Location: Northwest Irrigation and Soils Research
2023 Annual Report
Objectives
The research in this project includes a series of studies conducted under three broad objectives of improving water use efficiency and water quality in irrigated crop production. Water use efficiency research focuses on a variety of crops and conditions that occur in the northwestern U.S. Much of the water quality research focuses on the Upper Snake Rock (USR) watershed which is part of the ARS Conservation Effects Assessment Project (CEAP).
Objective 1: Characterize plant-climate-management interactions to optimize water productivity in intensively irrigated systems.
Subobjective 1A. Determine the effect of barley cultivar type (food and malt) on water use efficiency under full and deficit irrigation.
Subobjective 1B. Quantify the relationships between irrigation level and barley grain and straw yields under optimum and sub-optimum N supplies.
Subobjective 1C. Evaluate crop water use and agronomic response of sorghum-sudangrass hybrids under multiple management practices.
Subobjective 1D. Develop an IoT canopy temperature measurement system for crop stress monitoring.
Objective 2: Clarify climate and management impacts on water quantity and quality at the field edge and beyond the vadose zone.
Subobjective 2A. Evaluate the impact of tillage, cover crop, and fertilization management on surface and groundwater processes under linear-move irrigation system.
Subobjective 2B. Evaluate subsurface water quality dynamics under sprinkler and furrow irrigation at the field scale.
Subobjective 2C. Develop a furrow irrigation-induced erosion prediction tool.
Subobjective 2D. Develop a soil parameterization database for a center pivot infiltration model.
Subobjective 2E. Evaluate the impact of variable soil depth on water balance and nutrient leaching.
Objective 3: Identify environmental conditions and adaptive management strategies that improve water quality in surface and subsurface drainage networks in irrigated landscapes.
Sub-objective 3A: Develop a machine learning technique to detect and map in-field irrigation methods.
Sub-objective 3B: Evaluate the effect of long-term changes in irrigation methods and interannual variations in crop area on water availability and quality.
Sub-objective 3C: Evaluate the SWAT model for highly managed irrigated watersheds of the Northwest.
Sub-objective 3D: Determine P sorption capacity and equilibrium P concentration (EPC0) for a range of agricultural and canal soil/sediments in Idaho.
Approach
This project involves a combination of experimental field studies, watershed studies, model developments and tool development. The overall objective to optimize water productivity in intensively irrigated systems will be achieved through four field studies. A three-year study will measure the response of two barley cultivars to four irrigation levels ranging from full irrigation to 25% of full irrigation. A second study will further clarify the interrelation between irrigation level and barley straw and grain yield under optimum and sub-optimum nitrogen. This study will provide valuable data on evapotranspiration requirements under a variety of barley management scenarios. A third study will evaluate the performance and the viability of sorghum-sudangrass as an alternative forage crop while a fourth study will apply state-of-the-art sensing and wireless networking technologies (Internet-of-Things) to develop a canopy temperature measurement system to monitor crop water stress. This system will provide a practical canopy temperature measurement platform for the application of the crop water stress index (CWSI) to manage deficit irrigation for a wine grape cultivar.
In a second objective cover crop and no-till practices will be evaluated to devise sustainable and climate-resilient management systems by monitoring runoff, erosion, infiltration, soil water content and surface and groundwater quality on experimental fields. Another study will compare furrow irrigation to sprinkler irrigation on surface and subsurface water quantity and quality using field-installed lysimeters, soil moisture sensors and runoff measurements. A third study will develop a furrow irrigation soil erosion model by contrasting a machine learning approach with a process-based approach to erosion prediction in eroding furrows. In a fourth study, a database to parameterize an infiltration model for center pivot irrigation will be developed to more accurately account for surface sealing in infiltration prediction. A fifth study will apply a combination of field monitoring of soil processes, in-situ remote sensing, cutting-edge imaging and 3D reconstruction technologies including Ground Penetrating Radar to model soil depth and its impact on water and nutrient dynamics.
The third objective will be accomplished through watershed research. Irrigation water diverted into the 82,000 ha Upper Snake Rock watershed and water returning to the Snake River will be monitored for water quantity and quality to determine water, sediment, and nutrient balances for the watershed. Watershed research will evaluate potential associations between the extent of specific crops in a watershed and water quality outcomes. A methodology will be developed by applying deep learning techniques and computer vision to map the types of irrigation used on agricultural fields of the watershed. One study will parameterize the SWAT model for highly managed irrigated areas and evaluate improved irrigation routines developed for this model. A final study will use carefully designed benthic sediment sampling strategies to determine equilibrium phosphorus concentration in irrigation return flow drainage systems.
Progress Report
Crop production studies for Objective 1 are ongoing or have been completed. The first year of a three-year study was completed comparing water production functions of malt and food barley. Initial results indicate that irrigation can be reduced about 25 percent (%) without significant yield loss, although protein will likely increase, which can be positive for food barley and negative for malting barley. The first year of another three-year study was completed. This study is determining barley yield for different irrigation amounts with optimum and sub-optimum nitrogen supplies. This data will help improve irrigation management under full and deficit irrigation scenarios under high and low yield conditions. Field studies have been completed to determine planting strategies and water use for sorghum-sudangrass in southern Idaho. Crop yield was greatest when sorghum-sudangrass was planted in seven-inch wide rows and harvested once during the growing season compared to wider row spacing (30 inch) or two harvests per season. Irrigation water requirements and water productivity are being calculated. Sugar beet and corn water use data were summarized for sugar beet and dairy industries in Idaho. Relationships between crop evapotranspiration and sugar beet and corn yields were determined to assess mitigation options for potential irrigation water shortages.
Two ongoing field studies and two modeling efforts are being conducted for Objective 2. A second year of data was collected from a study evaluating tillage and cover crop impacts on runoff and water quality under sprinkler irrigation. A custom wireless sensor network system has been developed to monitor soil water potential and soil temperature at the 12 experimental plots, resulting in 72 sensors being logged every 30 minutes. Runoff and surface and subsurface water quality data are also collected during irrigation and natural rainfall events. A SQLite database has been developed to combine all the data. Early data suggest an anecdotal benefit of cover crop and no-tillage treatments. A fourth year of data was collected from the lysimeter study to compare leaching under sprinkler and furrow irrigation. Leachate volume is highly variable and tends to be greater with furrow irrigation early in the growing season.
Additional research is continuing to improve a furrow irrigation erosion prediction model. A data set containing over 2000 measurements of furrow irrigation sediment loss in the western U.S. was used to develop a data driven model to estimate sediment loss using eight common hydraulic and field condition variables. The unsupervised machine learning technique Kohonen self-organizing maps (KSOM) was initially used to predict furrow sediment loss but the inherent variability in measured furrow sediment loss limited the ability of a KSOM model to provide reliable predictions. A transfer learning approach was used to increase prediction performance of the model. Additional improvement of the transfer learning approach have increased the coefficient of determination between measure and predicted values to 0.83 and average predicted furrow sediment loss is 16% less than measured sediment loss.
Multiple field soils were subjected to different levels of droplet kinetic energy flux (specific power) in the laboratory using a Cornell sprinkle infiltrometer. The data from these tests will be used to calibrate a center pivot infiltration model for various soil textures. Infiltration rate was the greatest when specific power was zero and decreased as the level of specific power increased. Infiltration rate response to uniform, increasing or decreasing application rates were investigated. The was little difference in infiltrated volume between different water application rate patterns when the same total amount of water applied with the same total specific power, indicating that cumulative droplet kinetic energy applied to the soil surface largely determines infiltrated volume of water, regardless of application pattern.
Also in support of Objective 2, cooperative research with University of Nevada, Reno, and University of Idaho, used ground penetrating radar on a research farm to detect variations in soil depth. Seven transects were measured on a 160-acre field with variable depth to bedrock. Ground penetrating radar information will be compared with soil cores that were previously collected from the field.
For Objective 3, water quality monitoring continued at 30 sites in the Upper Snake/Rock watershed for the Conservation Effects Assessment Project (CEAP). Monitoring focuses on irrigation return flow from the Twin Falls Canal Company (TFCC) irrigation project. Recent data analysis showed a significant decrease in dissolved phosphorus concentrations after a large snowmelt and flooding event in Spring 2017. Data for the TFCC irrigation project have been analyzed to evaluate the effect of management practices on quality and quantity of irrigation return flow. Overall, analyses of data from 2006 to 2021 showed return flow volumes increasing at two of seven sub-watersheds while average phosphorus concentration and nitrate concentration showed decreasing trends at some sites. A combination of changes in irrigation methods and the adoption of other best management practices explain these long-term improvements in water quality. Research for the Legacy Phosphorus special project under CEAP has focused on collecting field soil samples for common analysis, compiling data for simulations with the Annual Phosphorus Loss Estimator (APLE) model and initiating comprehensive sampling of several water quality ponds. Three sets of channel sediment samples were also collected for phosphorus sorption studies.
A deep learning model using a U-Net architecture has been developed to map irrigation methods in agricultural areas of Utah and within the TFCC irrigation project. Data from the Utah Water Related Land Use (WRLU) dataset for 2006 to 2021 provided irrigation method information for all agricultural areas in the state of Utah. Irrigation methods were classified as Flood, Sprinkler or Other (drip and non-irrigated). Ten models were trained on the WRLU dataset supplemented with 235 additional polygons in TFCC within which irrigation methods were manually labelled for years 2006 to 2021 with available aerial imagery. Model inputs were derived from publicly available Landsat 5 and Landsat 8 satellite imagery. Training was done on the SCINET Ceres High Performance Computing system and the 10 models averaged into an ensemble model. The overall accuracy of the ensemble model was 78%. Precision achieved were 73%, 82% and 80% for the Flood, Sprinkler and Other classes, respectively. Codes used to construct model inputs, train models and develop prediction maps have been posted on a publicly available GitHub library. Step-by-step directions and other support documentation are currently being developed to be posted on the GitHub page. Irrigation method maps have been developed for the TFCC irrigation project for 2006 to 2021 using the deep learning method. The proportion of sprinkler increased at a rate of 1.4% per year. In-depth analyses at the sub-watershed scale will link return flow water quality changes through time to the proportion of sprinklers in a sub-watershed.
In collaboration with researchers at the University of Idaho, the Soil and Water Assessment Tool (SWAT) is being used to simulate hydrologic processes in sub-watersheds of the TFCC irrigation project. The SWAT model was successfully set up for one sub-watershed in the TFCC irrigation project, however, parameterizing SWAT to accurately represent the timing of irrigation has proven more difficult than anticipated because irrigation water supply is limited. Additional evaluations of the SWAT model on field-scale data are being conducted.
Accomplishments
Review Publications
King, B.A., Shellie, K.C. 2023. A crop water stress index based internet of things decision support system for precision irrigation of wine grape. Smart Agricultural Technology. 4. Article 100202. https://doi.org/10.1016/j.atech.2023.100202.
Pan, P., Qi, Z., Koehn, A.C., Leytem, A.B., Bjorneberg, D.L., Ma, L. 2023. Modification of the RZWQM2-P model to simulate labile and total phosphorus in an irrigated and manure-amended cropland soil. Computers and Electronics in Agriculture. 206. Article 107672. https://doi.org/10.1016/j.compag.2023.107672.
