Lead Investigators 

Jeffrey Goldman, Deborah Estrin, Eric Graham

Overview

Many research teams aim to quantify and forecast changes in biodiversity in at multiple spatial scales, and to understand the intrinsic dynamics of biodiversity and its responses to anthropogenic drivers of change. One approaches to this goal involves implementing standard biodiversity monitoring protocols for data collection at field sites. Achieving the goal requires comparing high-quality data among different sites, which, in turn, requires strict adherence to standard practices and protocols by those collecting data. Therefore, tools that help technicians follow standards and have built-in quality control functions can help improve data quality. Furthermore given the distributed nature of much of this work, tools that facilitate rapid and robust reporting of data to shared databases can be an important management device for ensuring compliance with protocols and making data available in the most timely fashion. Many groups struggle with these issues. Data are often recorded first on paper and then transcribed to spreadsheets, before being uploaded to a data store. This practice, in addition to being time consuming, can result in transcription errors, incompatibilities in data formats, and data loss.

Approach

We are leveraging the emergency of extremely powerful handheld computers that incorporate low-power processors, wireless communication capabilities, and flexible sensor interfaces, devices which currently are widely used in personal and industrial applications. These personal data assistants (PDAs) and cellular telephones are merging into so-called smart phones, which are programmable and increasingly contain integrated sensors, such as cameras, microphones, and global positioning satellite (GPS) receivers. To date, environmental monitoring efforts have incorporated these sophisticated devices in basic ways. Over time, they will create, in effect, a new type of mobile monitoring system.

Although many environmental data sets are still initially recorded by hand on paper before being encoded digitally, rugged handheld computers are increasingly being used as a primary data collection tool. The use of such devices eliminates a manual (and often error-prone) paper-to-digital encoding step and also improves data quality in the following ways:

• Prompting a user to enter both data and metadata according to standard protocols
• Automating certain routine tasks (e.g., time and date stamping)
• Checking data validity (e.g., using drop-down menus or pre-determined rules)
• Automatically geo-referencing data
• Enabling users to record contextual information, such as photographs and voice annotations, using integrated cameras and microphones
• Uploading data regularly and automatically via wireless connections (e.g., cellular or WiFi) to central data stores.

As technology evolves in the years ahead, additional capabilities will be available. For instance, improvements in processing and storage will enable automatic species recognition and adaptive sampling approaches. As wireless internet access expands to remote monitoring locations, technicians in the field will be able to communicate with more senior personnel if problems arise, coordinate sampling among distributed groups, and access remote computing resources. Finally, over time more sensors will be available in small rugged packages with plug-and-play interfaces to handheld computers, creating the opportunity to collect more data and to do so extremely efficiently.

We are investigating ways to use this technology to improve the quality of data collected by humans in the field and making the process more efficient.

Systems/Experiments

Recently, CENS began working with the Tropical Ecology Assessment and Monitoring (TEAM) network, a group of ecologists seeking to quantity tropical biodiversity and understand how it is affected by changes in land use and climate. The initial goal of the collaboration was to explore the use of handheld computers during the data collection process to assist with promulgating standards and improving data quality throughout the network. The approach involved the development of prototype software for a handheld computer platform that is capable of being used in the field to collect data according to the TEAM protocol for monitoring butterflies. Software was developed for the Nokia N80 mobile phone platform and involved the Python programming language

Accomplishments

Over the course of about six months, CENS staff and students developed two versions of the software, which featured a local database, simple user interface and navigation, methods for directly entering all of the data associated with the protocol, and the ability to link photographs taken by the computer's integrated camera to specific records in the database. Initial research into the use of the camera to read barcodes did not lead to a suitable solution, but very early tests with a dedicated wireless laser barcode scanner are promising. The prototype was demonstrated and tested in the field at the second TEAM network meeting in Manaus, Brazil from 1-5 December 2006. Informal feedback from the meeting indicates that members of the network are enthusiastic about incorporating handheld computers into their data collection practices. This proposal lays out an approach to developing similar applications for actual use in data collection

Future Directions

Move from prototype to robust tools for biodiversity monitoring

If successful in securing external support, we will select several TEAM protocols to develop into workable applications. Candidates include bird and butterfly sampling and forest tree censuses. We will analyze the protocols to generalize features that can be developed and applied across a number of protocol applications. The software will be developed in collaboration with TEAM personnel via a requirements definition stage.

Future work will move away from the Nokia platform and onto an off the shelf and less expensive platforms using the Windows Mobile operating system. Devices will be evaluated on the basis of cost, durability, and built-in features or expandability.

Research on input methods, user interfaces, and hardware options

Through our prototyping work and discussions with TEAM personnel, we have identified several areas in which data collection via handheld computers can be improved. We will be conducting feasibility studies of the following topics to assess whether they are suitable for incorporation in EcoPDAs, and, if they are, to incorporate them into software.

• Map interfaces
• Rapid species name input methods
• Use of Barcodes and radio frequency identification (RFID) tags as data input methods
• Canopy openness measurement via hemispherical imagery
• Determination of appropriate hardware and accessories for rugged and low-cost devices

People

Faculty:

• Deborah Estrin

Staff:

• Jeffrey Goldman
• Eric Graham
• Yeung Lam
• Eric Yuen

Graduate Students:

• Chris Mar
• Andrew Parker

External Research Partnerships

Tropical Ecology Assessment and Monitoring (TEAM) program, Conservation International