geocens source talk: results from an atlantic rainforest micrometeorology sensor network pilot study

Source Talk: Results from an Atlantic Rainforest Micrometeorology Sensor Network Pilot Study

Humberto da Rocha, Professor Titular, University of São PauloAndreas Terzis, Associate Professor, Johns Hopkins UniversityRob Fatland, Program Manager, Microsoft Research

Microsoft ResearchEfforts, projects, and tools towards advancing environmental science

Presented by Rob Fatland: rob.fatland@microsoft.comExternal Research “Earth Energy and Environment” themeBrazil Rainforest Sensor Network projectMODIS AzureWorldwide TelescopeSciScopeOther tools and projectsNext steps

General Idea (getting to the science)

Data Systems FIELD

sample

in situ

model

remote

proxies

• Replicate methods + solutions

• Iterate iterate iterate

• Join the model ecosphere

• Communicate

• Educate

• Policy impact

• Citizen science

Specifically Mata Atlantica Micrometeorology

• Is the data system robust?• Is the data scientifically valid?• Can this experiment be adopted or replicated elsewhere?

Physical system

Sensor networks: In practice

• Computers are reliable and foolproof

• Networks are reliable and foolproof• Therefore use powerful network

nodesAlternative idea• Simple is good• Simple uses little power• But still have to figure out where to put

the smarts$$$

011101

tradeoffs

Johns Hopkins: Life Under Your Feet:

Koala

Sensor networks: In practice

• The network node

BatteryPack

EnclosureMounting Hardware

Mote

Dessicant

)))(((Transparent plastic

Sensors: InternalSensors: External

Radio antenna

Sensor networks: In practice

The network node

Sensor networks 2: Koala in practice

Gateway mote

Data recovery operation:Once every 24--48 hours

The network

Data manipulation using .NET

Data and Results

Data and Results

Black: Over 32 days how do 3 adjacent sensor vary compared to 0.1 deg Celsius?

Red (right vertical axis): What is the temperature in the canopy?

0.1

0.2

0.3

10 days

20 days

Temperature degrees Celsius

Temperature averaged standard deviations, deg Celsius?

Summary

• 20 million data values: One small step • Data appears to be scientifically valid• 100% mote survival rate• Study site ready for further studies• Scalability an open question

Microsoft ResearchEfforts, projects, and tools towards advancing environmental science

Presented by Rob Fatland: rob.fatland@microsoft.comExternal Research “Earth Energy and Environment” themeBrazil Rainforest Sensor Network projectMODIS AzureWorldwide TelescopeSciScopeOther tools and projectsNext steps

Data preparation

• MODIS– MOD04 (aerosol)– MOD05 (precipitable water)– MOD06 (cloud)– MOD07 (atmospheric profile)– MOD11 (land surface temperature)– MCD12 (land cover)– MOD13 (vegetation index)– MCD15 (LAI)– MCD43 (albedo)

Upscale to continental scale

Upscale to the global scale

Questions

• 1. How do spatial-temporal patterns of ET and GPP vary over multiple years at global scale

• 3. How does global dimming/brightening influence GPP, ET, WUE and LUE

• 4. How does diffuse radiation modulate GPP, ET, WUE and LUE?

• ET = Evapotranspiration• GPP = Gross Primary Production (some of which is consumed by the

producer during nighttime respiration)• NPP = Net Primary Production (after producer has respired, hence

towards biomass)• WUE = Water Use Efficiency• LUE = Light Use Efficiency

Microsoft ResearchEfforts, projects, and tools towards advancing environmental science

Presented by Rob Fatland: rob.fatland@microsoft.comExternal Research “Earth Energy and Environment” themeBrazil Rainforest Sensor Network projectMODIS AzureWorldwide Telescope: DemoSciScopeOther tools and projects

Microsoft ResearchEfforts, projects, and tools towards advancing environmental science

Presented by Rob Fatland: rob.fatland@microsoft.comExternal Research “Earth Energy and Environment” themeBrazil Rainforest Sensor Network projectMODIS AzureWorldwide TelescopeSciScope: DemoOther tools and projects

Microsoft ResearchEfforts, projects, and tools towards advancing environmental science

Presented by Rob Fatland: rob.fatland@microsoft.comExternal Research “Earth Energy and Environment” themeBrazil Rainforest Sensor Network projectMODIS AzureWorldwide TelescopeSciScopeOther tools and projects

Other tools and projects• Excel Add-Ins: Web Charts, NodeXL, Power Pivot• Worldwide Telescope Earth: Imagery, topography, vector and

point data• Coastal Glacier project• Building machinery to couple BGC from terrestrial

watersheds to marine productivity• Building a data analysis stack to work across multiple

spectral methods (absorption, MS, fluorescence, NMR)• TRIDENT workflow• Stream Insight, SQL Server Spatial, LINQ, Azure, WM7• X-Express• .NET and Visual Studio