Final Report: New Technologies for Rapid Water Quality and Bioassessment of Agricultural Streams by Citizen ScientistsEPA Grant Number: SU835518
Title: New Technologies for Rapid Water Quality and Bioassessment of Agricultural Streams by Citizen Scientists
Investigators: Durtsche, Richard , Frost, Joshua , Guerrero, Josue , McCombs, Brad , Moses, Julie , Rice, Alexus , Shields, Nathan , Steinitz-Kannan, Miriam , Walker, Madeline
Institution: Northern Kentucky University
EPA Project Officer: Levinson, Barbara
Project Period: August 15, 2013 through August 14, 2014
Project Amount: $15,000
RFA: P3 Awards: A National Student Design Competition for Sustainability Focusing on People, Prosperity and the Planet (2013) RFA Text | Recipients Lists
Research Category: P3 Challenge Area - Agriculture , P3 Challenge Area - Water , Pollution Prevention/Sustainable Development , P3 Awards , Sustainability
The goals of the phase I portion of this P3 grant were to create the prototypes of mobile technologies for a range of users, including 4-12 students, STEM undergraduates, citizen scientists, and professionals for the collection of credible data that can be put into a database for use in the bioassessment of headwater and wadeable streams. Water is one of the greatest resources on the planet, and water quality is essential for the functioning of organisms and ecosystems. Agricultural alteration to the landscape often results in structural changes to streams and waterways. Agricultural runoff and pollutants, including sediment, and chemicals used in agricultural practices can modify the quality of the water. Streams and rivers are important drainage conduits of any watershed, as aquatic ecosystems can house a great diversity of organisms. They are integral parts of many human populations as they serve as the main source of drinking water. Bioassessment and evaluation of habitat and physicochemical parameters of streams and rivers are an important mechanism to monitor the water quality and health of watersheds that are impacted by agricultural and other anthropogenic practices. Typically, bioassessment of stream quality is measured in one of three different metrics, through the characterization of the biodiversity of the macroinvertebrates, the algae and bacteria, or the fish. Indices of biotic integrity often are used to evaluate species that are present. An index can be calculated based on the tolerances of organisms to contaminants and the types or assemblages of organisms present, the measures of various structure characteristics of streams, or the range of physicochemical values of the water, to determine if the water and stream ecosystem is of good quality or laden with pollutants and of poor quality. These indices can be used in the management of water resources across regions. However, the number of streams sampled is limited to the number of researchers available for an area, the degree of detail taken at each sampling site, and the time required to survey and record the data for the stream in question. Intensive monitoring for biotic parameters of all streams across a large geographic scale with measures of additional abiotic stream quality features impacted by agriculture (i.e., erosion, stream sinuosity, vegetative cover, stream structure and substrate composition) often is logistically impractical for ecological researchers. Citizen science (CS) offers a solution, while engaging and educating the public. The abundance of watershed streams potentially impacted by agricultural runoff or pollutants, and the interest by the public in understanding the quality of water in their area offer an opportunity to meet the public needs with the bioassessment and water quality mobile technologies we are developing. The educational opportunities for students and the willingness of CS organizations to assist in this process provide for a productive collaboration. CS can function as an indispensable means to “crowd-source” ecological tasks or establish community-based monitoring activities. With new technologies and methods that engage learners, CS can push the envelope on what ecologists can achieve and expand the potential for spatial ecological research while supplementing existing local research programs. The resulting data collected can be viewed as a public good because such collaborations support public participation in science and Earth stewardship. Thus, the use of emerging technologies by citizen scientists and researchers can span the gap of credible data collection while expanding the frontiers of ecological research and public engagement. This P3 grant has allowed for the redesign of our recently developed educational water quality app to the Water Quality Pro app to improve the ability to quantify water quality parameters and stream bioassessment, with the goal of sustained development of such technologies through testing and quality control for use by students, citizen scientists, and professionals alike on a global basis. In conjunction with research and design of this app, a goal of the P3 grant is to advance the design and construction of a Web active database for storage of credible data that can be accessed by professional, educational, and citizen science groups alike. The credibility of the data will result from the planning, design and testing of “smart filters” that will be used in the app itself and through the database to assist the user in the correct identification of organisms and to alert the user if the data appear invalid. To enhance global use of these technologies, the companion website and the Water Quality Pro app will be available in multiple languages, including English and Spanish. We also are partnering with Engineers without Borders and the engineers will be able to use our app throughout the world where they have water engineering projects from Ecuador to Uganda. The website will provide additional information about stream health, water quality parameters, and bioassessment. It will be where data can be retrieved from the open access database or linked to larger state, national, and international databases. Summary statistics and some graphing will be available from the website, and this will be a source for social media links (Facebook, Twitter, etc.). In addition, the development of educational videos and training workshops are designed to ensure students and CS have a complete command of the instrumentation and software interface. These mobile technologies, will eventually stimulate product lines of small businesses that focus on environmental monitoring or sustainable use of the environment.
This EPA P3 grant has stimulated the planning for the development of an advanced mobile app, Water Quality Pro (WQP), that will meet the needs of credible stream assessment by a range of people from the novice to the professional. This app will be modularized and include a variety of features that will make it efficient for the collection of data on stream quality, facilitative for those needing assistance with taxonomic identifications, enhanced to auto-locate the sampling site based on satellite GPS sensors, and integrated to such a degree that stream assessment can be made from a single (e.g., chemical parameters) to a suite of monitoring protocols (e.g., benthic macroinvertebrates, habitat quality, fish, algae and bacteria). Sampling reports can be printed, saved as .pdf or .csv files, or uploaded to our open access database linked to our companion website http://WaterQualityPROtection.nku.edu when Internet access is available. Smart filters in both the app and the database will flag entries outside the range of standard parameters and query users for information clarification or correction to increase the credibility of the data available from multiple users. Open API interface connectivity will allow any data collected to be transferred to other databases, be they local watershed CS databases, agency databases (e.g., KY DOW, US EPA, USGS), or federally sponsored international databases (e.g., DataONE). The companion website will provide information about stream quality and measures of stream assessment. It will host the database for access to records and for descriptive statistics, graphing and data summary for a variety of data fields. Social networking access (e.g., Facebook, Twitter, etc.) also will be accessible from this website, and educational activities, gaming and educational links will make science literacy a fun and unexpected benefit of website usage.
Once we established the design of the WQP and companion website, we invited professional stream biologists, watershed watch leaders, agency personnel, and college and university professors to a day-long focus group to get their feedback on our ideas and to make changes according to what they found important and essential in stream assessment. Their overall assessment of our program user interfaces was extremely positive. Their feedback was honest and tremendously useful in making the modifications to our user interfaces and approach to the app design so that it will function to meet the needs of users across a range of technical skill sets. Features added to aid in credibility included a section on the methods used for various types of stream assessment, the last recorded date of instrument calibration, and a record of the chain of custody for those samples that are passed on to a lab for further analysis. With their help, we determined how many photo vouchers would be adequate for the sampling site and for any invertebrates collected. The group also was helpful in identifying options that should be available based on the user’s level of expertise. A toggle option to include or exclude photo vouchers was suggested to speed up data collection for those professional biologists that know their taxa well enough, although the option is still available. Various forms of physicochemical parameters were recommended for inclusion in the water parameters module. As a result of these requests and as a mechanism to make the app more useful on a global basis, we added user-defined options for any of the measures taken. At least for benthic macroinvertebrate bioassessment, our app user interface profiles were presented so that identification could be recorded at the family taxonomic level. This was a vast improvement over our educational version of the water quality app that only goes to the order level. To increase credibility for a novice or CS, and for enhancing science literacy we designed several options to aid in correctly identifying macroinvertebrate taxa. We introduced a polychotomous key (as opposed to a dichotomous key) based on illustrations for quickly reducing the potential taxa from which to choose. For example, if the user selected the tails button, a series of illustrated options would simultaneously appear on the app, including: no tails, one tail, two tails, three tails, short tails, long tails, paddle-like tails, feathery tails, and based on the option selected, the potential number of taxa the user would have to choose from to make the identification could be reduced. The more anatomical body features the user covers with the polychotomous key, the smaller the list of possible taxa for identification. Another method was to have the user take a voucher photo of every invertebrate taxa recorded for the first encounter of that taxa on a sampling record (site location). When the camera option comes up for a taxa selected, a stencil of the body shape would appear in the viewfinder. If the invertebrate did not roughly match the stencil, then that would be a passive credibility filter that would alert the user to select different taxa. Our last and most innovative approach for a credibility filter is with edge detection. This approach has the user take a picture of the invertebrate. Behind the scenes, this image then is processed within the app to create an edge or outline of the features of the body shape and structure. This edge image then is compared against a series of known taxonomic outlines within a mathematical algorithm to find the taxa that share the greatest percentage of edge space. When completed, the filter produces a short list of taxa ranked by greatest possible likelihood of similarity to the invertebrate under question from which the user then can select the correct taxa. The process is similar to facial recognition with humans, but this technology does not lend itself to invertebrates. Our efforts continue to refine and improve the edge detection for use with the WQP app, and we plan to incorporate a similar function into the database that can have greater computing power and use machine learning to increase identification credibility and feedback for users submitting photo vouchers of taxa recorded.
Phase I of the EPA P3 grant has helped us to refine our design features of the Water Quality Pro (WQP) app to best fit the needs of a range of users from novice citizen scientists to professional biologists. After focus group input, the modified WQP app user interface wireframes are ready for the actual programming to complete three of the six modules of the app, including the Site Profile, Water Parameters, and Benthic Macroinvertebrates. These templates set the stage for more rapidly developing the user interfaces and wireframes for the programming of the other three modules for bioassessment of fish, algae and bacteria, and habitat quality. We have explored a variety of credibility smart filters, both passive and active, and means by which users can increase their accuracy in taxonomic identifications of macroinvertebrates, and have settled on the development of active edge detection (behind the scenes mathematical algorithms) that match a photo of the invertebrate against internal images to give a ranking of the most likely taxa for identification (like facial recognition). This feature in conjunction with polychotomous keys will be used to allow the researcher to identify taxa to the family level. While these concepts have been tested on a small scale, they still need to be evaluated with taxa on a global scale. Polychotomous keys will still need to be created for family level identification within invertebrate orders. A site report from the WQP app for the first three modules of stream conditions will include: (1) a site profile autopopulated based on satellite GPS sensors, three photos of the stream site (upstream right bank, downstream right bank, and over the stream – substrate), field site location and site number has been selected based on perimeters around the GPS location, and users (CS groups, self identified) selected based on geographic location or added manually; (2) water parameters for a range of measures (pH, water temperature, conductivity, turbidity, dissolved oxygen, TDS, BOD, E. coli, fecal coliforms, water hardness, phosphates, nitrates), which include options of recording methods of collection, instrument calibration dates, and chain of custody for additional laboratory analyses, and a water quality index measure based on these parameters; and (3) the macroinvertebrates found at the site with abundances recorded to the family level, including the potential for two voucher photos of each taxa recorded, and a bioassessment index based on the abundance of a taxa and its index value. Reports can be printed, saved as .pdf or .csv files, or uploaded to the interactive open access database and companion website (http://WaterQualityPROtection.nku.edu). The website will allow access to the database for transferring data and data manipulation. It also will act as a conduit for environmental education about stream health and monitoring, social media interactions (Facebook and Twitter), and educational gaming to improve science literacy. Phase II of the P3 grant will bring the WQP app to market with three additional modules for stream assessment (fish, algae and bacteria, and habitat quality). Additional machine intelligence smart filters will be added to the database for increased credibility of taxonomic identification. The WQP app also will be produced for both iOS (iPhone and iPad) and Android users to appeal to a global market, and both the app and the website will be in multiple languages, initially English and Spanish. Field camps already have been run to show people the use of these technologies while collecting stream data, and will continue with both onsite and Web-based workshops. Video tutorials have been produced to help teach the use of the app and gain website understanding of the app function.