Skip Header NavigationIntranet 
HomeAbout UsResearchEducationResourcesPeople

Research Project

Networks and Infrastructure for Environmental Sensing and Image Aquisition

Applications > Terrestrial Ecology Observing Systems > Networks and Infrastructure for Environmental Sensing and Image Aquisition

On this page: Overview | Accomplishments | Future Directions

Overview of Systems

Micro Climate Sensor Array (CMS/CMS2/ESS/CAD):

ESS: The environmental sensing system (ESS) is a key component of a collection of tools that together are a nearly complete, end-to-end, sensor-to-user facility for deploying and managing a sensor network.  The system is composed of a small number of well-modularized components.  These include: CC1000 radio stack (including BMAC/LPL low-power listen); standard Crossbow MDA300 sensor board driver; several experimental routing services; time management service; persistent data buffer; system status service; and a remotely configurable data sampling engine.  The components seamlessly work together to allow a wireless system of homogeneous nodes to reliably route their data back to a base station.

CAD/CMS2:  The CMS2 and CAD transects both employ ESS. CMS2 is now poised to become the successor to the original CMS which has been operation for just over five years and has collected more than 60,000,000 measurements.  Placing CMS2 sensors side-by-side in the same environment with the original CMS stations has allowed us to verify that the newer hardware and software in CMS2 produced results consistent with CMS and provides verification of the original CMS data. Analysis of the CMS2 data has shown an overall uptime of 92.7%.  All of the collected data is stored in local databases and available to researchers via the web through the James Reverse servers. 

CAD is a system of 27 wireless nodes sending back temperature and humidity data at five minute intervals.  The network topology consists of a long string of nodes returning data to a base station located approximately in the middle.  Nodes at the far ends of the topology successfully route their data over five to six wireless hops.  The transect has been in operation for over one and a half years.

Figure 1aFigure 1b

Figure 1. On the left is a combined CMS/CMS2 weather station.  Each weather station consists of a number of standard sensors.  Pictured are: rain, wind speed, wind direction, temperature, humidity, and leaf wetness.  On the right is part of the Cold Air Drainage (CAD) transect.  Data collected from this experiment is being added to the JR database and is viewable at

A new Campbell Scientific reference weather station was installed to replace the aging Texas Weather station.  Because the old tower was in an area where construction is planned in the coming months, a new tower was erected at a nearby location.  In addition to the existing suite of instruments (wind speed, wind direction, broad spectrum solar radiation, humidity. temperature, and leaf wetness), several new sensors were added.  They include: Photosynthetically Active Radiation (PAR), fuel moisture content, and a pyranometer. In addition to providing reference data to CENS, the tower is also being used as a test platform for a suite of NEON meteorological instruments (more information about the NEON deployment is included below).

Figure 2aFigure 2b

Figure 2.  At left the forms have been built and the electrical run in preparation for pouring three yards of concrete for the base of the 30’ tower.  On the right is the fully instrumented and operational tower.

Automated Image Acquisition: The wildlife observation cameras installed last year at remote UCNRS sites as well as the two additional tower cameras at the JR have continued to reliably produce and upload images.  They are collected automatically, documenting raptor nesting locations and bighorn sheep activity as well as plant phenology.  Monitoring Bighorn sheep is one example of why an unattended observation system is such a powerful aid to researchers because capturing images of the animals might take weeks of sitting in 100F temps, whereas a networked camera allows easy access from the comforts of an office – with far less disturbance to the research object. 

What began as an experiment last year utilizing a mote-based camera (Cyclops) has spawned two additional areas of research: wireless cameras monitoring pitfall traps and imagers as sensors in nest boxes. For the pitfall traps, just the determination of presence versus absence is valuable in alerting a researcher to tend to an experiment.  For the nest boxes, continuous, periodic sampling of the interior of a nest box can give information such as: percentage of time a bird is incubating her eggs, the number of eggs at any time during the incubation period, the amount of time the father bird returns to the nest, and any number of other questions.

Figure 3aFigure 3bFigure 3c

Figure 3. Cameras at Deep Canyon, Agave Hill, and a remote watering hole are challenging to install and maintain.  The mote-based Cyclops pictured below in left) requires less infrastructure for deployment and allows researchers to better protect cameras from external factors.) The image (below center) is of a nest box with four eggs.  Below right is an image from a pitfall trap showing three lizards.

Figure 3dFigure 3eFigure 3f

Utilities and Networks: Maintained the James Reserve’s local and wide-area network system (wired and wireless), data storage systems, and our locally generated electrical power distribution systems.  Replaced all of the JR deep cycle batteries (the batteries have about a five-year life and were last replaced at the start of the CENS program in 1992) and added an additional 2500W of photovoltaic panels. Though not optimal for sun exposure, they do reduce fuel consumption by the generator.  Long-term plans call for adding additional south-facing roof surface to the Lodge that will double as storage for additional batteries.  Due to multiple days of below freezing temperatures, the water pipes to Trailfinder Lodge froze, cutting off the water supply for just over a week.  A secondary water path was added to both restore water flow and to prevent such a failure in the future.

Figure 4aFigure 4bFigure4c

Figure 4.  On the left are some of the replacement batteries that were added this past year at the Reserve.  Additional solar panels (center) we’re added to the East side of Lolomi Lodge.   On the right is a portion of the new water line running to Trailfinder Lodge.

Communications Transect

Overview: The communications transect was implemented to record the relationship between microclimate changes and RF communication reliability at frequencies used by our wireless nodes. 

Approach:  We have deployed six wireless nodes with environmental sensors and permanent power sources (solar panels) to allow uninterrupted operation for approximately one year.  We plan to use connectivity data collected from communication among these nodes to correlate changes in temperature and humidity with RF signal propagation.

Systems:  As mentioned, there are six communication transect nodes currently deployed at James Reserve.  The node locations were chosen in an attempt to reach the outer limits of radio connectivity, thereby allowing us to observe connectivity between different pairs of nodes come and go as environmental conditions change over time.

Each node has been programmed to send out a series of beacons at defined intervals and, when not sending beacons, to listen for beacons from other nodes.  Upon receiving a beacon, a node will record from which node it heard the beacon along with various other measurements of the signal strength of the beacon.  The node will then return the recorded data, along with environmental data, back to a central base station.  Finally, these measurements are uploaded to a database at CENS. 

We now have one year’s worth of connectivity data, giving us information about a range of various microclimate conditions.  Towards the end of the experiment two of the six locations were severely damaged by falling trees, causing some loss of continuity.  The collected data has yet to be analyzed. 

Future work:  The next step will be to correlate the RF connectivity data with the environmental data.  If we determine any relationships, it will become possible to make future networks adaptable to their environment.  For example, if we find humid conditions cause connectivity to go down, we could allow the motes to raise their signal strength when detecting excessive humidity.  The end result will be improved connectivity but lower power consumption.

Figure 5

Figure 5.  We were able to utilize already existing mote housings to setup an RF communications transect to study the effects of environment and foliage on signal propagation

Collaborative Efforts Utilizing CENS Infrastructure:  The James Reserve has been able to provide the National Ecological Observatory Network (NEON) a test bed for prototype deployments and testing of NEON systems.  This is an excellent example of how the lessons learned from five years of field deployments and collaborative research efforts throughout a variety of disciplines is allowing others to benefit from our efforts moving technology and science forward.  Another example is our work with Mike Goulden from UCI in studying eddy covariance using above canopy towers.  Despite a variety of other potential locations, one of the reasons the James Reserve was chosen was because of extensive experience with field deployments and convenient infrastructure.

Figure 6aFigure 6bFigure 6c

Figure 6.  At left is the 100’ tower erected by UCI to study eddy covariance.  In the center is a view of some of the NEON meteorological instruments taken from a camera mounted atop the reference weather station tower.  At right is the inside of the weatherproof box that houses not only the CENS reference weather base station, but also the NEON base station.

Google Earth as a TEOS visualization platform

As the scale and complexity of TEOS sensor networks increase, so does the need for an intuitive, integrated systems and data visualization interface? Such a system should satisfy the practical engineering needs of developing, deploying and maintaining embedded sensor networks. CENS engineers need a way monitor the functionality and health of our sensor networks by visualizing live and archived data. They also need a centralized administration gateway to the numerous systems operating at the James Reserve. CENS participants also need a cohesive way to showcase and demonstrate their sensor networks to both researchers and the general public. Finally, CENS is seeking to develop field-portable tools for in-situ data visualization, spatio-temporal modeling and sensor network diagnostic tools. The beginnings of this conceptual product, Emissary, have been in use of Google Earth as a sensor network visualization platform.

Started in late 2005, the James Reserve Google Earth layer (JR GE layer) was our first attempt to draw together all the TEOS systems at the James Reserve into a single visualization interface. Google Earth, a widely available '3D virtual world' application, allows was easily integrated into the existing web-based tools and data tools in use at the James Reserve. Google Earth allows for the realistic 3D representations of sensor equipment, current snapshots from the Reserve’s various webcam networks, quick access to current and historic sensor measurement graphs, and the integration of various GIS (Geographic Information Systems) layers. Over the past year we have continued development of this interface; the new features and many improvements behind-the-scenes have been implemented in the James Reserve Google Earth layer since the last reporting period are summarized here.

New features and interface improvements

In order to make the experience more realistic and immersive, 3D models were created with Google Sketchup (a 3D modelling program) for all the major TEOS systems accessible via the JR GE layer (Figure X.1). As support for models are a new feature of Google Earth 4.x and KML 2.1 (the XML data format used by Google Earth), backwards compatibility to the original symbolic 3D polygons (Figure X.2) is still provided via a user controlled parameter. Examples of the original photos and both basic polygon and 3D model representations are show in Figure X.3.

Improvements were also made to the accessibility and handling of webcams in the JR GE layer (Figure X.4). Server-side caching helps to speed the download of images, and prevents direct requests from DMS database for archived images, as with earlier versions of the JR GE layer. More options are now available in the pop-up description balloons for the webcams which allow for more direct access to archives images from various time windows (e.g., the last hour, day, week and month - Figure X.5). As will be discussed in more detail later, these webcam images and other features of the layer can be optionally bundled into an all-inclusive “KMZ” download which allows for off-line viewing and giving of reliable demonstrations of the layer.

Better access to measurement graphs and image time sequences was also added to the JR GE layer (Figure X.6). 24-hour trend lines (also called “sparklines”) are now clickable, taking the user to a graphing interface of the last 24 hours for that measurement. The entire date range of data availability for each particular sensor is now linked in the description bubble, and also leads to an interactive graph of that data range. Deployment sites that have both archived images and sensor measurements are now linked to a previously existing version of the graphing interface that allows side-by-side stacking of this information (Figure X.7).

Additional GIS layers have been developed and exported for use in the Google Earth interface. In addition to the original set of James Reserve specific GIS layers -- such as a digital elevation model, slope, aspect, and contour lines -- these same types of layers have been added for the entire area of the Hall Canyon watershed (Figures X.8-13). Behind-the-scenes, the generating script was altered to allow for the new images and GIS layers to be added to the JR GE layer by simply adding them to a special directory on the webserver. This makes the addition of new image overlays and GIS layers very easy. A new, interactive plant community layer was also created from a GIS shapefile (Figure X.11). This layer breaks down each plant community type and sub-type in separate folders with unique polygon colors; this allows the user to quickly identify areas of interest. By hovering the mouse over the center-point of each polygon highlights the area in red, and clicking on the point reveals representative photos and links to botanical resources about that plant community (Figure X.14).

Programming Improvements

The majority of the improvements made to the James Reserve Google Earth layer this past year took place behind-the-scenes in the PHP (pre-Hypertext Processor) scripts that generate the KML layer dynamically. The first major improvement was the development of a PHP class for KML generation. This programming data structure allows for KML schema and tag value validation, thus helping developers from assembling incorrect KML code. The class also includes functions that will automatically convert column names from a source data table into custom KML tags. This allows for a data values to be encoded as XML within the KML, making it machine-readable for later parsing and for the eventual use of HTML templates for the more efficient generation of the description balloons. The class also features some basic functions which allow the developer to easily draw polygons of specified size, height, and shape.

By requesting the KML file from the PHP script from a web-browser, and by including a debugging parameter URL, developers are presented with a simple HTML debugging form. This interface allows easy access to the various configurable options of the layer (Figure X.14). These include several new features developed over the past year, all accessible via the debugger or by adding the correct parameters to the script's URL. These are summarized below.

The script can output with either KML 2.0 or KML 2.1 (including, in part, support of 3D Models, timestamps and time spans as an attribute of any feature, and the ability to specify geographic regions of feature visibility). As stated before, this differentiation determines if simple symbolic polygons are created, or if 3D models are used. Because of long load-times if the models are individually requested from the server, they can be packaged up in a “KMZ” or zipped KML file.

By default, 3D models are always bundled up in a KMZ. However, the other image files and GIS layers referenced in the JR GE layer can either be individually loaded on-demand from the webserver, or also packaged up as a complete snapshot of the layer. This allows for viewing of the layer as a single snapshot in time, without need for internet access. It is also particularly useful for giving demonstrations, as loading images and GIS layers on demand from out webservers can be slow. The drawback to this option is that the layer does not self-update every 5 minutes as with the “live” version, and the download size is considerably larger (15MB vs. 30-300KB).

There is also the previously described option to encode all current sensor measurement value as custom XML tags within the KML file. This retains machine-readability for the KML file, whereas otherwise the only reference to current measurement values is buried within the presentational HTML of the description balloon.

A few other options available via the debugging interface, and are primarily useful for problem solving, these include: outputting KML to the screen for viewing, the target file for download, or both; stopping the connection to the DMS MySQL database for current sensor measurements (speeds debugging and loading of the file when off-site or when the database is down); cache override  mechanisms which prevent public requests for the JR GE layer from generating a new KML file more than every 5 minutes; and finally, halting the bundling of live and archived images from webcams and database.

Actual Photo Sketchup Model (KML 2.1) Simple Polygon (KML 1.0)
Figure X.3a Figure X.3b Figure X.3c
Figure X.3d Figure X.3e Figure X.3f
Figure X.3g Figure X.3h Figure X.3i
Figure X.3: Photos, 3D Models and Simple Polygons. Columns, left to right: Actual photo, Google Sketchup “Collada” model, and simple polygon representation. Rows, from top to bottom: HOBO sensored Nest Box, Networked Info-mechanical System (NIMS) – mobile sensor platform on high-tension cables, and CMS Micro-climate station.


Figure X.5

Figure X.5: Detailed Description Bubbles. Clicking on a sensor location or webcam provides current measurements and snapshots. Miniature trendlines ("sparklines") show the 24-hour measurement trends. Dates of data availability are also provided.


Figure X.6

Figure X.6: Access to all images in the time-frame. Yellow bars with black outlines represent images displayed next to the graph; thin black bars denote un-displayed images. Clicking on any vertical bar brings a pop-up window with the corresponding image. In this case the pop-up window shows a baby bird that hatched between the two adjacent displayed images.


Figure X.7Figure X.7: Inside temperature and humidity for two nestboxes graphed with synchronized time-lapse webcams images from three other nestboxes. Clicking on a thumbnail provides a pop-up enlargement of that image. Clicking on the date of the image will refocus the entire page to view sensor data and images from that day. Users can also pan forwards and backwards in time, and well as zoom in and out.

Figure X.8 Figure X.9
Figure X.8: 2-meter contour and DEM model Figure X.9: GIS aspect model
Figure X.10 Figure X.11
Figure X.10: Four-inch aerial photographs Figure X.11: Interactive vegetation map
Figure X.12 Figure X.13
Figure X.12: Kriging model of florometery on Lake Fulmor; taken by NAMOS and Roboduck

Figure X.13: One of 24 frames available for the Cold Air Transect time-animation


Figure X.14

Figure X.14: Interactive vegetation map with represented photos and botanical links


Figure X.15

Figure X.15: Debugging interface to the James Reserve JR GE layer



Future Directions

Experimental transect

In order to better model belowground conditions, an experimental transect will be established to the southwest of the AMARSS transect with leaf litter and water content manipulations.  Specifically, three areas of dry soil and three of wet soil conditions will be maintained by inserting rigid cylinders 30 cm in diameter and 36 cm deep (five gallon plastic buckets) in the soil.  The buckets will have their bottoms removed for the wet treatment and can be capped to prevent precipitation from wetting the soil for the dry soil treatment.  Two buckets (one wet, one dry) of each bare soil, pine litter, and oak leaf litter will be established.  This will give us a dry standard and a wet standard for each of three litter types to use for comparisons of natural conditions.  Each bucket will have a heat flux plate placed at 8 cm below the mineral soil and a thermocouple at 8 and 2 cm depth and at the soil surface.  All sensors will be attached to a Campbell Scientific datalogger.  A manually placed IR sensor will be moved between locations to measure the surface temperature of each cylinder of soil to compare it to the temperature measured by the surface thermocouple and to that on the NIMS RD shuttle over the AMARSS transect.  Additionally, a small photodiode, calibrated against a precision radiometer, will be placed at the soil surface in each treatment and will be used as a radiometer (and corrected with nearby precision radiometric measurements).

Retirement of the original CMS

With CMS2 working reliably and CMS requiring heroic measures to be kept alive, we plan to remove the older system within the coming months.  The lessons learned from this one deployment have touched or influenced almost every mote-based deployment and area of research in CENS.  Areas ranging from sensor calibration and predictive failure to something as mundane as deciding what tools are required to do a deployment have all resulted because of the longevity of the deployment.

Development of the CENS Data Center

During the past five years it has become clear that many of the lessons learned through deployments at the James Reserve and other locations is that much information has been lost due to worker (staff, GSRs, students) turnover.  To address this issue, a CENS Data Center has been established with the goal of preserving institutional memory.  In the coming and future years deployments done at the Reserve will be documented so that future programs and research efforts can benefit more easily from what we have learned.


The next big step for the development of the James Reserve Google Earth layer and Emissary as a whole is the transition to Sensorbase as a sensor data source. Adapting the JR GE layer script to use this new database will facilitate the accessibility of other CENS systems in Google Earth. Furthermore, the inclusion of on-the-fly GIS analysis and modelling of sensor data will provide serious scientific utility to the interface: users will be able to request Kriging models of various measurements across space and time, and visualize the ebb and flow of these environmental factors from within Google Earth.

As is, the JR GE layer can facilitate the study of several biological phenomena at the James Reserve. Two such potential biological studies include the investigation of micro-climate preference in avian nesting biology, and the effects of micro-climate on the springtime emergence and distribution of plant communities. The JR GE layer also provides great promise as an accessible education tool for scientific outreach projects seeking to teach K-12 students the principles of science using real, live and compelling data and multimedia.