Location: Water Management and Systems Research2015 Annual Report
1a. Objectives (from AD-416):
Objective 1. Develop and apply new watershed modeling tools to evaluate the long-term effects of innovative cropping, limited water, and nitrogen management on water quantity, water quality, and crop production in selected agricultural sub-basins in Colorado. [Contributes to Problem Area #1, Effective Water Management in Agriculture, Problem Statements 1.1.3 and 1.4.2 of the new National Program (NP) 211 Action Plan (FY 2011 – 2015)] Objective 2. Using data from Colorado and the Midwest, improve model components to quantify and assess spatially targeted agricultural conservation effects on water quantity and quality. [Contributes to Problem Area #4, Improving Watershed Management and Ecosystem Services in Agricultural Landscapes, Problem Statement 4.1 of the new National Program (NP) 211 Action Plan (FY 2011 – 2015)] Objective 3. Simulate the combined effects of projected climate change on crop production, water use, and nitrate transport, and assess potential cropping system adaptations at field to sub-basin scales in Colorado. [Contributes to Problem Area #4, Improving Watershed Management and Ecosystem Services in Agricultural Landscapes, Problem Statement 4.3 of the new National Program (NP) 211 Action Plan (FY 2011 – 2015)]
1b. Approach (from AD-416):
As population increases and climate changes, we face global crises of conserving and managing water quantity and quality for agricultural and urban demands. Distributed agro-hydrologic modeling tools are needed to address complex system interactions in space and time for different soils and climates. Impacts of water and nutrient management and of targeted conservation practices within and adjacent to fields must be assessed in terms of water quantity and quality at designated watershed outlets. This project focuses on developing simulation tools for evaluating and proposing solutions to critical emerging problems in diverse agricultural systems over scales ranging from approximately 50 to 50,000 ha under current and future conditions. The component-based AgroEcoSystem-Watershed (AgES-W) model, developed in the Object Modeling System (OMS) framework, explicitly simulates the hydrologic and agronomic responses from spatially distributed land use, management, and weather conditions across inter-connected ecosystem response units (ERUs). AgES-W will be enhanced for: 1) routing water and nutrients across a watershed, 2) diverse cropping system responses to water deficits, 3) model uncertainty analyses and scaling, and 4) plant responses to atmospheric CO2. New OMS tools will include ERU delineation, sensitivity analysis, spatial visualization, statistical analyses of outputs, and web-based cloud computing. Selected conservation practices will be evaluated under existing and projected climates in the semi-arid West (Colorado), and spatially targeted conservation will focus on the sub-humid Midwest (Iowa), resulting in new agricultural adaptation strategies. These case studies address agricultural water and nutrient management issues in the American West and Midwest, while providing component-based modeling tools globally.
3. Progress Report:
Most of the progress made in this project relates to development and improvement of AgroEcoSystem-Watershed (AgES-W) model, which is used to address all three project objectives. Thus the progress below covers model development and testing first, followed by applications to Objectives 1 and 2. None of the current year milestones applied AgES-W to Objective 3 (impacts of climate change). ARS researchers in Fort Collins, CO and university collaborators comprising the AgES-W development team: 1) developed new components to regionalize climate inputs (i.e., precipitation, minimum/maximum temperature, solar radiation, wind speed) prior to the start of model simulations; 2) completed testing and evaluation of science components for irrigation scheduling, nitrogen (N) dynamics, and for calculating the effect of changes in soil physical properties by tillage, reconsolidation, or other practices; 3) improved and enhanced Object Modeling System core modules for AgES-W component connectivity, and 4) developed new science modules (via extraction of code from other agroecosystem models) for conservation effects, hillslope soil erosion, Green-Ampt infiltration, and NRCS curve number hydrology. AgES-W auxiliary tools were updated and enhanced. Testing of a semi-automated watershed delineation tool was completed and ArcObject code modifications were implemented to improve program operation (i.e., ease of use) under ArcGIS. The tool was implemented to delineate Hydrologic Response Units (HRUs) at different scales within the 800 km2 South Fork Watershed in Iowa and a 56 ha (140 acre) watershed in eastern Colorado. Field boundaries and hydrogeology were the primary geospatial inputs used to delineate HRUs in Iowa, while topography, management units and soil types were used in Colorado. The Model Optimization, Uncertainty, and SEnsitivity Analysis (MOUSE) software, an open-source, Java-based toolbox of visual and numerical analysis components, was enhanced through the addition of new single- and multi-objective optimization algorithms and a Bayesian Monte Carlo approach to quantify model uncertainty. AgES-W model integration within the Geospatial Modeling Interface (GMI) was upgraded through the addition of new features for geospatial visualization (i.e., color ramping) of model input data and expanded options for controlling AgES-W output variable selection and display. In addition, the new modular version of the Soil and Water Assessment Tool (SWAT) model was added to the suite of models currently operating under GMI. The AgES-W model was restructured and enhanced to enable parallel processing for surface and subsurface flow routing components. The state-of-the-art parallel processing components allow simultaneous simulation of multiple Hydrologic Response Units (HRUs) and stream reaches as long as spatial flow routing dependencies have been satisfied. Methods added for parallelizing surface and subsurface processes included parallelization by dependency layer, sorting order, and disjoint subtrees. Each of these has trade-offs in computer processor synchronization cost, added complexity of the AgES-W code, and the “breadth” of parallelism that may be achieved. In addition to implementation of parallel processing methods, two different methods for distributing the implementation of AgES-W were developed and are undergoing testing. These methods include distributing the computational work of surface and subsurface processes across a cluster of virtual machines and distributing required AgES-W input data across the entire computing cluster. The three parallelization methods reduce AgES-W simulation run times by a factor of 2-3x and are also applicable to other spatially distributed modeling technologies. The Unified Plant Growth Model (UPGM) was enhanced by: 1) development of a Shuttleworth-Wallace component as an alternative to the Penman-Monteith component for estimating evapotranspiration; 2) incorporating the responses of stomatal conductance and photosynthesis algorithms to atmospheric CO2 concentration; 3) analyzing genes/markers for influencing winter wheat phenology identified from experimental data collected for 300 entries under varying water deficits, thereby allowing for better simulation of Genetics by Environmental interactions; and 4) completing initial seed germination studies for improving the seedling emergence component. UPGM technical documentation has been completed. The enhanced UPGM has been integrated into the AgES-W model and further evaluation is being conducted. The AgES-W model was deployed as a cloud computing application (see Accomplishment 2) to explore relationships between the size of simulated sub-areas, calibration approaches, resulting spatially variable and averaged parameter values, and the model’s ability to replicate measured soil moisture. An instrumented field site in Colorado, which has been monitored for over a decade in cooperation with a farmer, was simulated as a small (140 acres or 56 hectares) watershed. Soil moisture was simulated at four depths over 16 landscape positions under winter wheat and fallow conditions. The watershed was sub-divided into simulation areas based on management strips, soil type, and probe locations to delineate the watershed at three different resolutions. Calibration using five different schemes for vertical and horizontal variation of parameter values provided comparisons of the effects of homogenizing or lumping parameters as done in other large watershed models. AgES-W can capture agricultural and watershed processes at the scales of management to reveal spatial variations within a farm field (in this case, or larger watersheds) due to management and natural soil variability. (Objective 1) Using measured water quantity/quality data collected from the South Fork Watershed in Iowa, AgES-W was evaluated and calibrated to simulate hydrologic (streamflow) fluxes, nitrate fluxes, and sediment transport at the watershed outlet. A combination manual and autocalibration (using an OMS3-based calibration tool called LUCA) approach was used to calibrate sensitive AgES-W parameters for streamflow, nitrate, and sediment transport model output responses. Evaluation of observed versus AgES-W predicted streamflow resulted in statistical evaluation criteria (i.e., Nash-Sutcliffe model efficiency) that are among the best found in literature for hydrologic/water quality (H/WQ) watershed model evaluation. (Objective 2)
1. Landuse and Agricultural Management Practices web-Service (LAMPS). Spatial models of watershed processes and conservation planning tools require crop rotation and management information for specified areas that are tedious to compile from available sources. ARS researchers in Fort Collins, CO collaborated with Colorado State University and NRCS partners to develop LAMPS, which automatically links three data sources: 1) annual Cropland Data Layers from the CropScape web service based on high-resolution remote sensing data provided by the National Agricultural Statistical Service, 2) maps of irrigated areas compiled by the U.S. Geological Survey, and 3) the Land Management and Operation Database (LMOD) by NRCS based on representative crop rotations. Annual crop sequences from CropScape are matched to regional crop rotations in LMOD using a similarity index developed to match genetic sequences. Information for deploying LAMPS is available at http://javaforge.com/wiki/4077903. ARS researchers in Oxford, MS, NRCS Science and Technology tool developers, and university researchers are testing and deploying LAMPS to automate and improve the process of providing crop management inputs, including tillage, to various software and web packages.
2. AgroEcoSystem-Watershed (AgES-W) model runs in the cloud. Calibration of watershed models with large numbers of land units and spatially variable input parameters requires substantial computer resources in terms of processing speed, memory, and parallel processing capacity. ARS researchers in Fort Collins, CO worked with university partners who developed the Cloud Services Innovation Platform to manage AgES-W calibrations that may run for days on commercial cloud computers (four virtual machines with 16 processing cores each). Jobs are submitted and monitored using the Object Modeling System Console, which allows modelers to disconnect and reconnect at any time during calibration without interrupting the remote cloud-based simulation process. This system is facilitating intensive computational research of hydrology and nutrient transport in watersheds at multiple scales. Related developments and activities allow collaboration with NRCS to host and deploy other cloud-based models and data services.
3. Simulating vertical soil hydrology using the AgroEcoSystem (AgES) response-function model. The dynamics of soil water flow and storage in soils must be correctly represented to accurately simulate hydrological processes in a watershed. ARS researchers in Fort Collins, CO developed and tested AgES, a vertical implementation of the AgES-W model, to simulate soil water dynamics in Colorado. AgES simulations compared favorably with previous simulations using a physical process based model while running in a fraction of the time. The results provide new understanding of the model responses and interactions between functions controlling the vertical flow and storage of water.
4. NRCS Curve Number hydrograph model (WinTR-20) as a web service. WinTR-20 is an essential tool for estimating surface water runoff for various conservation planning tools. It has been deployed and maintained by the NRCS, but has come to the end of its lifecycle as a desktop application. ARS researchers in Fort Collins, CO partnered with the NRCS and university collaborators to develop a web-service version of the computational engine from WinTR-20. The web service retains the full computational abilities of the WinTR-20 application but does not include the graphical user interface and its associated features. A web page client that provides access to a simple application of the web service is also currently available. The service is described at: http://www.javaforge.com/issue/420080. Implementing the computational engine as a web service permits its reuse and repurposing for other current and future applications.
Fathelrahman, E., Basarir, A., Gheblawi, M., Sherif, S., Ascough II, J.C. 2014. Economic risk and efficiency assessment of fisheries in Abu-Dhabi, United Arab Emirates (UAE): A stochastic approach. Sustainability. 6(6):3878-3898.
Ranney, K.J., Niemann, J.D., Lehman, B.M., Green, T.R., Jones, A.S. 2015. A method to downscale soil moisture to fine-resolutions using topographic, vegetation, and soil data. Advances in Water Resources. 78:81-96. DOI 10.1016/j.advwatres.2014.12.003.
Ascough II, J.C., Delgado, J.A. 2015. Modeling landscape-scale nitrogen management for conservation. In: GIS Applications in Agriculture, Volume Four, Conservation Planning, T. Mueller and G.F. Sassenrath (Eds.). CRC Press, Boca Raton, FL. pp. 101–118. doi:10.1201/b18173-8.
DeJonge, K.C., Ahmadi, M., Ascough II, J.C., Kinzli, K.C. 2014. Sensitivity analysis of reference evapotranspiration to sensor accuracy. Computers and Electronics in Agriculture. 110 (2015) 176-186.