where the water travels. factors. c tice. As is the case with any model, and An understanding of the depth to the with soil survey information in particular, groundwater table is also needed. ground-truthing at the field scale is neces- Given these issues, SAGBI may be most T.A. O’Geen is UC Agriculture and Natural Resources sary to verify results. useful when used in concert with water Cooperative Extension (UCCE) Specialist; M. Saal is Graduate Student in the Soils and Biogeochemistry Graduate Group at UC Davis; H. Dahlke is Assistant If agricultural groundwater banking becomes an important Professor in the Department of Land, Air and Water water security practice, the SAGBI may provide valuable Resources at UC Davis; D. Doll is UCCE Farm Advisor in Merced County; R. Elkins is UCCE Farm Advisor in Lake information to guide future changes in cropping systems. and Mendocino counties; A. Fulton is UCCE Irrigation and Water Resources Advisor in Tehama County; G. Fogg is Professor, T. Harter is UCCE Specialist, and J.W. We acknowledge limitations to our infrastructure models and hydrogeo- Hopmans is Associate Vice Provost and Professor in the model. It does not consider proximity to logic models — which generally do not Department of Land, Air and Water Resources at UC a surface water source, which is an is- incorporate soil survey information in a Davis; C. Ingels is UCCE Farm Advisor in Sacramento County; F. Niederholzer is UCCE Farm Advisor in Sutter- sue especially in areas that are irrigated comprehensive way — to develop a fuller Yuba counties; S. Sandoval Solis is Assistant Professor solely from groundwater wells and are assessment of the processes and limita- and UCCE Specialist in the Department of Land, Air and not connected to conveyance systems tions involved in a potential groundwater Water Resources at UC Davis; P. Verdegaal is UCCE Farm that supply surface water. The SAGBI also banking effort. Advisor in San Joaquin County; and M. Walkinshaw is does not consider characteristics of the GIS Analyst in the Department of Land, Air and Water Information delivery vadose zone (the unconsolidated material Resources at UC Davis. This research was conducted with funding from the below soil and above the groundwater Our goal is to make SAGBI an interac- UC ANR Competitive Grants Program. We also thank the table) or depth to groundwater. In arid tive, web-based app. The decision support anonymous reviewers for their helpful feedback. regions, deep vadose zones may contain tool will display SAGBI groups as a map contaminants such as salts or agricul- in Google Maps. Users will be able to tural pollutants that have accumulated navigate via standard map interface op- over years of irrigation and incomplete erations such as zoom tools and panning, leaching. These deep accumulations of or by entering a location in a search field contaminants could be flushed into the to obtain SAGBI ratings. Users will also be water table when excess water is applied able to query and display the individual

References Amundson R, Guo Y, Gong P. 2003. Soil diversity and land Howell TA. 2001. Enhancing water use efficiency in Soil Survey Division Staff. 1993. Soil survey manual. Soil use in the United States. Ecosystems 6:470–82. irrigated agriculture. Agron J 93:281–9. doi:10.2134/ Conservation Service. US Department of Agriculture Barnett TP, Pierce DW, Hidalgo HG, et al. 2008. Human‐in- agronj2001.932281x Handbook 18. duced changes in the hydrology of the western United Kourakos G, Klein F, Cortis A, Harter T. 2012. A ground- Tindula G, Orang M, Snyder R. 2013. Survey of irriga- States. Science 316:1080–83. water nonpoint source pollution modeling framework tion methods in California in 2010. J Irrig Drain Eng: Bachand PAM, Horwath WR, Roy SB, et al. 2011. On-farm to evaluate lon-term dynamics of pollutant exceedance 139:233–8. flood flow capture – Addressing flood risks and ground- probabilities in wells and other discharge locations. Wa- USDA-NRCS. 2014. California Soil Survey Status Map. water overdraft in the Kings Basin, with potential applica- ter Resour Res 48: W00L13. doi:10.1029/2011WR010813. www.nrcs.usda.gov/Internet/FSE_DOCUMENTS/ tions throughout the Central Valley. Davis, CA: Bachand Le Bissonnais Y. 1996. Aggregate stability and assessment nrcs144p2_063838.pdf (accessed October 2014). & Associates. 4 p. http://aquaticcommons.org/11289/ of soil crustability and erodibility: I. Theory and method- Ward FA, Pulido-Velazquez M. 2008. Water conservation (accessed May 2015). ology. Eur J Soil Sci 47:425–37. in irrigation can increase water use. P Natl Acad Sci USA Canessa P, Green S, Zoldoske D. 2011. Agricultural Water Megdal S, Hamann R, Harter T, et al. 2009. Water, People, 105:18215–20. doi:10.1073/pnas.0805554105. Use in California: A 2011 Update. Center for Irrigation and the Future: Water Availability for Agriculture in the Water Association of Kern County. 2014. Water banking. Technology, Fresno State University, November. www. United States. CAST Issue Paper 44. 20 p. www.wakc.com/index.php/water-overview/sources-of- californiawater.org/cwi/docs/CIT_AWU_REPORT_v2.pdf. [NASS] National Agricultural Statistics Service. 2014. Na- water/87-water-banking (accessed October 2014). Cayan DR, Das T, Pierce DW, et al. 2010. Hydrology of the tional Agricultural Statistics Service website. www.nass. White EM, Gartner FR, Butterfield R. 1981. Range claypan early 21st century Southwest drought: Possible harbinger usda.gov/. soil improvement: response from furrowing and rip- of future decades. P Natl Acad Sci USA 107:21271–76. Orang M, Matyac SJ, Snyder R. 2008. Survey of irriga- ping in northwestern South Dakota. J Range Manage Charbonneau R, Kondolf GM. 1993. Land use change in tion methods in California in 2001. J Irrig Drain Eng 34:119–25. California: Nonpoint source water quality impacts. Envi- 134:96–100. ron Manage 17:453–60. Ruud N, Harter T, Naugle A. 2004. Estimation of ground- Department of Water Resources. 2015. California’s water pumping as closure to the water balance of a Groundwater Update 2013. A Compilation of Enhanced semi-arid, irrigated agricultural basin. J Hydrol 297:51–73. Content for California Water Plan Update 2013. State of Schoups G, Hopmans JW, Young CA, et al. 2005. Sustain- California Natural Resources Agency. April 2015. www. ability of irrigated agriculture in the San Joaquin Valley, waterplan.water.ca.gov/docs/groundwater/update2013/ California. P Natl Acad Sci USA 102:15352–6. content/statewide/GWU2013_Combined_Statewide_Fi- nal.pdf.

84 CALIFORNIA AGRICULTURE • VOLUME 69, NUMBER 2