Crowdsourcing for Land Cover and Land Use Change Research and Applications

Unprecedented amount of remote sensing data is being collected and archived by the different space agencies. Efficient use of these datasets, especially for monitoring of land cover and land use change, require contextual information in the form of LULC ground truth data. Ground truth data is utilized for training supervised classifiers or to interpret outputs from unsupervised classifiers. At present, LULC ground truth data is collected through surveys conducted by individual researchers or government agencies and are often static and not readily accessible by the public. Whereas, spectral and spatial resolution of satellite imagery is continually improving, the ability of researchers to utilize such data for LULC mapping is hampered by the lack of easily accessible and up-to-date ground truth datasets. We will discuss a framework that has been developed to effectively utilize crowdsourcing of LULC ground truth data and seamlessly interface this database with cloud based technologies for classification of satellite imagery. A mobile application based on Open Data Kit, designed for use by general public, guides the user though the process of collecting LULC ground truth data, including geolocation. This data is exported to Google Fusion Tables which can then be interfaced with classification algorithms available through Google Earth Engine framework. Near-real time applications of this framework will be presented. Training modules being developed to enable incorporation of hands-on activities into K-12 and higher education curriculum will also be presented.