San Luis Delta Mendota Water Authority#

Written by Wendy Haw

Model Region and Spatial Coverage#

The databases encompass the San Luis Delta Mendota Water Authority (SLDMWA) regions. There are 29 distinct regions within this boundary.

Table 1 Modeling regions utilized in the database#
WDNAME	HR NAME
BANTA-CARBONA I.D.	San Joaquin River
BROADVIEW W.D.	San Joaquin River
Byron Bethany (CVP)	San Joaquin River
CENTRAL CALIFORNIA I.D.	San Joaquin River
CITY OF TRACY	San Joaquin River
COLUMBIA CANAL CO.	San Joaquin River
DEL PUERTO WATER DISTRICT	San Joaquin River
EAGLE FIELD W.D.	San Joaquin River
FIREBAUGH CANAL W.D.	San Joaquin River
FRESNO SLOUGH W.D.	Tulare Lake
GRASSLANDS W.D.	San Joaquin River
HENRY MILLER RECLAMATION DIST. 2131	San Joaquin River
JAMES I.D.	Tulare Lake
LAGUNA W.D.	San Joaquin River
MERCY SPRINGS W.D.	San Joaquin River
ORO LOMA W.D.	San Joaquin River
PACHECO W.D.	San Joaquin River
PANOCHE W.D.	San Joaquin River
PATTERSON W.D.	San Joaquin River
PLEASANT VALLEY W.D.	Tulare Lake
RECLAMATION DISTRICT 1606	Tulare Lake
SAN BENITO CO. W.D.	Central Coast
SAN LUIS W.D.	San Joaquin River
SANTA CLARA VALLEY W.D.	San Francisco Bay
THE WEST SIDE I.D.	San Joaquin River
TRANQUILLITY I.D.	Tulare Lake
Turner Island Water District	San Joaquin River
WEST STANISLAUS I.D.	San Joaquin River
WESTLANDS W.D.	Tulare Lake

Crop Categories and Assignments#

A key challenge in assembling these databases is how crops are classified across different data sources. California agriculture includes over 400 commodities, and it is not feasible to model each crop individually. As a result, crops must be grouped into broader categories.

Individual commodities in the land cover datasets from LandIQ grouped into 23 crop categories to reduce quantities of data required for modeling. Categories were based on existing DWR crop groups and were revised to improve resolution of important commodities such as almonds, pistachios, and walnuts. More information on this is given in the “Land Cover” section. Each combination of crop category and hydrologic region was assigned a proxy commodity which was used to bridge between the land cover in the model and the economic data from the USDA NASS dataset. Economic proxies were chosen following analysis of acreage and revenue trends for NASS commodities within each hydrologic region to reflect both the prevalence of individual crops and total economic value. Relationships for category development and proxies are given in Table 2.

Table 2 Model crop categories, commodities, and economic proxies#
Crop Group	LandIQ Commodities	Economic Proxy Commodities
Alfalfa	Alfalfa and alfalfa mixtures	HAY ALFALFA
Almonds	Almonds	ALMONDS ALL
Beans Dry	Beans (dry)	BEANS DRY EDIBLE UNSPEC. BEANS LIMA LG. DRY
Berries	Bushberries Strawberries	BERRIES STRAWBERRIES F MKT BERRIES BLUEBERRIES
Subtropical	Citrus and Subtropical - No Subclass Subtropical Fruits Misc. Olives Kiwis	OLIVES ORANGES NAVEL
Corn	Corn, Sorghum or Sudan (grouped for remote sensing classification only)	CORN GRAIN CORN SILAGE
Cotton	Cotton	COTTN LINT UPLAND COTTN LINT PIMA
Cucurbits	Melons, Squash, and Cucumbers	SQUASH MELONS CANTALOUPE
Field and Grain	Wheat Field Misc. Grain and Hay - Misc. Grain and Hay - No Subclass	WHEAT ALL
Grapes	Vineyards - No Subclass	GRAPE WINE GRAPE TABLE GRAPE RAISIN
Lettuce	Cole crops (mixture of T22-T25) Lettuce or Leafy Greens (grouped for remote sensing classification only)	LETTUCE HEAD
Onions and Garlic	Onions and Garlic	ONIONS
Orchards	Apples Avocados Cherries Dates Deciduous - Misc. Deciduous - Mixed Peaches and Nectarines Pears Plums, Prunes or Apricots (grouped for remote sensing classification only) Pomegranates	PLUMS DRIED CHERRIES SWEET PEACHES FREESTONE
Pasture	Pasture - Miscellaneous Grasses Pasture - Mixed Pasture - Native Improved	N/A
Pistachios	Pistachios	PISTACHIOS
Potatoes	Potato or Sweet potato (grouped for remote sensing classification only) Potatoes	POTATOES SWEET POTATOES IRISH ALL
Rice	Rice Rice - Wild	RICE MILLING
Safflower	Safflower	SAFFLOWER
Sunflowers	Sunflowers	SUNFLOWER SEED PLANTING
Tomatoes	Tomatoes (Processing)	TOMATES FRESH MARKET TOMATES PROCESING
Truck	Carrots Peppers (Chili, Bell, etc.) Truck Crops - Misc. Truck Crops - No Subclass	PEPPERS BELL
Walnuts	Walnuts	WALNUTS ENGISH
Young Perennial	Young Perennial (grouped for remote sensing or when CLASS C, D or V is not determined)	N/A

Table 3 shows the 2 crop classification schemes that are used: OpenAg and DWR. DWR uses a more aggregated system with 19 crop groups while OpenAg uses more detailed and expanded classifications with 23 crop groups. Either classification system may be used in the database assembly process depending on the context and the intended application.

Table 3 Crop classification schemes#
OpenAg	DWR
Alfalfa	Alfalfa
Almonds	Almonds & Pistachios
Beans Dry	Citrus & Subtropical
Berries	Corn
Corn	Cotton
Cotton	Cucurbits
Cucurbits	Dry Beans
Field and Grain	Grain
Grapes	Onions & Garlic
Lettuce	Other Deciduous
Onions and Garlic	Other Field Crops
Orchards	Pasture
Pasture	Potatoes
Pistachios	Rice
Potatoes	Safflower
Rice	Sugar Beets
Safflower	Tomato Processing
Subtropical	Truck Crops
Sunflowers	Vineyard
Tomatoes
Truck

Land Cover#

Field-level parcels for 2014, 2016, and 2018-2024 was provided by LandIQ through the California Open Data portal. These geospatial layers were intersected with the modeling region in Python using the GeoPandas library and crop acreages were then calculated for each model region and hydrologic region. When computing the land use acreage, there were parcels that overlapped with multiple hydrologic regions. To remedy this, the hydrologic region associated with the overlapped water district (SLDMWA) was used.

In the LandIQ datasets, young perennials are reported as an unspecified category without identifying the specific crop type. To allocate this acreage, the young perennial acreage was distributed across standing-bearing perennial crops (Almonds, Grapes, Orchards, Pistachios, Subtropical, and Walnuts) based on each crop’s share of total standing-bearing acreage within a modeling region. For example, if a region’s standing-bearing acreage consists of 50 percent almonds, 20 percent walnuts, and 30 percent grapes, then the young perennial acreage is assigned to these crops in the same proportions.

Price and Yield#

Price and yield data were obtained from the USDA National Agricultural Statistics Service (NASS) County Agricultural Commissioner Reports, which provide information on crop prices, yields, and acreage in California. Values were averaged over the 2018-2020 period, and prices were adjusted to 2024 dollars using the Consumer Price Index (CPI). After compiling these data, representative price and yield values were assigned to each aggregated crop category through a proxy-crop selection process. When a single crop clearly dominates a category – defined as having an acreage share more than 10 percent higher than any other crop – that crop is designated as the proxy. If multiple crops have similar acreage shares, an acreage-weighted average is computed, and the crop whose price and yield are closest to that average is selected as the representative proxy.

Some price and yield values were missing in the NASS reports. To address these gaps, a hierarchical imputation procedure was applied, using progressively broader spatial aggregates. Missing values were first filled using averages for the same crop in neighboring counties. If unavailable, averages within the same hydrologic region were used. Remaining gaps were filled using averages from neighboring hydrologic regions, and finally, statewide averages were applied as a last resort.

Several crop-specific adjustments were also required. For tomatoes, fresh and processing types were distinguished based on regional production patterns: coastal hydrologic regions were assigned fresh tomatoes, while non-coastal regions were assigned processing tomatoes. For grapes, multiple types (wine, table, raisin) were aggregated using revenue-weighted averages at the hydrologic region level to reflect their economic composition. For lettuce, consistent regional data were unavailable, so values from the 2017 UC Davis Iceberg Lettuce cost and return study were applied across all regions. Finally, when converting from OpenAg to DWR crop classifications, multiple crops were combined into broader groups. Acreage-weighted averages were used to ensure that aggregated price and yield values reflect the relative land shares of the underlying crops.

Applied Water#

Water use data was obtained from the California Department of Water Resources (DWR) Agricultural Land and Water Use Estimates dataset for the years 2018-2020. This dataset provides annual estimates of applied water, irrigated crop acreages, crop evapotranspiration, and evapotranspiration of applied water for 20 crop categories. A Python script was used to extract the irrigated crop area (ICA; acre) and applied water (AW; acre-ft/acre) data, aggregate them at the hydrologic region scale, and compute the average applied water per acre-ft for each crop, restricting the analysis to entries where AW values were greater than zero.

\[Avg\ AW = \frac{AW}{ICA} \qquad (1)\]

While DWR uses a 20 crop classification system, the OpenAg database expands a few of these crop classifications into separate categories. For instance, DWR groups almonds and pistachios together while OpenAg considers them as separate crop classifications. Additionally, lettuce and berries are singled out of the truck crop group as individual categories. To account for crop level water use disaggregation, an additional Python script was employed to compute the average applied water for these expanded crop categories using data from the previously mentioned dataset for the years 2016-2019. The DWR applied water values were scaled using crop-specific ratios to estimate crop-level applied water:

\[{Final\ AW}_{c} = s_{c} * g \qquad (2)\]

where $s$ represents the split factor, which is the ratio of the recent (2016-2019) average applied water from crop $c$ to the baseline (2011-2013) applied water for the crop group. The term $g$ represents the DWR crop group average applied water value. The split factor is essentially a scaling ratio used to disaggregate applied water values when converting between the DWR crop groups to the OpenAg classifications. The following table presents the split factors applied to derive individual crop-level applied water values. These per-crop averages serve as the coefficients used to allocate water use across individual crops.

Table 4 Individual Crop Split Factors#
Crop Group (DWR)	Crop Group Baseline AW (2011-2013)	Single Crop (OpenAg)	Individual Avg AW (2016-2019)	Split Factor
Almonds and Pistachios	3.9427	Almonds	4.3333	1.10
Almonds and Pistachios	3.9427	Pistachios	3.997528567	1.01
Truck	2.063	Berries	2.676115961	1.30
Truck	2.063	Walnuts	3.829978110	1.86

A key limitation of the Agricultural Land and Water Use Estimates dataset is that it does not report applied water values for fresh tomatoes; therefore, processing tomato data was used as a proxy. The dataset also does not report applied water values for certain crops in specific hydrologic regions. To address these gaps, non-zero crop-specific applied water values from neighboring hydrologic regions were averaged and used as proxies. Applied water for sugar beets is not reported in any hydrologic region, so values for “Other Field Crops” within each region were used as a substitute. Omegawater for Pasture was reduced to $10/acre-ft and for Field and Grain to $25/acre-ft to maintain positive net revenues.

Costs#

The cost and return data are sourced from the UC Davis Agricultural and Resource Economics Cost and Return Studies, which provide detailed per-acre production cost information for a wide range of commodities, including labor, land, supplies, and other inputs. The dataset includes both current studies, which are preferred, and archived studies that are used when more recent data are unavailable. A key challenge is that these studies do not offer complete coverage across all crops and hydrologic regions, making additional supplementation and harmonization necessary to ensure consistent cost structures throughout the dataset.

To create a consistent cost dataset, production costs are expressed as shares of total revenue rather than absolute dollar values. This standardization allows meaningful comparison across crops with very different price levels or production scales. Each cost component – such as land, labor, or supplies – is divided by total commodity revenue to obtain its proportional share. Equation 3 provides the formula used to calculate the cost shares:

\[cs_{x} = \frac{c_{x}}{c_{total}} \qquad (3)\]

where $cs_x$ represents the cost structure of the individual cost component $c_x$ and $c_{total}$ is the total commodity revenue.

Several crop-specific adjustments were required to achieve full coverage. Some adjustments are similar to those implemented with the price and yield data to ensure consistency. Cost data for pasture were unavailable, so the cost structure from the ‘Field and Grain’ category was used as a proxy. For tomatoes, fresh and processing types were distinguished based on regional production patterns, with coastal hydrologic regions assigned fresh tomatoes and non-coastal regions assigned processing tomatoes. For grapes, multiple types – wine, table, and raisin – were aggregated using revenue-weighted averages at the hydrologic region level to reflect their economic composition. For lettuce, consistent cost data were not available across regions, so values from the 2017 UC Davis Iceberg Lettuce cost and return study were applied statewide. Additional assumptions were made for water costs: a standard value of $50 per acre-foot was applied to most crops, with lower values for pasture and field crops and a higher value for lettuce. When converting from OpenAg to DWR crop classifications, acreage-weighted averages were used to ensure that aggregated cost shares reflect the relative land contributions of each crop.

Addressing gaps in spatial and crop coverage was another crucial step. Because cost and return studies do not cover all crops or all hydrologic regions, missing values were supplemented using data from nearby hydrologic regions or from the same crop in other regions when available. In some cases, additional adjustments were required to maintain consistency. For example, cost values for certain crops in the Sacramento River region – such as grapes, cucurbits, and safflower – were replaced with values from the San Joaquin River region to ensure alignment in cost structures across regions.

Supply Elasticities#

Elasticity of supply values were adapted from previous work by Jose Manuel Rodriguez Flores for other agricultural economics modeling projects focusing on the southern San Joaquin Valley and Westlands Water District.

Crop supply elasticities were computed through the use of panel data analysis and the partial adjustment framework for counties belonging to the San Joaquin Valley. Proposed by Nerlove 1956, the framework relies on the assumption that farmers use crop price information from the previous year to determine the production size of the same crop in the present year. Panel data estimations were computed via single-equation estimation. The approaches used were Feasible Generalized Least Squares (FGLS), Pooled Ordinary Least Squares (OLS), Fixed Effects Models (FE), and Random Effects Models (RE). Model selection for crop type was dependent on the characteristics of the panels and data. The dataset has a small number of panels (N = counties) and a large number of time periods (T = years).

The following single equation formulation was used:

\[{lnA}_{t} = \beta_{0} + \beta_{1}{lnA}_{t-1} + \beta_{2}{lnP}_{t-1} \qquad (4)\]

where $\ln A_t$ represents the natural logarithm of harvested acres in year $t$. This value is modeled as a function of $\ln A_{t-1}$ and $\ln P_{t-1}$, the previous year’s harvested acres and price, respectively. These two values serve as a proxy for farmers’ price expectation in year $t$. Both $\beta_1 \geq 0$ and $\beta_2 \geq 0$ are expected, in which the number of acres increases due to an increase in anticipated prices. The coefficient $\beta_2$ represents the short-term supply price elasticity while the ratio $\frac{\beta_2}{(1-\beta_1)}$ is the long-term supply response.

The data used was obtained from USDA Agricultural Commissioner reports, which provides annual data on harvested acres, price, yield, production, and value of each crop for all counties in California. To have balanced panels, counties for each crop were selected only if they had a complete time series from 1980-2016. Proxy crops were derived from those utilized in the Statewide Agricultural Production Model (SWAP). These crop supply elasticity values were then expanded for the OpenAg crop classifications where certain groups were separated into individual categories.

The original elasticity values are defined using the OpenAg crop classification. To align these values with the DWR classification, certain crops in the OpenAg scheme are aggregated together. The following table summarizes the crop groupings used. For each DWR crop group, the aggregated elasticity value is computed as the simple average of the individual elasticities of the corresponding OpenAg crops.

Table 5 DWR Crop Group Conversion to OpenAg Single Crop#
Crop Group (DWR)	Single Crop (OpenAg)
Almonds and Pistachios	Almonds
Almonds and Pistachios	Pistachios
Truck	Berries
	Lettuce
	Walnuts
Other Deciduous	Orchards
Other Deciduous	Walnuts

Getting Access#

To get access for additional staff, please contact the WSM Lab.

San Luis Delta Mendota Water Authority

Contents