Events
Rob Fatland
Microsoft
Microsoft External Research Division
Redmond, WA, USA
Dr. Dennis R. (Rob) Fatland works as a program manager for the External Research division of Microsoft Research, specifically on the use of Microsoft-driven (and other) technologies in service to environmental research. His work currently includes sensor network adaptation to rainforest micrometeorology in Brazil and data and model-driven studies of the biogeochemistry of terrestrial-marine coupling in Southeast Alaska. Dr. Fatland received a B.S. in physics from the California Institute of Technology in 1987 and a Ph.D. in Geophysics from the University of Alaska Fairbanks in 1998. He has worked for 6 years at NASA-JPL in radar remote sensing and for an additional eight years for (then) Vexcel Corporation in Boulder Colorado with emphasis on remote sensing and ground-based sensor networks and applications of geospatial technology to environmental science and applications.
Abstract
Microsoft Research Data Segment Collaboration
I work as a program manager in Microsoft Research, in the External Research division “Earth Energy and Environment” theme. I have a PhD in geophysics and a background in satellite radar interferometry. This remote sensing method is useful for discerning small-scale (centimeter) surface displacements over time, for example in the context of ice motion, seismic activity, or subsidence in response to sub-surface aquifer depletion. My work has also extended to sensor network development, deployment, and data harvest, and to in-depth water sample characterization related to terrestrial-marine nutrient coupling. I was part of the USP-FAPESP-MSR-JHU collaboration to deploy a sensor network in Mata Atlantica in late 2009, a pilot study that has led to my current involvement in facilitating a “next generation” study in Brazil. I am based in Redmond, Washington, USA.
My interest in environmental research can be expressed in terms of two perspectives: First I would like to see better development of technological solutions in response to scientific motivation. Second, I would like to see more completeness in technology solutions towards an eventual idea where the technological detail has become secondary to use, as is the case in the consumer market for example with mobile phones. To achieve these twin objectives of technology adoption and ease of use the technologist must cross over into the environmental science domain, first to learn what the problems look like, and second to offer what is possible in a given time frame and budget. For example deploying 5000 temperature sensors may be feasible, but 5000 CO2 sensors will be prohibitively expensive. At what point is the best science going to be done: With 5000 temperature sensors or 1000 temperature sensors and 20 CO2 sensors? To that end the following is a partial list of candidate technology options:
• In situ sensing
o Scale-up of pilot study to 2000 + sensors { temperature, soil moisture, RH, PAR, TSR, NDVI, “expensive” limited quantity sensors like CO2 or wind speed }
o Sensor model 1: It produces analog voltage: Can mate to data logger / network node
o Sensor model 2: It produces a digital signal: Can mate to data logger / network node
o Multiple meteorological stations
o Sensors in tree boles, in soil, leaf undersides/topsides, suspended in air, above canopy,…
o Water transport: Stream gauges, vadose / hyporheic / groundwater flow
• Remote sensing
o MODIS, other multi-spectral imaging platforms: Validation and multiple spectrally-derived pa
rameters
o Airborne LIDAR and optical: High resolution characterization including canopy structure
o Spaceborne imaging radar: Penetrate clouds, backscatter related to foliage characeristics
• Sample analysis
o Water samples: DOM, nitrogen, phosphorous, spectral analysis (EEM, MS, absorption…)
o Soil samples: soil moisture, organics characterization, roots, mycorrhizae
• Modeling
o Climate modeling is turning to local sub-scale models coupled to GCM boundary drivers.
Throughout this experimental process it is our hope (MSR with collaborators) that we can address and support engineering challenges in support of science objectives. These include but are not limited to sensor calibration and validation, data provision and provenance, standards and interoperability, maintenance of error and precision estimates, and visualization options.