Agriculture Information Service Built on Geospatial Data Infrastructure and Crop Modeling

Kiyoshi Honda,Amor V. M. Ines,Akihiro Yui,Apichon Witayangkurn,Rassarin Chinnachodteeranun,Kumpee Teeravech

DOI: https://doi.org/10.1145/2637064.2637094

2014-01-01

Abstract:An agricultural information service platform, called FieldTouch, is being built and tested on geospatial data infrastructure and crop modeling framework. More than 100 farmers in Hokkaido, Japan, have been participating on this development and are utilizing the services for optimizing their daily agricultural practices, e.g., planning and targeting areas where to apply fertilizer more to enhance homogeneity of growth and robustness of crops in their fields. FieldTouch integrates multi-scale sensor data for field monitoring, provides functionality for recording agricultural practices, then supports farmers in decision making e.g., fertilizer management. RapidEye satellite images are being used for monitoring vegetation status updated every two weeks. Field sensor data from 25 nodes record soil moisture and temperature data at different soil depths, and suites of meteorological variables e.g., rainfall, minimum and maximum temperature, solar radiation, wind, etc. every 10 minutes. Data from national weather observation network, AMeDAS, is also a source of daily weather data. We used "cloudSense" sensor backend service that serves meta-data and data to FieldTouch via a standard web service called SOS (Sensor Observation Service), which brought great flexibility and enhanced automation of system's operation. Using agronomic data from experimental station, the cultivar parameters (genetic coefficients) of a local wheat variety were calibrated for the DSSAT (Decision Support System for Agrotechnology Transfer) crop model using data assimilation. These were built in a web-based DSSAT wheat crop model called Tomorrow's Wheat (TMW) where in a user can explore the effects of timing of sowing at a given climatic condition, soil and crop management. TMW accesses long-term weather data from the on-line observation station up to the most recent archive, parameterize a built-in weather generator, then generate 100 weather scenarios then runs the wheat model at the chosen planting date, then two weeks, and one week before and after that. The yields are presented as distribution of yields at these different planting options. Future developments are going-on to personalize more the system so that the user can input fertilizer scenario, and be able also to apply seasonal climate forecast, and link to the 25 sensor nodes to simulate current plant conditions given a management scenario. In this way, the user can be informed better on how to manage their sources of vulnerabilities in their fields.