Implementation of wetland adaptive water quality management strategies under real-time salinity TMDL’s

Principal Investigators:

Thomas C. Harmon, Ph.D.UC Merced School of Engineering and Sierra Nevada Research Institute

tharmon@ucmerced.edu

Nigel W.T. Quinn, Ph.D., P.E.Lawrence Berkeley National Laboratory and

UC Merced Sierra Nevada Research Institutenwquinn@lbl.gov

UC Water Resources Center Technical Completion Report - Project No. SD007

December 2008

1

ACKNOWLEDGMENTS

Funding. The research described in this report was made possible by funding from the University of California’s Water Resources Center (UC WRC) Salinity Drainage Program (Award No. WRC-SD007) which primarily supports graduate student stipends. Additional project support was obtained from the California Department of Water Resources (PI’s Quinn and Harmon) and the State Water Resources Control Board (PI Quinn). Funding from the State Water Resources Control Board Real-time Wetland Water Management Project provided the real-time water quality monitoring system and the remote sensing imagery for years 2006, 2007 and 2008. The project benefited significantly from in-kind support from the Department of Fish and Game and Grassland Water District. Supplemental funds supporting UC Merced students and investigators were provided by the National Science Foundation’s Center for Embedded Networked Sensing (CENS).

Personnel. Many students and staff collaborated on this UC WRC project beyond the lead investigators Harmon and Quinn. These include John Beam, William Cook, Lara Sparks, and Ricardo Ortega of the California Department of Fish and Game. Ernie Taylor and Joe Tapia of the California Department of Water Resources also participated on the project. UC Merced faculty member Qinghua Guo, graduate students Patrick Rahilly and Donghai Li, staff research scientist Alex Rat’ko, and undergraduate assistants Ruby Gonzalez-Jimenez and Carrie McGarraugh. Mr. Rahilly completed his M.S. thesis while a graduate student, with primary funding from this project.

2

Implementation of wetland adaptive water quality management strategies under real-time salinity TMDL’s

Table of Contents

ABSTRACT....................................................................................................4

1. PROJECT BACKGROUND AND OVERVIEW....................................7

1.1 SEASONAL WETLAND RESOURCE MANAGEMENT........................................71.2 MOIST-SOIL MANAGEMENT....................................................................81.3 MOIST-SOIL PLANT VEGETATION............................................................91.4 REAL-TIME WATER QUALITY MANAGEMENT IN THE SAN JOAQUIN BASIN

....................................................................................................................101.5 EXPERIMENTAL PLAN............................................................................11

2. REAL-TIME FLOW AND WATER QUALITY MONITORING........12

3. USE OF REMOTE SENSING TO ESTIMATE CHANGES IN WETLAND MOIST SOIL PLANT HABITAT............................................13

3.1 OBJECTIVES............................................................................................133.2 METHODS AND RESULTS........................................................................13

3.2.1 VEGETATION CLASSIFICATION USING MULTI-SPECTRAL IMAGERY133.2.2 SWAMP TIMOTHY PRODUCTIVITY UNDER A MODIFIED HYDROLOGY16

3.3 SUMMARY............................................................................................18

4. USE OF ELECTROMAGNETIC BULK SALINITY MAPPING TO ASSESS IMPACTS OF A MODIFIED WETLAND HYDROLOGY........19

4.1 EXPERIMENTAL METHODS....................................................................194.1.1 ELECTROMAGNETIC (EM) SPATIAL SOIL MAPPING METHODS (2007)................................................................................................................194.1.2 ELECTROMAGNETIC (EM) SPATIAL SOIL MAPPING METHODS (2008)................................................................................................................214.1.3 SOIL MOISTURE, TEMPERATURE, AND SALINITY SAMPLING METHODS................................................................................................22

4.2 RESULTS AND DISCUSSION....................................................................234.2.1 EM SALINITY MAPPING RESULTS (2007)........................................234.2.2 EM SALINITY MAPPING RESULTS (2008)........................................264.2.3 SOIL MOISTURE, TEMPERATURE, AND SALINITY TEMPORAL RESULTS

................................................................................................................304.3 SUMMARY................................................................................................31

5. SUMMARY, CONCLUSIONS, AND APPLICABILITY....................32

REFERENCES CITED................................................................................34

3

4

ABSTRACT

This project has provided scientific support to a multi-year, interdisciplinary study of potential impacts to wetland moist soil plant habitat resulting from modification of the scheduling of seasonal wetland drawdown within the San Joaquin River Basin. Seasonal wetland drainage contributes salt loading to the San Joaquin River – changing the timing of these wetland contributions to the River, if part of a comprehensive, basin-wide real-time water quality management system, can improve compliance with State salinity objectives and the current salinity TMDL. Sustaining seasonal wetland habitat for waterfowl is mandated by Federal law and supported by State agencies and private foundations. These wetlands are an important over-wintering resource for waterfowl and shorebirds on the Pacific Flyway as well as supporting the local economy of Merced County through duck hunting and more than 160 private duck clubs. This research project centers around six pairs of seasonal wetland ponds (adjacent except in the case of Volta Wildlife Management Area), which were selected from amongst the State and private managed wetlands within the 170,000 acre Grasslands Ecological Area. Each pair of matched sites allows traditional wetland drawdown management practices to be compared to those where drawdown is delayed until April 15 each year – with respect to impacts on water quality, wetland soils and moist soil plant habitat. The study aims to answer the question “How long can modified wetland management practices be sustained, that are designed to improve water quality in the San Joaquin River (SJR), without negatively impacting the biological value of waterfowl habitat?”

The project has made the following accomplishments during:

The project team assisted in the deployment of 24 telemetered (radio and cellular modem) flow and water quality monitoring stations measuring continuous electrical conductivity, temperature, and stage were established at the inlet and outlet of the six paired wetland pond sites. The stations now measure salt fluxes in and out of each pond.

Three sets of high-resolution multi-spectral images were acquired for the project study area (pre-treatment in 2006, and post-treatment 2007, 2008). Images for 2006 have been processes and analyzed, while those for 2007 and 2008 will be completed in early 2009. We have developed techniques using plant association-specific spectral signatures to identify 29 of the most important wetland plant associations. Over 500 ground truth locations were sampled to verify the accuracy of the technique. Of the 29 signatures, 9 were moist-soil plant associations that included Crypsis schoenoides

5

(swamp timothy), which is a dominant moist-soil plant in the Grasslands Ecologic Area. With the spectral signatures, classification was performed to estimate areas of swamp timothy presence and absence across the entire study area.

In spite of suboptimal timing of the 2006 images, several pond sites exhibited strong correlation between several spectral signatures and plant productivity in terms of total above-ground biomass and seed production have been identified using multi-spectral data and ground truth based on vegetative sampling. Given improved flight timing, the 2007 and 2008 results are expected to yield improved results.

High-resolution soil salinity maps were created using a Geonics electromagnetic field instrument (EM-38). The EM-38 produces a relative soil salinity map that must then be calibrated to salinity values measured from physical samples. Twelve soil samples per pond were used to calibrate the maps. Results from 2007 were of limited value scientifically due to the timing of the data collection, but served as guidance for the 2008 campaign, which resulted in good correlation between EM results (using the next generation EM-38 MK2) and ground-truth samples and ultimately reliable salinity maps for the pond sites.

The time series data on moisture, temperature and salinity can provide data for creating temporal interpolations between the spatial mapping times. Because the spatial maps require significant effort in both collecting and analyzing the data, having the time series data is important. Additional work to determine the best scheme for fusing these two types of data is clearly warranted.

Methods have been developed for correlating reflectance spectra from aerial imagery with manual swamp timothy productivity survey data was completed for 2007 and are currently being applied to the 2008 data sets. These results are also being compared with the EM-based salinity maps, with salinity-moisture time series data being used to account for any significant time differences in the survey maps. The integrated results will enable us to provide significant input on the question of how long modified wetland management practices be sustained, that are designed to improve water quality in the San Joaquin River, without negatively impacting the biological value of waterfowl habitat.

Supplemental funding was acquired through a California Department of Water Resources Proposition 204 grant (PI Quinn, co-PI Harmon) to develop mathematical models of seasonal wetland hydrology and to improve the representation of these wetlands in the current WARMF-SJR model.

6

In terms of professional development, Patrick Rahilly completed his M.S. in Environmental Systems at UC Merced based on this project, and the project led to ongoing collaborative interactions between the UC project team and the California Departments of Fish and Game and Water Resources.

Significant scientific outputs from the study include: (a) the development of a robust monitoring system platform for measuring real-time flow and water quality data; (b) the finding that moist soil plant vegetation associations vary sufficiently between wildlife management areas and that spectral signatures need to be developed for each independently if accurate representation is to be achieved. Initial classification was performed using e-Cognition software for image segmentation and ERDAS Imagine for classification. We developed a methodology whereby the analysis can all be done within the e-Cognition software. The team developed the first accurate map of major moist soil plant associations for the ponded areas, which have provided the first quantification temporal shifts in wetland plant habitat. (c) Soil salinity surveys revealed a likely mechanism of salt deposition within the wetland soil profile. This has allowed the protocol for sampling bulk soil salinity to be adapted – thus providing a more accurate assessment of soil salinity changes over time as a result of modified hydrology practices designed to improve SJR water quality.

7

1. PROJECT BACKGROUND AND OVERVIEW

This project provided scientific support to an interdisciplinary study of potential impacts to wetland moist soil plant habitat resulting from modification of the scheduling of seasonal wetland drawdown within the San Joaquin River Basin. Changing the timing of seasonal wetland drainage to the River has been proposed as part of a comprehensive, basin-wide real-time water quality management system aimed at improving compliance with State salinity objectives and the current salinity TMDL in the Basin.

The project focused on paired seasonal wetland ponds, which were selected from amongst the State and private managed wetlands within the 170,000-acre Grasslands Ecological Area. Each pair of matched sites allowed traditional wetland drawdown management practices to be compared to those where drawdown is delayed by approximately one month each year – comparisons were with respect to impacts on water quality, wetland soil conditions (moisture and salinity) and moist soil plant habitat. The study aimed to address the question of how modified wetland management practices would affect the biological value of waterfowl habitat in terms of a key wetland moist soil plant community.

1.1 Seasonal wetland resource management

Preservation and enhancement of wetlands in California’s Central Valley is important to ensuring wildlife and habitat diversity. The regional wetlands are home to millions of waterfowl and shorebirds, a diverse community of moist-soil vegetation, and other common and endangered wildlife (Mason, 1969; Cogswell, 1977; Stoddard and Associates, 1998). Because of the great importance of this wildlife, best management practices (BMPs) for wetland management have been developed. Depending on the goals, these BMPs can include grading, discing, mowing, grazing, burning, herbicide application, dry season irrigations, and the timing of wetland flood-up and drawdown. The fall flood-up occurs during the months of September and October, and the spring drawdown occurs during the months of February, March, and April. By timing flood-up and drawdown in the San Joaquin Valley, managers mimic the wet/dry seasonal cycle that these historical wetlands once experienced. This seasonal cycle aids life’s processes and can be adapted to promote desired species.

Under “natural” conditions, this diversity would be supported through seasonal flooding and natural disturbances (drought, fire) that historically followed the seasonal cycle. However, due to anthropogenic effects (water projects, agricultural and urban development, etc.), the hydrologic regime that once defined these annual cycles in the Central Valley no longer exists.

8

To mimic these natural processes, research has been undertaken to understand the role of water manipulation, irrigation, waterfowl habitat requirements and both vegetation and waterbird responses to different management techniques. Altering wetland drainage schedules affects the timing and rate of drawdown of wetland ponds and hence the forage value of the wetlands for migrating and wintering shorebirds and waterfowl. Wetland salinity management measures also affect the productivity and diversity of vegetation that can be grown in the watershed (Mushet et al., 1992).

Wetland drawdowns are timed to make seed and invertebrate resources available during peak waterfowl and shorebird migrations and to correspond with optimal germination conditions (primarily soil moisture and temperature) for naturally occurring moist-soil plants (Smith et al., 1995). Spring drainage that is timed for optimal habitat conditions occurs at a sensitive time for agriculture in the South Delta in that these drainage releases occur during the time crops are being irrigated or the first time and are germinating – potentially affecting crop yields. Studies suggest that approximately 10% of the San Joaquin River’s annual flow, and 30% of its annual salt load, passes through wetlands within the Grasslands Basin, which includes the Grassland Water District (Grober et al., 1995; Quinn et al., 1997; Quinn and Karkoski, 1998).

1.2 Moist-Soil Management

The wetland “best management practice” (BMP) specific to this research project focuses on water level manipulation and is most often called “moist-soil management”. Moist-soil management refers to a process of water level manipulations to promote productive habitat conditions and beneficial vegetation such as smartweed (Polygonum punctatum), watergrass (Echinochloa crus-galli), and swamp timothy (Crypsis schoenoides) for foraging waterfowl (Figure 1.1).

9

Figure 1.1 - Desirable moist-soil plant vegetation associated with Central California’s managed wetlands: (A) Crypsis schoenoides; swamp pricklegrass or swamp timothy; (B) Echinochloa crus-galli; watergrass or barnyard grass (both photos 2003 George Hartwell).

Water-level manipulations include flood-up in the fall and wetland drawdown in the spring, and provide optimal conditions at each stage of vegetation development. In addition to flood-up and drawdown, several summer irrigations are conducted by wetland managers to sustain and improve growth characteristics of the desired vegetation (Figure 1.2). The seeds of moist-soil plants are recognized as a critical waterfowl food source, providing essential nutrients and energy for wintering and migrating birds (Fredrickson and Taylor 1982; Bundy, 1997). Not only does the desirable vegetation provide direct nutritional value through consumption, but it also encourages healthy invertebrate populations, a high-protein food source at critical times of the year (Swanson, 1988; Mushet et al., 1992; Smith et al., 1995; Bundy, 1997; Stoddard and Associates, 1998).

It is generally accepted by wetland managers that during cool wet years, and for wetlands of greater depth, it is better to drain them later because the optimal conditions of soil temperature and soil moisture tend to occur later. Conversely, during warm dry years, and for shallower type wetlands, it is better to drain them earlier because the optimal conditions of soil temperature and soil moisture tend to occur earlier. However, in intensively managed wetland complexes such as the GWD, the heterogeneity of wetland soils, year to year variations in the weather and the complex dynamic ecology of the wetland resource require constant hydrologic manipulation and fine tuning of management decisions by wetland biologists.

1.3 Moist-Soil Plant Vegetation

Many different species of moist soil plant vegetation grow within the GWD. Together they form a mosaic of vegetation communities that provide the habitat required to sustain wildlife. Wetland managers often classify this vegetation, either native or naturalized, into two categories: desirable or non-desirable. Desirable plants include native species that form a healthy mixed marsh or that can provide shelter or food stores to migratory waterfowl and shorebirds. Non-desirable plants are often invasive/introduced species and may consume resources (such as light and soil) that otherwise would go to desirable species.

There are generally three major desirable moist-soil plant communities that are targeted for waterfowl forage potential. These targeted communities are found in a mixed marsh setting and are either dominated by smartweed, swamp timothy, or watergrass. A healthy mixed marsh for the San Joaquin

10

Valley could include several other desirable species such as sprangletop (Leptochloa fascicularis), brass buttons (Cotula coronopifolia), and alkali heath (Frankenia grandifolia). While targeting one of the highly desirable plants in the mixed marsh such as swamp timothy, wetland mangers also promote the other listed species (Smith et al., 1995). Several other acceptable plants work well in a mixed marsh community and can include, but are not limited to, tule or hardstem bulrush (Scirpus acutus), cattail (Typha latifolia), spikerush (Haleocharis palustris), purple ammannia (Ammannia coccinea), alkali bulrush (Scirpus robustis), fat-hen (Atriplex patula), and beggar-ticks (Bidens spp.).

The three desirable plants above, swamp timothy, watergrass, and smartweed, have a tendency to grow in large stands, bordered by mixed marsh consisting of desirable plants along with other acceptable plants. As conditions change (drainage plans, for instance), so does the composition of the stands and border areas. Wetland mangers target species by means of water manipulation and other management practices (i.e. flood-up and drawdown plans, disturbance, dry season irrigation, alternative land use). There are several non-desirable plants that tend to establish a stronghold when conditions are not ideal for the more desirable plants. These non-desirable plants include, but are not limited to, aster (Aster spp.), cocklebur (Xanthium strumarium), salt grass (Distichlis spp.), Bermuda grass (Cynodon dactylon), and dock (Rumex spp.). These species grow in dense stands and can dominate the more desirable wetland species if unchecked (Smith et al., 1995).

1.4 Real-time water quality management in the San Joaquin Basin

To improve flow and water quality conditions in the San Joaquin River system, the California Department of Water Resources formed the San Joaquin River Management Program (SJRMP), a stakeholder group representing many of the agencies, landowners and other parties interested in improving the San Joaquin River ecosystem. One of the SJRMP’s mandates was to reconcile and coordinate the various uses and competing interests along the river. The SJRMP created a number of working subcommittees – one of which was the Water Quality Subcommittee. This subcommittee applied for grants, one of which supported early work on real-time water quality management in the SJR. One of the Water Quality Subcommittee’s initial tasks was to develop solutions like real-time drainage management to address the occurrence of high salinity levels in the lower San Joaquin River at critical times of the year such as the onset of pre-irrigation in Delta agricultural lands (Figure 1.2).

Studies conducted initially under the SJRMP and subsequently by Berkeley National Laboratory, have suggested that wetland drainage from the GWD

11

could be scheduled to coincide with peak assimilative capacity in the San Joaquin River to help improve downstream water quality (Grober et al., 1995; Quinn et al., 1997; Quinn and Karkoski, 1998). Increased surface water supply allocations under the Central Valley Project Improvement Act (CVPIA) have created greater opportunity than existed previously to coordinate the release of seasonal wetland drainage with the assimilative capacity of the San Joaquin River.

Figure 1.2 - Timing of wetland drawdown to coincide with periods of San Joaquin River assimilative capacity (Quinn and Hanna, 2003).

Coordinated releases will help achieve salt and boron water quality objectives and improve both downstream agricultural draws and fish habitat in the main stem of the San Joaquin River and Sacramento-San Joaquin Delta. Improved scheduling of west-side discharges can assist in avoiding conflict with critical time periods for early season irrigation as well as fish rearing and remove an important stressor leading to improvements in the San Joaquin River salmon fishery.

One of the main activities in this cooperative research project involved the installation, operation, and maintenance of wetland pond inlet and outlet monitoring stations. These monitoring stations were installed, mostly with SWRCB funds, to encourage improvements in coordination of salt loads from wetland units, within the Grassland Water District and Los Banos Wildlife Management area, with SJR assimilative capacity. Management of wetland drainage through scheduling of releases to coincide with periods of SJR assimilative capacity can improve the river’s water quality. This project provides a systematic data collection program to evaluate the short and long-term consequences of real-time wetland drainage management.

12

1.5 Experimental Plan

Through leveraging of UC Salinity Drainage Program funding with additional funds from the California Department of Water Resources, and with funding and in-kind services from the California Department of Fish and Game, an expanded experimental scope was developed relative to the initially proposed work. Primarily six paired wetland ponds were included in the study instead of the two pairs originally proposed. Each paired site included one control (traditional practice) and one treatment (delayed drainage) pond. Each year, the control units were drawn down during the typical timeframe (March 15 to April 1), while treatment units were drawn down between May 1 and May 15. This later scheduling of wetland drainage will coincide with the Vernalis Adapative Management Program (VAMP) prescribed fish flows in the San Joaquin River to aid annual salmon migration. The VAMP prescribed discharge can more than double the usual seasonal flow in the River, creating considerable assimilative capacity for salt.

The experimental plan that was carried out during the course of the 2-year study is presented in sections 2-4 described below. These cover the following topics:

2. Real-time flow and water quality monitoring. 3. Remote sensing for mapping vegetation and plant productivity. 4. Mapping bulk salinity in wetland soils using electromagnetic

sensing.

2. REAL-TIME FLOW AND WATER QUALITY MONITORING

Cooperative projects supported by the State Water Resources Control Board and CALFED and led by Principal Investigators Nigel Quinn (LBNL/UCM) and Ric Ortega (DFG) installed monitoring stations equipped with temperature, stage/flow, and salinity (EC) sensors, data-loggers and data communications telemetry at the inlets and outlets of six paired wetland sites. These ponds range in size from 20 to 100 acres and, in all but one case are situated adjacent to each other. One of the paired wetlands (control) is managed traditionally and drained prior to April 15 each year. The adjacent wetland (modified hydrology) is drained after April 15 to coincide with high San Joaquin River assimilative capacity. Data from these stations are available on the web http://www.ysieconet.com/public/WebUI/Default.aspx? hidCustomerID=99 using the YSI EcoNet system which polls each monitoring station every 15 minutes and reports the data to the NIVIS data

13

server. The data from these sites are being used to develop salt mass balances for these seasonal wetlands, which will ultimately be used to inform wetland adaptive management practices.

3. USE OF REMOTE SENSING TO ESTIMATE CHANGES IN WETLAND MOIST SOIL PLANT HABITAT

This section describes the methods and results from efforts to map vegetation community structure and productivity of a pivotal moist soil plant species, swamp timothy (Heleochloa schoenoides), using high-resolution multispectral aerial imagery. The goal here is to provide spatial data on desirable and undesirable plant communities in the context of the modified hydrology experiment. Clearly, if the delayed draw-down scheme is detrimental to the key moist soil plant communities, then it is not a viable management option. Aside from this specific question, remote sensing tools for assessing wetland plant community structure and changes at scales relevant to management decisions are greatly needed in settings such as the GEA.

3.1 Objectives

There were two main objectives associated with remote sensing work: (1) aerial mapping of vegetative communities, and (2) swamp timothy productivity estimation tools based on remote sensing products (e.g., vegetative indices). The methods for analyzing the imagery are based on standard remote sensing techniques, which have been applied previously, but mainly in agricultural settings. The novelty of this work is the extension of these techniques to managed seasonal wetlands, and the simultaneous spatial and temporal assessment of soil salinity, moisture content, and temperature. Analysis of 2006 data has been completed, 2007 is undergoing analysis at the time of this report, and the 2008 images are undergoing preparation by the contractor and will be analyzed in early 2009.

14

3.2 Methods and Results

Remotely sensed imagery (3 band multi-spectral) was flown in 2006, 2007 and 2008 for all the study sites as part of the SWRCB project. All imagery was flown at 6 inch spatial resolution – the 2006 and 2007 imagery was flown by HJW Geospatial Inc. of Oakland, CA and the 2008 imagery by the University of Utah (Professor Christopher Neale). UC Merced students were involved in developing training datasets for this wetland mapping project which were used in conjunction with e-Cognition and ERDAS Imagine software to develop moist soil plant habitat maps.

3.2.1 Vegetation Classification Using Multi-Spectral Imagery

In this analysis e-Cognition was used to perform segmentation of the imagery into a number of polygons with similar spatial characteristics. Imagery was divided into 30 unique spectral classes, representing the typical maximum spectral distance available in the data, based on past remote sensing research at Berkeley National Laboratory. Recent research at LBNL has demonstrated the ability of segmentation algorithms combined with high-end image processing software to discern populations of common moist soil plants from high resolution multispectral imagery. Study site interpretative moist soil plant species maps were generated during each year of the study to compare vegetation response affected by adaptive drainage drawdown practices. These polygons were imported into ERDAS and the training data used to associate the spectral signals with the more common moist soil plant associations. Each individual pixel within the aerial images has three unique values representative of the very near infrared (VNIR), red, and green bands. The VNIR, red, and green bands as well as multiple vegetation indices (Table 3.1) were investigated to discover their predictive capabilities of vegetation classification and key species (swamp timothy) productivity. A supervised classification was performed using ground truth data to check the classification and evaluate its accuracy. This analysis was repeated each year and desirable moist soil plant abundance mapped and compared over the three years of the study. Large areal extents of moist soil plants such as swamp timothy, watergrass and smartweed are indicative of optimal wetland habitat.

3.2.2 Swamp Timothy Productivity under a modified hydrology

15

Manual sampling was undertaken late each season after senescence to develop estimates of swamp timothy productivity under delayed drawdown. Sampling points were randomly generated within the swamp timothy communities on the wetland ponds sites. The number of points within each field depended on the area, one sample per acre. For each sample location, a 10cm x 10cm core was removed from the pedon, allowing the entire aboveground biomass to be clipped. The samples were then oven dried at 105 °C for 24 hours. The total mass of each sample was weighed. The swamp timothy mass was removed from sample and weighed giving the percent composition of swamp timothy per sample. The swamp timothy seeds were then segregated from the vegetative matter and weighed. Various other characteristics of the swamp timothy were measured, including plant height and inflorescence length, width, and mass.

16

Table 3.1 - Vegetation indices investigated with respect to swamp timothy productivity.

VegetationIndex Name Abbreviati

onDescription Reference

Normalized difference vegetation

indexNDVI , where and stand for the spectral reflectance

measurements acquired in the red and near-infrared regions, respectively.

Sellers 1985Myneni, 1995

Simple ratio SR ,Near-infrared/Red reflectance ratio.Baret and Guyot, 1991Tucher, 1979

Soil-adjusted vegetation index

SAVI L ranges from 0 for very high vegetation cover to 1

for very low vegetation cover. L=0.5 is used in this study. Huete 1988Rondeaux, 1996

Transformed soil-adjusted vegetation

index

TSAVI Where a and b are, respectively, the slope and the

intercept of soil line ( )

Baret, 1989Thenkabail, 2000

Modified soil-adjusted vegetation

index

MSAVIQi et al., 1994

17

Global environment monitoring index

GEMI

GEMI can reduce both the soil and the atmospheric effects on satellite data.

Pinty, 1992Rondeaux,1996

18

A number of vegetation indices (VI) have been proposed and their merits have been listed in the Table 3.1. Each of these (along with individual bands) was tested for correlation with swamp timothy aboveground biomass and seed productivity. The results are summarized in Table 3.2, with typical regression results plotted in Figure 3.A. Using thee correlation coefficient (R-squared) as a criteria for correlation strength, the results demonstrate that vegetation indices calculated from the aerial photo taken in June 2006 were more strongly related to swamp timothy seed mass than those taken from the May images. More specifically, the simple ratio was a relatively strong indicator of swamp timothy seed mass in the Duck Strike North using June aerial photo (R2 = 0.603), total swamp timothy (0.716) and total biomass (0.676). For the Duck Strike South site, the seed mass was more strongly related to TSAVI but the correlation was somewhat weaker than was the case for Ducky Strike North. Biomass productivity estimation as a function of swamp timothy seed production is shown in Figure 3.B for the Ducky Strike ponds using the simple ratio (SR) vegetative index.

Table 3.2 – Correlation coefficients for the vegetation properties and vegetative indices tested based on the 2006 high-resolution multi-spectral imagery.

May 2006 Vegetation Property

NDVI SR SAVI TSAVI MSAVI GEMI

DSN Total Biomass 0.613P 0.622 E 0.510C 0.619 C 0.464 C 0.234 C

Total ST 0.405 Q 0.469 Q 0.282 Q 0.399 Q 0.273 Q -ST seed production

0.402 Q 0.483 Q 0.242 E 0.391 Q 0.235 E -

DSS Total Biomass 0.426 L 0.443 E 0.409 Q 0.403 C 0.406P 0.388 E

Total ST 0.446 L 0.403 Q 0.436 Q 0.418 Q 0.438 E -ST seed production

0.407 L 0.338 L 0.370 Q 0.378 E 0.376 C -

June 2006

DSNTotal Biomass 0.656 Q 0.676 Q 0.650 P 0.663 P 0.642 C 0.549 E

Total ST 0.693 E 0.716 E 0.654 E 0.697 E 0.645 Q 0.482 E

ST seed production

0.581 E 0.603 E 0.539 Q 0.581 Q 0.537 E 0.440 E

DSSTotal Biomass 0.441 Q 0.443 Q 0.459 E 0.426 E 0.275 Q 0.181 Q

Total ST 0.476 L 0.472 L 0.356 Q 0.465 L 0.488 E 0.201 Q

ST seed production

0.407 L 0.421 L 0.344 Q 0.448 E 0.311 Q 0.202 Q

DSN/S - Ducky Strike North/South; ST is swamp timothy.Values shown are R2 for best-fitting regression lines for the three measured vegetation properties shown (g/m2) as a function of the vegetation indices (L linear; Q quadratic; C cubic; E exponential; P power law).

19

±

Swamp Timothy Seed Mass distribution at Strike Duck Wetland, Los Banos, CA in 2006

Seed Mass(g/m )2

Non Plant Area

DSN Boundary

DSS Boundary

0 - 30

30.1 - 60

60.1 - 90

90.1 - 120

120.1 - 150

150.1 - 330

0 160 32080Meters

Figure 3.B – Seed mass production estimated for Ducky Strike site based on multi-spectral image data using the simple ratio (SR) vegetative index.

20

Figure 3.A – Data and regression models for selected Ducky Strike South and North site measured vegetation properties and vegetative indices calculated from multi-spectral image data.

3.3 Summary

The goal of the vegetation change mapping using high resolution imagery was to provide a means of tracking the impacts of delayed drawdown on the experimental wetlands within the Grassland Ecological Area. In the project - aerial imagery flown in 2006 was used to develop methods for measuring both wetland moist soil plant areal extent and biomass using various vegetative indices. Interpretation of the 2006 imagery provided the following insights:

The vegetation classification schema developed in this project helped to confirm the potential for using high-resolution aerial photography to map community structure in the Grassland Ecological Area. The 2006 imagery point to the importance of timing in the flights. The May images yielded substantially less useful data than the June images because the vegetation had not yet sufficiently emerged in time for the earlier photos. However each year has a different hydrological signal and the best guide for choosing imagery data collection dates is to closely observe moist soil plant germination on the ground. Imagery from the June 2006 aerial survey revealed reasonable correlations between several vegetative indices and swamp timothy biomass in terms of total biomass and, more importantly from the perspective of waterfowl management, seed biomass.

4. USE OF ELECTROMAGNETIC BULK SALINITY MAPPING TO ASSESS IMPACTS OF A MODIFIED WETLAND HYDROLOGY

This section describes the methods and results from efforts aimed at characterizing drained wetland soil moisture content, temperature, and salinity in time and space. These properties are critical to moist soil plant ecology, and are impacted by hydrologic management practices. Key questions remain as to how these properties change on both short and long timescales. In this study we used an electromagnetic (EM) device to map soil salinity of ponds after drawdown.

4.1 Experimental Methods

For spatial soil mapping (section 4.1.1), two major EM campaigns were undertaken, the first in 2007 and the second in 2008. Maps were generated for all 12 wetland ponds in the study. As will be shown below, the first campaign yielded some results for soil salinity, which correlated poorly with manually collected soil salinity samples. The cause for this was most likely the relatively dry conditions under which the survey was undertaken. In 2008, the EM campaign was undertaken earlier and the resulting EM values correlated well with measured soil salinity values.

Two temporal soil moisture, temperature, and salinity assessments (section 4.1.2) were also undertaken, with an abbreviated (roughly 3-week) time series collected in spring 2007, and a long-term time series collected from before drawdown until after flood-up in 2008. The temporal data were collected at one site in several locations.

4.1.1 Electromagnetic (EM) Spatial Soil Mapping Methods (2007)

The 2007 spatial mapping process employed an EM-38 MK-1 ® (Geonics Ltd. Ontario, 21

Canada) along with calibration software, based on the DPPC (Dual Pathway Parallel Conductance) model developed by Rhoades et al. (1989). The EM-38 MK-1® utilizes dual coil electromagnetic induction to obtain soil salinity maps employing non-invasive methods. The EM-38, along with a backpack GPS system (Trimble, Sunnyvale, CA), was connected to a hand-held PC (Juniper Systems, Allegro Cx, Logan, UT). The logging was conducted using a program called TrackMaker ®. EM mapping has been proven to be effective and accurate in the prediction of soil salinity across vast landscapes in agricultural settings (Corwin and Lesch, 2003; 2005a; and 2005b, Isla et al., 2003, Lesch and Corwin 2003, Lesch et al., 2005; Cassel, 2007). However, its use in wetland settings has largely been unexplored. Wetland settings lack the uniformity and the homogenous nature of soils in agricultural fields, and therefore may require additional or different interpretative schemes. Factors to be considered include variations in soil texture and taxonomy, soil moisture, topography, vegetation and litter cover which all affect electromagnetic response (Hanson and Kaita, 1997; Suddeth et al. 2005; Brevic et al. 2006). The most significant factors determined by Corwin et al. (2003) in a westside San Joaquin Valley cotton field (Broadview Water District, Fresno County) were ECe, gravimetric water content, and texture.

The EM-38 can be used in two different orientations; vertically or horizontally. The maximum depths of the horizontal and vertical orientations, representing 75% of the response signal, are roughly 1m and 2m, respectively. The 75/25 response pattern was considered to be the maximum reading depth by McNeill et al. (1980) based on their theory and field trials. The peak signal strength for the horizontal and vertical orientations are between 0-0.3m and 0.3-0.6m respectively. Further EM-38 theory can be found in Geonics EM-38 Technical Notes #6 (McNeill et al., 1980).

For each pond surface, transects were paced by foot in parallel where tules and cattails allowed with 15m spacing between transects (Figure 4.1). The EM system was set to auto-sample every 2 seconds. At a walking speed of 7kph, that is roughly one sample every 4m along each transect. The device was suspended at a consistent 10cm height above ground. The dates of the 2007 and 2008 surveys conducted are summarized in Table 4.1. The output from the GPS and EM-38 was in xyz format.

Figure 4.1 – Researcher pacing parallel transects with EM-38 in drained wetland pond unit.

The ESAP software package was created by the USDA Salinity Laboratories to convert EM-38 xyz (apparent EC) data to actual electrical conductivity (EC). Within the program is a Response Surface Sampling Design (RSSD) that uses

22

the raw ECa xyz data to design a sampling strategy in order to optimally validate high and low EM response values. Presuming that the ECa data is normally distributed, sample locations are designated by the RSSD software based on standard deviations from the mean. In our case, samples were designated within a standard deviation of 3.5 (ESAP default) on each side of the mean. In all, 12 locations for each field were selected from which to collect soil samples for analysis.

Given that our primary objectives are to (1) create base-line soil salinity maps in order to assess changes in soil salinity due to modifications in wetland hydrology and (2) to develop a relationship between soil salinity and vegetative productivity, a depth of centered around 15cm was selected for soil sample collection. The 15cm depth of sampling was chosen to ensure that sample was mineral soil, to sample within the effective rooting zone of swamp timothy, but also at a shallow enough depth that changes in salt concentration due to variations in hydrologic management might be quantified. These samples were taken one month after the EM-38 survey had taken place, but before the next flood-up event.

An additional set of soil samples was collected during the EM-38 surveys. These samples were collected randomly in an attempt to sample the extremes of high and low moisture areas within each field. These samples were used to assess soil moisture at the time of survey. Because the ESAP-RSSD designated soil samples were collected one month after the EM-38 survey had been conducted, these samples were not used for moisture measurements since they did not represent the ambient soil moisture that was present at the time of the surveys. Instead, the random samples that were collected during the EM-38 surveys were used for soil moisture measurements. These moisture values were used to infer the moisture content of the samples designated by ESAP-RSSD. In instances where moisture values near the ESAP sampling point were not sampled, moisture content was estimated using a value from moisture sampling locations of similar topography and vegetation to the sample location in question.

Samples collected as prescribed by the RSSD software were analyzed for salinity in the lab (12 samples per pond, 12 ponds, 144 samples total). Each sample was crushed with a wooden rolling pin to break up aggregates and then passed through a 2mm sieve. None of the pebble fraction was crushed during processing. Qualitative notes of percentages of pebbles to soil, as well as the parent rocks were taken. For each sample, a fixed ratio of 15 grams of soil and 30 mL of deionized water were added to 50mL vials. The vials were mixed by hand to ensure that all of the soil was wet. Vials were then placed in a shaker for one hour. After shaking, samples were left upright overnight to allow the suspension to settle. The following day, samples were centrifuged at 3,000 rpm for 30 minutes. After 30 minutes, if the supernatant was not clear, samples were re-centrifuged. The conductivity (EC1:2) and pH of the supernatant was measured using a Myron Ultrameter II and values recorded.

The USDA Salinity Lab’s ESAP-Calibration software was used to convert the EM-38 response distribution (apparent EC: ECa) to an actual EC distribution (ECe) across the wetlands. The program performs a calibration based on an empirically fit regression model which employs the DPPC equation as developed by Rhoades (1989) and USDA Salinity Laboratory in Riverside, California.

4.1.2 Electromagnetic (EM) Spatial Soil Mapping Methods (2008)

23

In 2008 the timing of the salinity mapping campaign was shifted to capture the soil maps under times closer to the draw-down times based on the analysis of the 2007 salinity mapping data. In addition a new prototype EM38-MK2 was purchased by Lawrence Berkeley National Laboratory for use on the project. The MK2 device has two lengths of dipole separation: the traditional 1m-separation and a new 0.5m-separation. The 0.5m-separation is the equivalent to the 1m device operating in the horizontal orientation. With the MK2, both surface and sub-surface soil salinity can be measured simultaneously in the vertical orientation. The change in timing and the added EM instrument capability greatly enhanced the quality of data collected. Other than timing, EM-38 survey was conducted in the same fashion described above. The ESAP-RSSD program was used to determine the sampling strategy. For each field, 12 sampling locations were designated. At each location, soil samples were collected from two depths, 10-20cm and 45-55cm (12 ponds at 12 sites/pond and 2 samples/site making a total of 288 samples). The soil samples were taken immediately to the laboratory at UC Merced and a portion of the sample (roughly 250g) was oven dried at 105oC to determine gravimetric water content. The remaining portion of the sample was air dried in open plastic bags. When dry, samples were crushed using a soil crusher (Humboldt Mfg. Co., Schiller Park, IL), which pulverizes the soil aggregates leaving rock fraction largely intact. Qualitative notes of percentages of pebbles to soil, as well as the parent rocks were taken. For each sample, a fixed ratio of 15 grams of soil and 30 mL of deionized water were added to 50mL vials. The vials were mixed by hand to ensure that all of the soil was wet. Vials were then placed in a shaker for one hour. After shaking, samples were left upright overnight to allow the suspension to settle. The following day, samples were centrifuged at 3,000 rpm for 30 minutes. After 30 minutes, if supernatant was not clear, samples were re-centrifuged. The conductivity (EC1:2) and pH of the supernatant was measured using a Myron Ultrameter II and values recorded.

The USDA Salinity Lab’s ESAP-Calibration software was used to convert the EM-38 response distribution (ECa) to an actual EC distribution (ECe) across the pond areas. In addition to the salinity determination at all 12 study sites, the effect of soil moisture on the EM-38 MK2 signal strength was investigated in a spatio-temporal study at one of the study sites, Mud Slough 3b. Mud Slough 3b was initially mapped April 11, 2008. The spatio-temporal study commenced May 22, 2008 post drawdown of the summer irrigation. This study continued weekly with the last survey on June 23, 2008 totaling 6 iterations.

4.1.3 Soil Moisture, Temperature, and Salinity Sampling Methods

To provide a better understanding of the rate of change of soil moisture during wetland drawdown – since soil moisture is an important consideration for soil salinity mapping – additional moisture content, temperature, and salinity data were obtained by installing sensors (Model ECH2O-TE, Decagon Devices, Pullman, WA) at multiple levels in the soils underlying several ponds. Project budget constraints limited this portion of the study to one pond pair (Los Banos 31b and 33). The volumetric water content portion of this sensor operates by

24

Figure 4.2 – Schematic diagram of the soil pylon installation.

measuring the dielectric constant of the media through the utilization of frequency domain technology. Salinity was measured as electrical conductivity (EC) by measuring the current resulting from a fixed voltage applied through the media between two stainless steel screws on separate sensors prongs.

In 2007, the multilevel sensor systems, referred to as soil pylons, were tested at Gadwall for an abbreviated period post drawdown (March 15 to April 30, 2007). In 2008, the pylons were tested at Los Banos 31b and 33 for the entire draw-down to flood-up season (time March 3, 2008 to September 15, 2008). The sensor deployment in 2008 consisted of four stations, two per pond. Each station consisted of 3 decagon TE sensors connected to a decagon EM-50 logger. Deployment occurred while ponds were still flooded. A three-inch auger hole was dug and sensors placed at three depths, 15cm, 30cm, and 45cm. The logger was attached to 1.5m length, 5cm diameter PVC pipe which ensured that datalogger would be situated well above summer irrigation pond levels. Figure 4.2 shows the station design. Sensor stations were deployed through the entire summer and were removed post flood-up. The sensors logged data every 10 minutes.

4.2 Results and Discussion

4.2.1 EM salinity mapping results (2007)

The 2007 soil salinity mapping experiments yielded data that exhibited poor correlation between the lab-measured EC values and the EM-38 response. The best data obtained during the 2007 drawdown were from fields that were mapped earlier in the season when there was more soil moisture and therefore a better quality EM-38 signal response. The fields that exhibited the poorest relationships between EM-38 readings and field data were the fields that were mapped last. This finding suggests that the timing of these surveys was suboptimal with respect to soil moisture content in the field in 2007. Due to the lack of correlation between actual EC (EC1:2) and EM-38 response values (ECa) we determined that we could not obtain sufficient reliable regression models for prediction of field EC values. In spite of our inability to calibrate the EM results for the first season, the relatively salinity values were mapped and provided guidance for the 2008 mapping campaign.

There were several reasons for the 2007 calibration problems. First, our initial assumption that soil salinity within the upper soil profile was uniformly distributed was subsequently found to be incorrect. Through sampling with an auger, salt crusts were found at 60cm depth in many of our sites (Figure 4.3). Although not all of our sites are vertisols, they nearly all have vertic cracks and superactive shrink-swell clays and clay loams. As the soils crack, downwards beyond 60cm, the water wicks off the vertic faces. As the water evaporates, the salts are left behind resulting in salt crusts on the vertic faces at 60cm depth. As the soils are re-flooded in fall, the vertic cracks close and the salt crusts become encapsulated at depth creating a marbled like appearance. This phenomenon was not found in the surface horizons, only between depths of 50cm to 80cm; greater depths were not investigated. This finding suggests that our assumption of soil profile salinity uniformity is false. Hence the survey which used the EM-38 in the vertical orientation most likely did not accurately represent the near-surface soil salinity. This was an unforeseen factor contributing to the noise in our data.

25

In addition to the signal noise created from the significant salt nodule formation at depth, our method of determining EC with a fixed water/soil ratio (EC1:2) was also suspect. ESAP was created based on the traditional methods of soil EC measurements using the saturated soil extract (ECe) - this method mimics field capacity which accounts for minor textural differences between soil samples(Tanji 1990). However, this method is very time-consuming and can be difficult to perform in fine textured samples such as the heavy, 2:1, superactive clays such as those found in our study areas. It was suggested that some conversion factor be investigated to convert the EC1:2 to ECe, but that factor varies substantially with soil texture - without a texture analysis, it is hard to determine what which method is more accurate. Further investigation into this question is warranted.

Soil texture is one of the major factors in determining the quality of data not only in the calibration process but also during the EM-38 survey. As illustrated by the NRCS Soil Survey using soil series polygons soil texture varies substantially across the landscape. In some fields the textural differences between soil series’ are not substantial but may alter the EM-38 signal response just enough to distort the values. The depth to restrictive layers, bulk density, and horizontal textural differences also plays a role in the EM-38 signal response. For example, Salt Slough 24 (Figure 4.4) has apparent soil salinity values, which appears to follow the soil series delineation between the Alros clay loam to the West and the El Nido sandy loam to the east, where the signal values are much higher over the Alros than the signals over the El Nido. In the future, in such instances, it would be advisable to conduct two soil surveys - one for each soil type.

More significant than the effect of soil texture on the quality of EM-38 signal response is that of soil moisture content. Adequate moisture content important when measuring electrical conductance – without which the electromagnetic signal of the EM-38 deteriorates. After reviewing the literature associated with the application of this device, a unifying field condition was soil moisture content at or near field capacity (FC): the greater the moisture - the better the signal. Our salinity surveys during 2007 continued well into June when soil moisture content was significantly below field capacity. This fact alone could explain much of the noise in our data. In addition our samples were collected at 15cm and most likely did not represent the soil moisture content at 60cm, the theoretical EM-38 depth of observation in the vertical orientation.

26

Figure 4.3 – Marbling caused by salt crystals (white) in 60cm depth in Mud Slough 3b clay loam.

Another potentially important factor is that, for many soils, the upper foot of the profile accounts for less that 10% of the electromagnetic signal that is received in the vertical orientation (Scott Lesch, USSL, personal comm.). Even if one were to assume that the EC is consistent throughout the entire profile, the moisture gradient between the surface and subsurface soil is steep during late spring and early summer. By the middle of May, the upper foot of soil is practically desiccated whereas the subsurface, considering these soils are upwards of 50% smectite and soil moisture at these depths may still be near field capacity. There is currently no known method to account for this type of moisture gradient in analyzing EM-38 data.

In light of the effects of soil moisture on the EM-38 response,

why then was the correlation between EC1:2 and ECa decent in Ducky Strike South as well as Mud Slough 3B? When reviewing the dates of the surveys (Table 4.1) those two fields were the earliest surveys for the season, May 16 and May 14 respectively. This fact suggests that the majority of the remaining surveys, which were conducted later in the season had adequate soil moisture to provide representative EM-38 readings. For future efforts, it is suggested that the surveys are conducted the moment that the soils are dry enough to be walked on. In some cases, this could take well over a week, but fortunately, these tight clay soils don’t release water readily and still may be at or near FC.

Though the calibration failed to link the observed soil ECa values to the true EC distributions at the study sites, the ECa maps provide an informative qualitative view of the distribution and relative concentrations of salts across the field sites. Considering the calibration of the EM-38 was consistent between surveys, the ECa values for all fields can be compared on a parallel basis. Maps were produced for all 12 sites for the EM-38 generated ECa distributions using ordinary kriging in an effort to evaluate the spatial distribution of salinity qualitatively in spite the failure to obtain a working calibration for the survey.

One of the most interesting discoveries was made in Volta 23. A historic channel, now level, was discovered (Figure 4.5). The ECa map was overlaid with the 2007 NRCS soil survey map suggested a channeled Fluvaquent to the east, outside the field, coinciding with the low salinity band though the middle of the field. Three soil samples that were

27

taken from within the revealed historic channel, all had significantly lower EC1:2, lower pH, and a pebble fraction that was near 50% volumetrically. Three other samples had pebble fractions <10%, two being in proximity of the historic channel, and the other being in an isolated pool to the south of the managed portion of the field. This finding suggests that the soil used to fill and level this channel was from a different source, resulting in inherently lower salinity levels and/or that the lower apparent bulk density and coarse particle fraction allow a greater capability for leaching, removing the high salts and alkalinity from the upper 2 meters.

4.2.2 EM salinity mapping results (2008)

The data acquired during the 2008 salinity survey were a major improvement on the 2007 data. Our assumption that soil moisture was the most significant factor influencing EM-38 MK2 response quality proved to be correct. The relationships between EM-38 MK2 response values and lab tested EC1:2 were well correlated and allowed the calibration of the soil salinity maps to actual EC. Calibrated soil salinity maps representing surface soil salinity were created for all 12 of the study sites.

A key point with respect to analysis of “apparent” EC maps and the EC maps based on a salinity regression model is that, although the general separation of ECa and actual EC within a given field show similar patterns - the EM-38 MK2 signal strength is affected additionally by texture and soil moisture content. Hence, significant differences can occur between ECa and model-determined EC. This is illustrated in Figure 4.6 for Ducky Strike North (DSN) and South (DSS) which shows show typical differences between salinity maps developed for apparent ECa and regression model-determined EC.

28

29

Figure 4.6 – 2008 Salinity maps of Ducky Strike North and South ponds showing differences between apparent ECa (figure on left) and regression model-determined EC (figure on right).

).

30

Analysis of the 2007 data and apparent ECa maps of Salt Slough (SS) (Figure 4.4) suggested that soil texture had a large affect on EM-38 signal response. The 2008 (Figure 4.7) EC maps of Salt Slough (from the regression model) shows the opposite effect. In the case of the 2008 EC map of Salt Slough (north pond - SS 24) has the opposite response to that shown in the 2007 ECa map. Considering that the eastern portion of SS24 is a sandy loam and the western portion is a clay loam, it is now clear the effects of soil texture, as seen in the 2007 ECa map, was due to the residual soil moisture. The sandy loam will not hold the same amount of water as the clay loam and therefore, with lower overall salinity levels, and therefore have a lower EM-38 signal response. The 2008 EM-38 survey was conducted much earlier than the 2007 survey resulting in an overall uniformity of soil moisture. The opposite response of the 2007 EM-38 to the calibrated map of 2008 is most likely due to the differences in soil texture. The water table is near the surface most of the summer and subsequently, the saline ground water wicks upwards in the sandy loam by capillary action. As the ground water reaches the soil surface, it evaporates leaving the salts behind. The ground water does not wick up in the clay loam in the western portion of the field resulting in an overall lower salinity level.

Because soil moisture content at time of EM-38 survey is so crucial in acquiring quality EM data a temporal study was conducted to ascertain the effects of diminishing soil moisture to EM-38 signal response over a period of 5 weeks post irrigation at the MS3b site of the MHP study. Figure 4.8 displays the results of this study. The 1m and 0.5m values represent the distance of dipole separation within the EM-38 MK2 instrument where the 1m portrays sub-surface and 0.5m portrays surface salinity. As seen in Figure 4.8, the signal weakens considerably over the 5 weeks period. Soil samples were collected for each survey date according to ESAP RSSD software to measure soil moisture and soil salinity but have not yet been processed.

31

4.2.3 Soil Moisture, Temperature, and Salinity Temporal Results

The in situ soil sensors tracked moisture, temperature, and salinity levels locally the Los Banos (LB) and Mud Slough (MS) units through the entire summer season. The results for these ponds were similar and the LB results are shown for a depth of 45 cm below ground surface (Figure 4.9). From the March drawdown to the summer irrigation event, as the soil moisture content decreased, the soil salinity increased. The sensors responded rapidly to the summer irrigation event in late May. The rapid increase in salinity during this event suggests that some salt precipitates may have been re-dissolved at this time. Once the irrigation water was drawndown, the soil temperature again tends to increase steadily with summer air temperatures and solar radiation.

An interesting result was the consistent decrease in the observed soil salinity after the summer irrigation event. This may be an artifact of insufficient soil moisture for accurate EC response by the in situ sensors. A similar trend was observed in the temporal EM-38 study at MS3b (Figure 4.8). The limitations of both the in situ and EM-38 salinity sensors with respect to measuring soil salinity at reduced moisture levels is a topic that warrants further investigation.

The soil moisture and soil temperature sensor readings appeared to operate reasonably at the lower moisture levels. These data suggest that soil networks, if employed across a wildland setting, may provide pertinent decision support to land managers and help them in making management decisions.

Figure 4.9 – In situ time series for one sensor location (Los Banos 31B South, 45 cm depth) for (a) volumetric water content (VWC), (b) salinity as EC (dS/m), and (c) temperature (C).

32

4.3 Summary

Over the course of two seasonal campaigns, we refined electromagnetic soil salinity mapping procedure, enabling it to be employed more quantitatively in wildland soils such as those underlying Central California managed wetland ponds. In advancing the EM techniques, valuable lessons were learned in support of subsequent salinity characterization efforts.

First, and in accord with Lesch and Corwin (2003), a primary criterion for successful EM surveying is to minimize soil material and status variability across the landscape as much as possible, whether spatial (i.e., minimizing texture variations) or temporally (minimizing soil moisture variability). While this is typically reasonable to accomplish in agronomic systems, it is less controllable in wildland soils. In this work, soil moisture content and textural variations across the landscape seem to be the most dominant variables controlling the quality of EM-38 survey data. For the most part, our study sites offer similar soil parameters, outside of the sandy loam of the Salt Slough sites; therefore, minimization of variability temporally is recommended as a means of producing the desired results.

Second, the best time to conduct soil salinity surveys in the systems under consideration in this study is directly after the initial draw-down allowing for uniform soil moisture content both vertically in the soil profile and also spatially across the landscape. Unfortunately, due to the topographic and spatial variations between the fields, some drain much faster and some areas within the fields don’t drain entirely. Surveys should take place as soon as the majority (>90%) of the field has been drained and as soon as the soils are dry enough to walk on without sinking. Temporal data from key location can be used to inform the decision about when to perform salinity mapping. It is ideal to survey on the initial drawdown cycle and not to rely on the summer irrigations for adequate soil moisture content. The summer irrigations are only held for a week at most and it is uncertain that with the heavy clay soils that moisture percolates very deep. As well, the temperatures during the irrigations are much higher and, upon drawdown, the fields dry out differently. In some of our sites, not all fields were inundated by the irrigation resulting in large dry spots surrounded by soils with 40% moisture content. Careful considerations should be made reflecting on the circumstances and condition of the fields before the time to survey takes place.

With the addition of the EM-38 MK2, the surveying quality and reliability of data improved dramatically. This device, along with soil sampling at two depths in the soil profile, ultimately allowed for the successful prediction of surface soil salinity. This device also helped to overcome the discovery of massive salt marbling in the sub-surface soil profile.

Finally, the time series data on moisture, temperature and salinity can provide data for creating temporal interpolations between the spatial mapping times. Because the spatial maps require significant effort in both collecting and analyzing the data, having the time series data is important. Additional work to determine the best scheme for fusing these two types of data is clearly warranted.

33

5. SUMMARY, CONCLUSIONS, AND APPLICABILITY

This UC Salinity Drainage Program research project was a two-year, interdisciplinary study of potential impacts to wetland moist soil plant habitat resulting from modification of the scheduling of seasonal wetland drawdown within the San Joaquin River Basin. Seasonal wetland drainage contributes salt loading to the San Joaquin River – changing the timing of these wetland contributions to the River, if part of a comprehensive, basin-wide real-time water quality management system, can improve compliance with State salinity objectives. However, such modifications may also adversely impact moist soil plant communities in these managed wetlands, the main focus of this project.

This research project centered around six pairs of seasonal wetland ponds, ranging from 20 to 100 acres in size. These were selected from amongst the State and private managed wetlands within the 170,000-acre Grasslands Ecological Area. Each pair of matched sites allows traditional wetland drawdown management practices to be compared to those where drawdown is delayed until April 15 each year – with respect to impacts on water quality, wetland soils and moist soil plant habitat.

The project approach utilized a combination of remote sensing (multi-spectral aerial imagery) and on-the-ground sensing (water quality monitoring, electromagnetic (EM) spatial salinity mapping, and local temporal monitoring of soil salinity). The following is a list of project accomplishments and findings to date:

The project team assisted in the deployment of 24 telemetered (radio and cellular modem) flow and water quality monitoring stations measuring continuous electrical conductivity, temperature, and stage were established at the inlet and outlet of the six paired wetland pond sites. The stations now measure salt fluxes in and out of each pond.

Three sets of high-resolution multi-spectral images were provided to the project which included a pre-treatment in 2006 and post-treatment images for 2007 and 2008). Images for 2006 have been processed and analyzed, while those for 2007 and 2008 will be completed in early 2009. We have developed techniques using plant association-specific spectral signatures to identify 29 of the most important wetland plant associations. Over 500 ground truth locations were sampled to verify the accuracy of the technique. Of the 29 signatures, 9 were moist-soil plant associations that included Crypsis schoenoides (swamp timothy), which is a dominant moist-soil plant in the Grasslands Ecologic Area. With the spectral signatures, classification was performed to estimate areas of swamp timothy presence and absence across the entire study area.

In spite of suboptimal timing of the 2006 images, several pond sites exhibited strong correlation between several spectral signatures and plant productivity in terms of total above-ground biomass and seed production have been identified using multi-spectral data and ground truth based on vegetative sampling. Given improved flight timing, the 2007 and 2008 results are expected to yield improved results.

High-resolution soil salinity maps were created using a Geonics electromagnetic field instrument (EM-38). The EM-38 produces a relative soil salinity map that must then be calibrated to salinity values measured from physical samples. Twelve soil samples per pond were used to calibrate the maps. Results from 2007 were of limited value scientifically due to the timing of the data collection, but served as guidance for the 2008 campaign, which resulted in

34

good correlation between EM results (using the next generation EM-38 MK2) and ground-truth samples and ultimately reliable salinity maps for the pond sites.

The time series data on moisture, temperature and salinity can provide data for creating temporal interpolations between the spatial mapping times. Because the spatial maps require significant effort in both collecting and analyzing the data, having the time series data is important. Additional work to determine the best scheme for fusing these two types of data is clearly warranted.

Methods have been developed for correlating reflectance spectra from aerial imagery with manual swamp timothy productivity survey data was completed for 2007 and are currently being applied to the 2008 data sets. These results are also being compared with the EM-based salinity maps, with salinity-moisture time series data being used to account for any significant time differences in the survey maps. The integrated results will enable us to provide significant input on the question of how long modified wetland management practices be sustained, that are designed to improve water quality in the San Joaquin River, without negatively impacting the biological value of waterfowl habitat.

Supplemental funding was acquired through a State Water Resources Control Board Grant (PI Quinn) and a California Department of Water Resources Proposition 204 grant (PI Quinn, co-PI Harmon) to develop mathematical models of seasonal wetland hydrology and to improve the representation of these wetlands in the current WARMF-SJR model.

In terms of professional development, Patrick Rahilly completed his M.S. in Environmental Systems at UC Merced based on this project, and the project led to ongoing collaborative interactions between the UC project team and the California Department of Fish and Game and California Department of Water Resources.

The work completed under this UC Salinity Drainage Program award yielded results of interest to both wetland managers and broader-based agencies. With respect to the wetlands, the methods and results developed here are the beginnings of a combined remote and ground-based sensing approach to providing management decisions support in terms of hydrologic manipulation of these ecosystems. Understanding the effects of drawdown timing on wetland plant community structure is vital to maintaining the function of this ecosystem, especially with respect to providing waterfowl habit.

The remote sensing products developed in this work suggest the need to regularly contract aerial imagery. This is clearly undesirable with respect to both cost and timing issues. Fortunately, the techniques developed here are transferable to the analysis of satellite imagery, and these products are expected to improve in terms of resolution and the availability of additional spectral bands in the near future.

In a broader context, the methods and results from this project provide a framework for salinity management in Central California wetlands and similar systems in the San Joaquin River and Bay-Delta regions. Managed water bodies that are appropriately instrumented at Internet-enabled sites, such as the inlets and outlets of wetland ponds in this study, can be integrated into larger-scale monitoring networks (e.g., California Digital Exchange Center). The remote and ground-based sensing products can then be coupled to the knowledge of the assimilative capacity of receiving water bodies. With appropriate analytical tools, including simulation and multi-objective resource optimization models, an adaptive management program based on real-time water quality conditions can be developed.

35

REFERENCES CITED

Brevik, Eric C., Thomas E. Fenton, Andreas Lazari. 2006. Soil electrical conductivity as a function of soil water content and implications for soil mapping. Precision Agriculture 7: 393-404.

Bundy, R.M. 1997. Hydrology influences on vegetation response in a managed moist-soil impoundment. M.S. Thesis, University of Missouri-Columbia.

Cassel, Florence S. 2007. Soil Salinity Mapping using ArcGIS. In GIS Application in Agriculture, Chapter 8, pg 141-162. Francis J. Pierce and David Clay (eds.) CRC Press, New York.

Cogswell, H.L. 1977. Water Birds of California. California Natural History Guides: 40. University of California Press. 399 pages.

Corwin, D.L. and S.M. Lesch. 2003. Application of Soil Electrical Conductivity to Precision Agriculture: Theory, Principles, and Guidelines. Agronomy Journal. 95: 3.

Corwin, D.L., Lesch, S.M., Shouse, P.J., Soppe, R., Ayars, J.E., 2003. Identifying soil properties that influence cotton yield using soil sampling directed by apparent soil electrical conductivity. Agronomy Journal 95 (2): 352–364.

Corwin, D.L. and S.M. Lesch. 2005a. Characterizing soil spatial variability with apparent soil electrical conductivity. Computers and Electronics in Agriculture 46:103-133.

Corwin, D.L. and S.M. Lesch. 2005b. Apparent soil electrical conductivity measurements in agriculture. Computers and Electronics in Agriculture 46: 11-43.

Fredrickson, L.H., Taylor T.S. 1982. Management of seasonally flooded impoundments for wildlife. Resource Publication 148. U.S. Dept. of the Interior, Fish and Wildlife Service, Washington, D.C.

Grober, L.F., Karkoski J., Poole T. 1995. Water quality impact of wetlands on San Joaquin River, California, paper no. 00149 In T.G. Cleveland [ed.], Advances in the development and use of models in water resources: Proceedings of the American Water Resources Association held in Houston, Texas, November 5-10, 1995. Department of Civil and Environmental Engineering, University of Houston, Houston, Texas.

Hanson, B.R., and K. Kaita. 1997. Response of Electromagnetic Meter to Soil Salinity and Soil-Water Content. Journal of Irrigation and Drainage Engineering March/April: 141-143.

36

Isla, R., R. Aragues, and A. Royo. 2003. Spatial variability of salt-affected soils in the middle Ebro Valley (Spain) and implications in plant breeding for increase productivity. Euphytica 134: 325-334.

Lesch, S.M., D.L. Corwin, D.A. Robinson. 2005. Apparent soil electrical conductivity mapping as an agricultural management tool in arid zone soils. Computers and Electronics in Agriculture 46: 351-378.

Lesch, S.M., and D.L. Corwin. 2003. Using the Dual-Pathway Parallel Conductance Model to determine how different soil properties influence conductivity survey data. Agronomy Journal 95, 395-379.

Mason, H.L. 1969. A Flora of the Marshes of California. University of California Press. 879 pp.

McNeill, J.D., 1980. Electromagnetic Terrain Conductivity Measurement at Low Induction Numbers. Tech. Note TN-6. Geonics Ltd., Ontario, Canada.

Mushet, D.M., Euliss, Jr., N.H., Harris, S.W. 1992. Effects of irrigation on seed production and vegetative characteristics of four moist-soil plants on impounded wetlands in California. Wetlands 12, 3.

Rhoades, J.D., N.A. Manteghi, P.J. Shouse, and W.J. Alves. 1989. Soil electrical conductivity and soil salinity: New formulations and calibrations. Soil Science Society of America Journal 53: 433-463.

Quinn N.W.T. and W.M. Hanna. 2003. A Decision Support System for Adaptive Real-Time Management of Seasonal Wetlands in California. 2003. Environmental Modelling and Software. Vol. 18, Issue 6, pp 503-511. Elsevier Science Ltd. (LBNL-50238)

Quinn N.W.T. and W. M. Hanna. 2002. Real-Time Adaptive Management of Seasonal Wetlands to Improve Water Quality in the San Joaquin River. Advances in Environmental Research. Vol. 5, pp 309-317. Elsevier Science Ltd.

Quinn, N.W.T., Chen, C.W., Grober, L.F., Kipps, J., Cummings, E. 1997. Computer model improves real-time management of water quality. California Agriculture, 51, 5.

Quinn, N.W.T., Karkoski, J. 1998. Real-time management of water quality in the San Joaquin River Basin, California. American Water Resources Association 34, 6.

Smith, W.D., Rollins, G.L., Shinn, R.L. 1995. A Guide to Wetland Habitat Management in the Central Valley. California Dept. of Fish and Game, California Waterfowl Assoc., Sacramento, CA. 34 pp.

37

Stoddard & Associates, 1998. Water management plan for Grassland Water District. Report for the Grassland Water District, Los Banos, CA. 72 pp.

Swanson, G.A. 1988. Aquatic Habitats of Breeding Waterfowl. In: Hook et al. (Eds.) Ecology and Management of Wetlands, Vol. 1: Ecology of Wetlands. Timber Press, Portland, Oregon. 592 pp.

Suddeth, K.A., N.R. Kitchen, W.J. Wiebold, W.K. Batchelor, G.A. Bollero, D.G. Bullock, D.E. Clay, H.L. Palm, F.J. Pierce, R.T. Schuler, and K.D. Thelen. 2005. Relating apparent electrical conductivity to soil properties across the north-central USA. Computers and Electronics in Agriculture 46: 1-3, 263-283.

38