OVERVIEW

Microclimate, Acoustic, and Video Enabled Sensor Networks:
Microclimate Sensor and Image Acquisition Networks

1.  Research Objectives:

Our primary long term objective is to design, develop, deploy, support, and evaluate distributed, continuously operating networks of embedded sensors and image capture devices for measuring environmental, physiological and ecological variables within diverse natural ecosystems, and to design, develop, deploy, support, and evaluate data storage and user interface systems that allow the sensor networks and the measurement data and images to be used efficiently for biological and environmental research. Our overall approach is an iterative process of developing system components collaboratively with both researchers and engineers, and then building and deploying those components and having researchers use them.  We use what we learn in this process to improve the existing components and design new ones.

We have organized our development and experiments in four areas.  The first is the Micro Climate Sensor Array.  This includes sensor testing and deployment for above and below ground environmental sensors, avian nesting sensors, and plant ecophysiological sensors, as well as the electronics, communications hardware and system software to control and record the operation of the sensors in the field.

The second area is the Image Collection System. Our objective is to develop an automated image collection system that will support a variety of soil, plant, and animal studies. Currently we are capturing images of nest box activity, plant growth, moss hydration status, and scene observation. These images are used in biology studies, and we also plan to use them as training images for automatic categorizing images by the activity in the image.  If we are successful with the automatic categorizing we hope to use the algorithm to on a processor co-located with the image capture hardware so that we can categorize the image in real time and avoid the requirement for high bandwidth connection to the image capture sites.

The third area is the User Interface for the sensor array and image capture systems.  The initial objective is to allow users to browse captured data and images, and control the operation of the systems. The long-range objective is to deliver real-time and historical data displayed in detailed geographic context, along with data from sensors as they are moved to various locations (the Emissary project).

The fourth area is the use of all of these experimental systems for initial biological investigations.  These investigations test the whole network in operation.  The planning for the investigations directs the design in each of the areas, and by using the whole system we discover the problems with the design as well as opportunities for creating new capabilities.

APPROACHes, EXPERIMENTs and SYSTEMS:

2. There are four areas of experimentation and development of the microclimate sensor and image acquisition network:

1. Micro Climate Sensor Array
2. Automated Image Collection System
3. Sensor Array and Image Capture User Interface
4. NIMS
5. Initial Biological Investigations

2.a.  Microclimate Sensor Array

We continued to operate the CMS wireless MICA-based sensor array, now starting it's third year of operation.  The system details were described in previous annual reports, with the exception of minor maintenance issues continues to operate as originally deployed in the spring of 2003.  We have collected over 15 million measurements in the database from this system, and the data have been used in several 'Initial Biological Investigations' described later, as well as providing real field data for CENS systems group investigations into data compression, calibration, and sensor faults.

Figure 1 - nest box (wired with Ethernet) and microclimate array (wireless using mica 1)

Last year we expanded the area of coverage, adding 12 sites with data loggers collecting temperature, humidity, PAR, and soil moisture, and we continued to operate this system throughout this year. Each of these is located at nest boxes around the Reserve, and also collect the temperature and humidity inside the nest box. These sites are part of the nest box image collection system discussed below. The construction if this system was described in last year 's annual report.  This year we have experimented with extending our Ethernet and connecting the data loggers using a Moxa serial port to Ethernet interface so that we can report the data closer to real time. The initial tests of are working, so we expect to add this to the data loggers that are within Ethernet range.

We also continue to work with the CENS Systems Group in their development of the Extensible Sensing System (ESS2) for the mica/stargate sensormicronets in ecological applications. Field deployments have been conducted for a range of software and hardware testing, as well as deployments to capture a watershed scale 4-D snap-shot  of micro-temperature suitable for characterizing daily motion dynamics of high density cold air across complex topography. Such studies may provide data about a key parameter affecting the small-scale distributions of plants. Please refer to the report about the ESS in the CENS Systems Research section of this report.

2.b.  Automated Image Collection System

This year we operated the nest box image capture system deployed last year, adding over 200,000 images to the previous nest box image library of 130,000 images.

Figure 2 - micro-video camera detail, and example imagery from nest box array, and moss cam

We also experimented with using a tower mounted Pan-Tilt-Zoom camera to collect images to augment plant phenology studies of fern growth and manzanita sap flow. The success of these experiments has lead us to plan the installation of 3 additional tower-mounted Pan-Tilt-Zoom cameras.  The three towers are installed (described in the infrastructure section of the Terrestrial annual report) and the camera is operational on one tower, starting in March 2005.

Figure 3 - example imagery from time-lapse phenology studies of Manzanita and Bracken Fern taken from Trailfinder Lodge Weather tower automated image acquisition system

We are also working with the CENS Systems Group to contribute test and training imagery in support of the CYCLOPS Imaging sensor node development project. CYCLOPS, which is described under the Systems Research section of this report, is intended to provide a technology that will greatly improve our ability to collect and process in-situ images relevant to our biological investigations. 

2.c.  Sensor Array and Image Capture User Interface

This year we extended the user interface for sensor data and images, and built an initial user interface for system configuration and control.  We also extended the database schema in order to store additional types of information. 

Interface objectives and software engineering requirements:

-       Storage and web-based browsing of measurements and images

-       Users:

-       Biology and ecology researchers

-       CENS system researchers

-       Secondary school students through the CENS Education group

-       General public through James Reserve web site

-       Intended Use

-       Viewing data trends

-       Making simple comparisons

-       Data export for more sophisticated analysis

-       Interface to CENS sensors, video capture, import from data loggers

-       Easy to replicate at field stations

Features:

-       Provide a consistent user view of measurements and images

-       Logical: Region, Site, Measurement or View, at Time

-       Physical: Region, cluster address, device address, sensor channel

-       Physical implementation can change over time

-       Fault filter raw measurement data

-       Remove out-of-range measurements from displayed data

-       Allows focus on error rate.

Figure 4 - example of raw values by sensor types

-       Limit filter raw measurement data

-       Limit valid (within spec) readings to display appropriately (for example Humidity sensors are accurate to +/- 3% at high humidity, so they can report 102% humidity.)

-       Original reading retained

Figure 5 - automatic limit filter of raw measurement data, and pre-calculated summaries by time and date

-       Pre-calculate Hour, Day, Week & Month Summaries

-       Resample values to hour boundaries

-       Calculate and store min, max, average, sum, and measurement count for the period

Allows us to create graphs which display years of data in seconds

-       Sensor GUI Wizard

-       Step-by-step form

-       Control all CMSgraph features

-       Create complex trace arrangements

-       For users with clear idea of needs

Figure 6 - Interactive graphical user interface for CMS database

-       Quick Start

-       Start with measurement catalogue

-       Add traces via checklist or site map

-       For users wanting to browse data

Figure 7 - Quick start interface (by sensor)

Site	Measurement	Current Value	First Data	Last Data
MossCam Weather	Barometer	841.50 mBar	2003-04-29 19:01:59	2005-04-06 12:59:43	Graph
MossCam Weather	Dew Point	unavailable	2003-04-29 19:01:59	2004-11-05 11:23:36	Graph
MossCam Weather	Enclosure Humidity	34.000000 %	2003-04-29 19:01:59	2005-04-06 12:59:43	Graph
MossCam Weather	Gust	0.000000 km/hr	2003-04-29 19:05:42	2005-04-06 12:59:43	Graph
MossCam Weather	Humidity	36 %	2003-04-29 19:01:59	2005-04-06 12:59:43	Graph
MossCam Weather	Leaf Wetness	0.000000 0-10 dry-wet	2003-04-29 19:01:59	2005-04-06 12:59:43	Graph
MossCam Weather	Rain	0.00 mm	2003-04-29 19:01:58	2005-04-06 12:59:42	Graph
MossCam Weather	Temperature	17.7 C	2003-04-29 19:01:59	2005-04-06 12:59:43	Graph
MossCam Weather	Wind Direction	242 Degrees	2003-04-29 19:01:59	2005-04-06 12:59:43	Graph
MossCam Weather	Wind Speed	0.0 km/hr	2003-04-29 19:02:12	2005-04-06 12:59:43	Graph
Nestbox 13	Humidity	30 %	2003-11-25 01:59:58	2005-04-06 13:02:35	Graph
Nestbox 13	Inside Humidity	27.000000 %	2003-11-25 01:59:58	2005-04-06 13:02:35	Graph
Nestbox 13	Inside Temperature	20.300001 C	2003-11-25 01:59:58	2005-04-06 13:02:35	Graph
Nestbox 13	PAR Current	538.000000 uA	2003-11-25 01:59:58	2005-04-06 13:02:35	Graph
Nestbox 13	Temperature	18.7 C	2003-11-25 01:59:58	2005-04-06 13:02:35	Graph

Time, Aggregation period	Aggregation function
Last 48 Hours, all measurements	auto
Last 48 Hours, Hourly	auto	Min	Avg	Max	Sum
Last Week, Hourly	auto	Min	Avg	Max	Sum
Last Week, Daily	auto	Min	Avg	Max	Sum
Last 4 Weeks, Hourly	auto	Min	Avg	Max	Sum
Last 4 Weeks, Daily	auto	Min	Avg	Max	Sum
Last 3 Months, Daily	auto	Min	Avg	Max	Sum
Last 12 Months, Daily	auto	Min	Avg	Max	Sum
Last 12 Months, Weekly	auto	Min	Avg	Max	Sum
All, Daily	auto	Min	Avg	Max	Sum
All, Weekly	auto	Min	Avg	Max	Sum

Figure 8 - Interface to graphs of Humidity at MossCam Weather:

-       Panning

-       Auto-pan 1/3 of visible range

-       Pan specified amount of time

-       Zooming

-       Click on graph to recenter / zoom

-       Fixed zoom of entire extent

Figure 9 - Panning and zooming is accomplished via an interactive java-script platting interface

-       Set Start/End Times manually

-       Choose unit system

-       Metric

-       English (for use by general public)

-       Y-axes: max and min

-       Manage up to 2 y-axes

-       Auto-scaled by default

-       User-specified ranged

Figure 10 - units, multiple axis, and auto scaling features of CMS GUI

-       Manage traces from Options Window

-       Add, Delete, Show/Hide traces

-       Change trace order

-       Auto-adjust / Manual control of sampling interval & data aggregation

-       Manage traces from Site Map

-       Create trace at new site

-       Add, Delete, Show/Hide sensor traces at existing site

-       Clone traces to other sites

Figure 11 - Map interface uses coordinate system to specify sensor GUI display. This feature will eventually be replaced by a geographic information system (GIS) server

-       Image Capture

-       Sample control:

-       starting/ending month & day (e.g. limit capture to nesting season)

-       start/end time of day: clock time or solar time

-       solar time is important for long-term time-lapse

-       period between image captures

Image stored in database (thumbnail calculated and stored too)

-       Image Display

Figure 12 - Catalogue of all images

Figure 13 - Browse images in ÒcalendarÓ format

2.d NIMS

The James Reserve is a field test bed for ecological research using NIMS technology. Our test bed also serves the broader NIMS program in testing environmentally rugged electro-mechanical systems, efficient power supplies, and network communication

Imaging using a pan-tilt-zoom camera mounted on the horizontally mobile node (Figure 20, below) was used in beginning phenology observations (refer to the next section, Initial Biological Investigations). Imaging and static light sensors are being used in combination for the quantification and modeling of light variation on the forest floor in order to correlate light levels with potential photosynthetic capacity for Bracken Fern.

Micrometeorological data collected by the vertically mobile mechanism (Figure 20, below) is planned to augment fixed clusters of sensors to address the question of variability of microclimatic parameters across small scales and across forest edges.

Planned biological experiments include AMARSS coordination with the portable micrometeorological station and the ability to make an extended soil energy budget.

Refer to ACTUATION section for the complete NIMS report.

Figure 14 - NIMS systems located at the James Reserve

2.e Initial Biological Investigations

Plant ecophysiology and avian reproduction monitoring

Several studies each with different species and using fixed and mobile sensors, video, and short-term placement specialized sensors was organized to investigate the appropriate sensors and measurements for plant ecophysiological studies, and avian nest monitoring using embedded networked systems. These are research programs designed to refine science applications of CENS technologies using the James Reserve testbed.

Plant ecophysiological studies

• Phenology of native azaleas, Rhododendron occidentalis, and phenology and microenvironment of bracken fern, Pteridium aquilinum, and star moss, Tortula princeps, were recorded using NIMS mobile imaging, robotic fixed video on the Trailfinder micrometeorological tower, a fixed camera installation (ÒMossCamÓ), and temporary arrays of temperature, photosynthetically active radiation (PAR), and soil moisture sensors.
• Sap flow data of Manzanita, Arctostaphylos pringlei, and ponderosa pine, Pinus ponderosa, were collected using temporary placement of thermocouple-based sap flow, temperature, and PAR sensors.
• Carbon balance modeling of star moss, T. princeps, has begun using networked fixed-video and micrometeorological sensors as well as laboratory-based gas exchange measurements; these measurements are providing approaches to collect ecophysiological data using simple digital video sensors.

Figure 15 - Two model species, R. occidentalis and Pt. aquilinum, were selected at the James Reserve and phenology was recorded by hand as well as remotely by video camera.  Images were then analyzed for simple correlations with ground-based measurements (Figure 21, above).  Work is currently being conducted (NIMS) on image analysis and 3-D modeling of images to further refine automatic estimations of leaf areas.

Figure 16 --Sap flow methods provide a direct measurement of whole-plant water use with high time resolution. The technique has distinct advantages over other methods such as leaf gas exchange, plant chamber, soil water balance. Sap flow through woody and herbaceous plants is tightly correlated with local micrometeorological conditions and can be used as an indicator of plant water status and plant stress temporally as well as spatially.

Two techniques for measuring sap flow, the constant heating method and the heat balance method, were investigated for applicability to temperate forest and tropical forest, ecosystems, as well as for both woody and herbaceous species, 

Short-term (days) observations of sap flow using the constant heating method in A. pringlei (top figures; narrow branches and thin xylem) maintained a tight correlation to local, instantaneous micrometeorological events recorded by sensors within the JR infrastructure as well as with temporarily-placed static sensor nets. 

Long-term observations of sap flow using the constant heating method in P. ponderosa  (middle figure, only three days shown, deep xylem) produced results consistent with the onset of winter and reduced soil water availability

Short-term (days) observations of sap flow using the heat balance technique in Philodendron radiatum (bottom figure, non-woody aerial roots of a tropical epiphyte) were correlated with mass-balance of water to verify the technique.

-       All techniques proved valuable in determining the quantity and patterns of water use by different growth forms of plants.  Correlations with micrometeorological and soil water availability data that vary temporally and spatially will allow experiments in native stands of plants at JR and future test bed locations.

Moss cam: Carbon balance modeling is under way using color data from historical video and micrometeorological data at the James Reserve in addition to laboratory-based measurements of gas exchange.  The CO₂ uptake responses of the star moss, Tortula princes, to three major environmental factors (water status, temperature, and light) were measured in the laboratory and are now being compared to the precipitation, temperature, and light levels at the site.  A productivity model was developed to relate physiological responses to daily values of these microclimatic variables thereby allowing evaluation of seasonal changes in productivity.  The model is a first-order estimation and is defined as the product of the water status, as correlated with the color of the moss, the air temperature, and the light levels at the site, each of which was assigned a value of unity when that factor was not limiting net CO₂ uptake.

Figure 17 - comparing two color evaluation models of daily video images averaged over monthly intervals to monthly rainfall data at an adjacent microclimate CMS node the James Reserve.

-      Avian nest box and reproduction monitoring: 13 nest boxes were monitored using environmental sensors and digital video. The following are results of the 2004 nesting season:

	Nest Building	Eggs Laid	Incubation	Eggs Hatched	Days to Fledging
Nest box 8	Successful
Duration	32 days	5 eggs	18 days	5 eggs	24 days
Start	05/02/04	06/03/04	06/03/04	06/21/04	06/21/04
End	06/03/04	06/07/04	06/21/04	06/21/04	07/15/04
Nest box 22	Successful
Duration	13 days	5 eggs	13 days	4 eggs	23 days
Start	04/20/04	05/03/04	05/06/04	05/19/04	05/19/04
End	05/03/04	05/07/04	05/19/04	05/23/04	06/11/04
Nest box 31	Successful
Duration	9 days	5 eggs	15 days	4 eggs	23 days
Start	05/01/05	05/10/05	05/12/05	05/27/05	05/27/05
End	05/10/05	05/14/05	05/27/05	05/28/05	06/19/05
Nest box 27	Abandoned		Nest box 47	Abandoned
Duration	1		Duration	8
Start	05/02/04		Start	05/21/04
End	05/03/04		End	05/29/04
Nest box 3, 14, 21, 45, 48, 54, 55   Vacant

Figure 18 - reproductive success as measured by nest building, egg laying, and successful fledging of young are now automatically monitored in 13 nest boxes. Sensors also record temperature and humidity (inside and outside of the box). Nest box failures are directly attributable to predation, nest abandonment due to parent mortality, and lethal temperature fluctuations.

3. Problems Encountered:

• The original CMS hardware is nearing its end of life, and is requiring more maintenance.  We anticipate replacing it in the next year.
• A lighting strike at the Reserve last summer damaged some equipment, but did not disrupt data collection from most sites.

• Phenology imaging of R. occidentalis was complicated by localization of moving branches, and phenology measurements of both R. occidentalis and Pt. aquilinum were complicated by variable ambient light conditions (sunflecks interfering with imaging).  Solutions involved fixing branches to poles and timing image capture events for diffuse light conditions.
• Sap flow measurements of A. pringlei were complicated by the lowest-range available size of sap flow probes.  Sap flow measurements of P. ponderosa are difficult because of wounding responses and variable sapwood thickness of trees, in addition to the unknown variables of sap flow during season transitions.  Repeated measures of sap flow and the application of the heat-balance method for sap flow will clarify data trends.
• Carbon balance modeling and phenology of moss using images was complicated by leaf litter, variable rock exposure, and sun flecks in the images captured resulting in inconsistent brightness and color data.  Solutions involved supervised classification of images to identify specific regions of interest that were not affected by spurious events.
• Additionally, the wavelengths used by the infrared camera were not the same as those in remote sensing and used by remote sensing vegetation models, requiring the development of new models for color estimations of photosynthetic capacity.

4.  Major Accomplishments:

• Created a basic user interface for sensor data and images, which researchers have successfully employed to explore the data archives and obtain data for analysis.

-      A new system to acquire plant and animal phenology measurements provides an important means of understanding the relationship between periodic phenomena, such as leaf flush or flowering, and micro-environmental and climatic conditions.  Phenology relates strongly to primary productivity and the energy that enters into ecological food webs, and thus is vital in understanding ecosystem function and the effects of climate and climate change.  The study of phenology requires repeated measurements on multiple individuals at different geographic locations and appropriate microclimatological data.  The collection of phenological data can be time-consuming and labor-intensive and thus basic phenological information about many key plant species is either not known or is estimated from incomplete data sets. The ability to remotely observe leaf flush, branch elongation, flowering, and fruit load at many locations, while simultaneously collecting micrometeorological data at each site, will revolutionize the field of plant phenology.

-      Wireless networks coupled with actuated imaging technologies at JR are currently collecting high frequency, multiple location measurements of microclimate with phenological images (both plant and animal). The data generated by this combination of technologies has immediate ecological significance in terms of understanding population dynamics of producers as well as herbivore communities and examining the climatic effects on plant distributions and primary productivity, wildlife Terrestrial utilization and reproductive success in a changing microclimate.

5.  Future Goals and Objectives:

-       Enhanced map/GIS interface (EMISSARY)

-       Saved/Bookmarked graphs

-       Integrated image and measurement display          

-       Provide on-line metadata for both environmental measurements and sensor information

-       Improve data export

-       Convert the CMS to Mica2/SIB2 hardware, and ESS system software.

-       Expand the number of nodes in CMS to approximately 50.

-       Expand the types of sensors installed.

-       Integrate with NIMS mobile sensors.

-       Make a user interface that is more interactive and accessible to non-expert users.

-       Create a meta-data description of the CMS that is connected to the user interface.

-       Integrate CMS data and the avian nest box microclimate and image data, and include all the data in the online interactive user interface

PEOPLE

Faculty:

Deborah Estrin
Eric Graham
Michael Hamilton
Brent Mishler
Mani Srivastava
Stefano Soatto

Staff:

Mohammad Rahimi
Mike Taggart
Mike Wimbrow
Vanessa Rivera del Rio

Students:

Sean Askay
Naim Busek
Caitlin Hamilton
Jeff Robertson
Marina Sharifi
Tom Schoellhammer
Thanos Stathopoulos