Integrated Sensing of Soil Moisture at the Field-Scale: Measuring, Modeling and Sharing For Improved Agricultural Decision Support

Date
2014-10-08
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Determining the best way to e ciently use limited water resources, for food and energy- dedicated crops, has become crucial due to the rise in extreme events (floods/droughts) and higher variability in rainfall attributed to global climate change. Changing climate condi- tions will require new crops to be adapted to a changing agricultural environment. Reliable information on seasonal trends in crop growth and evapotranspiration with associated uncer- tainty/con dence ranges is crucial to guide the development of new crops and management strategies to cope with future climate. Given that crop growth is strongly coupled to soil moisture, developing reliable growth curves requires a detailed understanding of soil mois- ture at the eld-scale. Typically, it is impractical to collect soil samples to adequately assess soil moisture that represents both spatial distribution at the eld-scale and temporal dy- namics on the scale of a growing season (e.g. 110 days for cereals). A novel way to address soil moisture monitoring challenges is through an integrated, agro-ecosystems-level approach using an integrated sensing system that can link data from multiple platforms (wireless sen- sors, satellites, airborne imagery, near real-time climate stations). Assimilated data can then be fed into predictive models to generate reference crop growth curves and predict regionally-speci c yield potentials. However, integrated sensing requires interagency coop- eration, common data processing standards and long-term, timely access to data. Large databases need to be reusable by various organizations and accessible in the future, with comprehensive metadata. During the 2012 growing season a feasibility study was conducted which involved measuring eld-scale soil moisture with sensor network technology. The ex- periment utilized radially-distributed sensors for tracking in-season soil moisture data was collect using both automated in-situ sensors and hand-held sensors. Box-plots of soil mois- ture data was collected with hand-held revealed an early season wet soil moisture regime and late season dry soil moisture regime. The data sampled on July 5, 2012 was selected for geostatistical analysis. Bayesian kriging models and Bayesian kriging models with polyno- mial trends using di erent combinations prior distributions for the range and nugget were tested. Models with rst order polynomial trend, a reciprocal2 prior distribution for the range and a reciprocal prior distribution for the nugget t tended to predict the sample distributions the best. Soil moisture was predicted at a set of random point using ordinary kriging, universal kriging, the two Bayesian kriging models and the Bayesian kriging models with rst and second order polynomial trends. Overall, the universal kriging and Bayesian trend models predicted similar data distributions. OpenGIS-compliant services and stan- dards were utilized to provide long-term access to sensor data and construct corresponding metadata. Sensor Model Language, an inter-operable metadata format, was used to create documentation for the sensor system and sensing components. Two di erent third party im- plementations of the Sensor Observation Service were tested for providing long-term access to the data. This work discusses a set of key recommendations for monitoring eld-scale soil moisture dynamics for integration with remote sensing and models, including: 1) Improved in-situ sensing technology that would allow for less restrictive soil moisture measurements. 2) Integration of eld-scale in-situ networks with regional remote sensing monitoring. 3) The development of software and web services to integrate data from multiple sources with models for decision support.
Description
Keywords
Agronomy, Soil Science, Hydrology
Citation
Phillips, A. (2014). Integrated Sensing of Soil Moisture at the Field-Scale: Measuring, Modeling and Sharing For Improved Agricultural Decision Support (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/26574