A series of rapid development workshops, or sprints, are being held to allow pulses in the development of applications and tools which can be used by the AgMIP research community. The applications developed at these workshops fall into the following categories, in order of priority:
- STEP 1: Translate data from the AgMIP harmonized format used in the ACE (AgMIIP crop experiment) database into the model-specific formats used by each participating AgMIP crop model.
- STEP 2: Translate simulated outputs generated by each participating AgMIP crop model to the AgMIP harmonized format used by the ACMO (AgMIP Crop Model Output) database
- STEP 3: Translate data from model-specific input formats, csv and other formats to the AgMIP harmonized data format used by the ACE database
- STEP 4: Developing the user interfaces for accessing and archiving the data and translators
The figure below shows the flow of data from various formats of raw data to AgMIP harmonized (ACE) format (Label "A" on the figure) to the model-specific formats for each of the participating AgMIP crop models (label "B"). After crop model simulations are performed, the model outputs are translated from each model format to the standardized format used for storage and archiving of simulated data in the ACMO database (Label "C"). These data are then available for analysis or as input to regional or global economic models (Label "D").
AgMIP data structures are based on the data definitions for crop experiments that are fully described by the ICASA Master Variable List, One of the first tasks that must be done is to map the ICASA variables to the names, formats, location and units of the corresponding variables used by each model. In many cases, the variables are in different units and unit conversions are necessary. The format of these workshops is to pair a crop model expert with a programming expert to rapidly develop the applications specific to each crop model (steps 1, 2 and 3 in the above figure). The crop model expert has the knowledge of the crop model input and output formats, the variables required by each model, and the units and descriptions of data inputs and outputs and can map these model variables to the corresponding ICASA variables. The programming expert can work with the crop model expert to help write the software applications to read the standardized AgMIP data formats and write the data into the model-ready format for each given model.
The first AgMIP IT development workshop was held at the South Asia Regional workshop in India, February 20-24, 2012. This workshop began the development of data translation tools for Aquacrop, Infocrop, ORYZA and STICS. A tool for importing data from a spreadsheet format was also developed.
The second development workshop was held in Gainesville, Florida, July 9-14, 2012. This workshop concentrated on developing data translation tools for APSIM, Aquacrop, STICS and WOFOST crop models. Many discussions were held during the course of the week that allowed us to improve the harmonized data structure to accommodate a wider range of modeling perspectives. Some of these changes slowed development of existing translators, but allowed for a much better overall system for development for these models and future AgMIP developers.
A third development sprint workshop was held at the AgMIP North America Regional workshop in Ames, Iowa September 4-7, 2012 which included EPIC, CropSyst, Salus and RZWQM2 models. Further enhancements to the data structures and modeling environment resulted from this workshop. .
The 4th development sprint (AKA The Austin Hackathon), hosted by the iPlant Collaborative, was held at the Texas Advanced Computing Center in Austin, TX, USA, February 4-8, 2013. Participants at this workshop developed data translation tools for APSIM, AquaCrop, CropSyst, InfoCrop, SarraH, STICS and WOFOST.