Location: Water Management Systems Research2013 Annual Report
1a. Objectives (from AD-416):
(1) Maintain and enhance the Object Modeling System (OMS) by (a) improving the development and debugging capabilities of OMS for Fortran 95 and mixed Fortran 95/Java/C++; (b) improving the annotation and connectivity capabilities of OMS for Fortran 95 and mixed Fortran 95/Java/C++; (c) developing and evaluating OMS/cloud-based data provisioning capabilities for soil, land use/cover, and DEM data and linkage to AgroEcoSystem-Watershed (AgES-W) model input files; and (d) deploying AgES-W to a cloud infrastructure to enable scalable applications for large data sets. (2) Maintain and enhance a library of Fortran 95 and Java modules for continued development of the AgES-W watershed scale model. (3) Continue implementation of tools in OMS for land unit delineation, parameterization/calibration, sensitivity/uncertainty analysis, and spatial-temporal output visualization.
1b. Approach (from AD-416):
The Object Modeling System (OMS) currently has essential core modules (taken from the SWAT, WEPP, RZWQM2, and the European J2K and J2K-S models) used for building and deploying the AgES-W watershed model. These core OMS-based modules will be further verified and new modules added (e.g., infiltration, tile drainage, and crop growth) to further improve AGES-W. OMS functionalities for model debugging, module connectivity, data provisioning, and cloud computing will be augmented as needed during this process. ARS scientists will evaluate the AgES-W model with experimental data from two to three Colorado and Midwest watersheds for water quantity and quality outcomes of land management and conservation practices. Based on the evaluation, the component modules will be upgraded or replaced with new improved modules. To facilitate the application of the improved AgES-W model, functionalities will be developed in OMS to access appropriate NRCS databases. Additional tools developed by ARS scientists for land unit delineation, parameterization, sensitivity/uncertainty analysis, output visualization, and scaling will also be fully implemented. The final package will be delivered to NRCS and ARS modelers for further evaluation and feedback for improvements.
3. Progress Report:
The AgroEcoSystem-Watershed (Ages-W) model is the primary simulation tool in this project for exploring soil, water and plant interactions over space (field to watershed scales) and time. The development team: 1) integrated science components for irrigation scheduling, water conveyance, and passive tile drainage into AgES-W to improve simulation of conservation practices and systems; 2) modified AgES-W to better simulate soil-water processes and provide automated calibration within the Object Modeling System 3 (OMS3); and 3) implemented further model improvements to simulate surface runoff at sub-daily time increments. In addition, model code for calculating the effect of changes in soil physical properties by tillage, reconsolidation, or other practices were extracted from RZWQM2 and the new component tested under OMS3 for use in AgES-W or other models. Evaluation of AgroEcoSystem-Watershed (AgES-W) model science components for crop growth, nitrogen dynamics, and sediment transport was continued using measured water quantity/quality data from the Upper Cedar Creek Watershed (UCCW) in Indiana. Object Modeling System 3 (OMS3) enhancements were continued for better support of multiple programming languages including FORTRAN and C++. OMS3 FORTRAN support is complete, and C++ integration is on-going. AgES-W was deployed as a prototype service to the Cloud Services Innovation Platform (CSIP), a collaborative effort between NRCS, CSU, and ARS to develop a scalable infrastructure solution for modeling using cloud computing. AgES-W source code was added to the CSIP repository and the CSIP computing infrastructure was extended to provide model calibration support. Initial testing for model execution distribution across multiple cloud computers was performed. In addition, the set of Domain Specific Language (DSL)-based tools within OMS3 for calibration, sensitivity analysis, and uncertainty analysis was extended with addition of the “Particle Swarm” and “DREAM” methods. A Java-based web-service tool, the Crop Rotation and Management Builder (CRMB) was developed that utilizes a spatially enabled data provider, the CropScape web service from the USDA National Agricultural Statistics Service (NASS). The NASS CropScape service offers a remote sensing-based raster Crop Data Layer (CDL) for a specific year and a spatial Area of Interest (AOI). Land units are represented as polygons in an ESRI shape file or as a list of point coordinates in an ASCII text file. CRMB queries the annual main crop information from the NASS CropScape web-service for available CDL years. After re-classifying crop information within each AOI to adjust the final crop area using NASS CropScape provided accuracy values, CRMB selects the crop information by majority area size. CRMB then detects a sequence of main crops for every AOI and year, matches the crop sequence to actual available crop rotation information in LMOD, and links the AOI to an actual crop rotation with all available management operation information. Finally, CRMB generates required management input files for the AgES-W model. Expected accomplishments over the next year include the following: 1) develop new AgES-W science modules for conservation practices (e.g., contours, grass waterways) in order to better assess environmental outcomes of alternative conservation and management systems; 2) finalize testing of AgES-W within the CSIP cloud infrastructure to enable scalable applications for large data sets; and 3) perform AgES-W model evaluations to assess conservation practices across Colorado and specific Midwest regions.