Spatial Data

Home , GeoServer, ID (software), Marble (software), NASA WorldWind, OpenLayers

big data and cognitive computing

Article Development of Framework for Aggregation and Visualization of Three-Dimensional (3D) Spatial Data

Mihal Miu 1, Xiaokun Zhang 2, M. Ali Akber Dewan 2 and Junye Wang 3,* ID

1 The Chang School of Continuing Education, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, Canada; [email protected] 2 School of Computing & Information Systems, Athabasca University, 10011, 109 Street, Edmonton, AB T5J 3S8, Canada; [email protected] (X.Z.); [email protected] (M.A.A.D.) 3 Faculty of Science and Technology, Athabasca University, 10011, 109 Street, Edmonton, AB T5J 3S8, Canada * Correspondence: [email protected]; Tel.: +1-780-394-4883

Received: 17 February 2018; Accepted: 19 March 2018; Published: 21 March 2018

Abstract: Geospatial information plays an important role in environmental modelling, resource management, business operations, and government policy. However, very little or no commonality between formats of various geospatial data has led to difficulties in utilizing the available geospatial information. These disparate data sources must be aggregated before further extraction and analysis may be performed. The objective of this paper is to develop a framework called PlaniSphere, which aggregates various geospatial datasets, synthesizes raw data, and allows for third party customizations of the software. PlaniSphere uses NASA World Wind to access remote data and map servers using Web Map Service (WMS) as the underlying protocol that supports service-oriented architecture (SOA). The results show that PlaniSphere can aggregate and parses files that reside in local storage and conforms to the following formats: GeoTIFF, ESRI shape files, and KML. Spatial data retrieved using WMS from the Internet can create geospatial data sets (map data) from multiple sources, regardless of who the data providers are. The plug-in function of this framework can be expanded for wider uses, such as aggregating and fusing geospatial data from different data sources, by providing customizations to serve future uses, which the capacity of the commercial ESRI ArcGIS software is limited to add libraries and tools due to its closed-source architectures and proprietary data structures. Analysis and increasing availability of geo-referenced data may provide an effective way to manage spatial information by using large-scale storage, multidimensional data management, and Online Analytical Processing (OLAP) capabilities in one system.

Keywords: spatial data fusion; environmental modelling; geospatial information; data mapping; software; big data

1. Introduction Recent developments in remote sensing data, monitoring networks, and geographic information systems (GIS), like movements of open access and open data lead to the unprecedented growth of data [1–3]. There currently exist many large repositories and cloud storage of analytical and subject-oriented databases, such as national censuses, statistical frameworks of the UN System of Environmental-Economic Accounting [4–10]. Big Data technologies together with cloud computing create new opportunities for data intensive research and massive data processing from earth remote sensing in a multi-disciplinary environmental domain. Sustainable development requires multidisciplinary approaches to process Big Data, including climate change, environmental health, water resources, land use and land cover, web technologies, digital earth, and so on [8–10]. This also requires dealing with big data complexity, such as large scale, structured and unstructured data, online

Big Data Cogn. Comput. 2018, 2, 9; doi:10.3390/bdcc2020009 www.mdpi.com/journal/bdcc Big Data Cogn. Comput. 2018, 2, 9 2 of 14 and real-time interaction and processing, cross formats, and so on [9–12]. This makes data management and processing very difficult using conventional methods in a reasonable time although big data offers an environment rich information and opportunities to explore the data in more precise ways. It is still challenging how to process such big amount of data to extract useful information of economic and environmental sustainability because of complexity of the big data obtained from various sources and scales. Several recent studies focus on using big data and cloud computing techniques in environmental data analysis. Lokers et al. [8] presented a theoretical framework to analyze data-intensive cases related to big data usage in agro-environmental domain. This research indicates that data-intensive research evolves around capturing huge heterogeneity of interdisciplinary data and around creating trust between data providers and data users. Votolo et al. [9] gave an overview of currently available implementations that are related to web-based technologies for processing large and heterogeneous datasets and discuss their relevance within the context of environmental data processing, simulation, and prediction. Cavallaroa et al. [13] presented how big data analytics take advantages of techniques from the fields of data mining, machine learning, or statistics, with a focus on analyzing big data in remote sensing with modern technologies. Kussul et al. [14] proposed a framework for solving the large-scale classification and area estimation problems in the remote sensing domain based on deep learning. In this framework, self-organizing maps, multi-layer perceptron, and geospatial analysis were used for segmentation, classification and fusion, and for some post-processing, respectively. Wolfert et al. [10] presented a review on how big data is being used to provide predictive insights in farming operations, drive real-time operational decisions, and redesign business processes for game-changing business models. The authors of this paper believe that the big data will cause major shifts in roles and power relations among different players in the current food supply chain networks. Yang et al. [15] investigated how cloud computing can be used to address big data challenges, especially in climate studies, geospatial knowledge mining, land cover simulation, and dust storm modelling. Fritz et al. [16] described the global land cover and land use reference data derived from the Geo-Wiki crowdsourcing platform. These global datasets provide information on human impact, land cover disagreement, wilderness, and land cover and land use. Geospatial and environmental data include not only maps of locations of land usage, but also multiple attributes, such as socioeconomic data extracted from state and national census, and this is also a type of big data. These data sets are heterogeneous across various data sources and their file formats are inconsistent, since different suppliers tend to use proprietary data and file formats. They may have static or dynamic characteristics. Thus, there may be little or no commonality between formats used. This has led to increased challenges in capturing and analyzing such spatial data. Advancements in computing, visualization, and Internet technologies [17–20] have allowed two-dimensional (2D) and three-dimensional (3D) images to aggregate map layers into one graphical representation. Aggregation of different data sources can generate a new map with higher levels of informational detail that is not possible by any of the individual systems [18]. By sharing and aggregating this information over the Internet, enhanced accessibility and time responses can effectively improve the interpretation of used information/data sources when compared to conventional distribution of maps or character based online systems [21]. Aside from environmental modelling, using maps to visualize data can enable better interpretation of complex geographical data [22], identify patterns, and help in planning resource distribution for policy and decision making [23]. Soulard et al. [24] reported the harmonization of forest disturbance datasets using multiple data sources. Villarreal et al. [25] used three vegetation land cover maps from different geographic scales to improve the accuracy of wildlife habitat models. Therefore, aggregating, in the environment and resource management, provides a visual assessment for investigating spatial distribution of agricultural and forest production and pollution diffusion, with potential associations and their underlying causes [26] and can help the calibrations and classifications of land uses from remote sensing images. The accurate modelling of spatial data requires the definition of two kinds of metadata: Big Data Cogn. Comput. 2018, 2, 9 3 of 14

(a) warehoused metadata that models data structures and maintains integrated data from multiple data sources, and (b) aggregation metadata that speciﬁes how the warehoused data should be aggregated to meet analysis needs of decision makers. New mapping technologies, such as Google Earth, offer free satellite and aerial photos of most of the Earth’s land surface, which has led to an increased uptake in mapping technology for uses that are relevant to environmental health [27–31]. Xiong et al. [30] developed an automated cropland mapping of algorithms to aggregate continental Africa using Google Earth Engine cloud computing. Google Earth Engine cloud computing is now providing a platform for data production and analysis using Python and JavaScrip [31]. However, Google Earth does not allow for user customization because it provides no external plug-in infrastructure. Developments in GIS have made mapping of spatial data more commonplace and is being used in a wide range of applications. Current commercial GIS vendors use closed-source architectures and proprietary data structures, limiting their ability to add libraries and tools, including those for transient watershed modelling, and limiting geospatial data to their own preferred sources. The vendor ESRI, through its ArcGIS platform [32], for example, creates a GIS infrastructure that provides users with servers, clients, and geospatial data. However, ArcGIS does not provide a plug-in functionality for the use of aggregating and fusing geospatial data from different data sources. Because of large scale of data, some formats cannot be treated in ArcGIS. Therefore, it is desirable to provide a plug-in infrastructure that allows users to perform their own customization. If multiple map services for the same geographical region could be aggregated, the resulting map produced would include more spatial information. Therefore, the aggregation and fusion of spatial data in GIS is a useful feature in the environmental sciences, although this only marginally presents in current commercial GIS, primarily through ad-hoc solutions [33–35]. The successful use of this data depends to a large extent on the user’s ability to access, integrate, and analyze the data [36–38]. The objective of this paper is to develop a framework called PlaniSphere [39] for the aggregation and visualization of various map sources. It allows for synthesizing raw data of geometric components and aggregate non-spatial information contained in a data warehouse associated to those components, deﬁned in GIS fact tables. Aggregation of map data can produce new fusion maps with a higher level of detail than what is currently available.

2. Methodology

2.1. Conceptual Model of Data Aggregations A software tool that aggregates and fuses various geospatial data sources needs to reconcile differences from multiple data formats and service providers. Alternatives to commercial vendors are open-source suppliers, but existing open-source suppliers, such as GeoServer [40] and MapServer [41], provide fragmented solutions for geospatial data, services, and infrastructure. Usually, these open-source projects are independent of each other, although GeoServer and MapServer will allow for clients to connect using a Web Map Service (WMS) protocol [42] from the OpenGeospatial Consortium (OGC) [34,43]. Techniques for performing complex analysis of information stored in a data and map warehouse has been developed, including such systems as the OGC [34,43] and Online Analytical Processing (OLAP) [35,44,45]. This provides specifications of interoperability among vendors [28,29]. The software tools that this project creates will facilitate the usage of various geospatial data, regardless of their format and location. Figure1 provides a high-level overview of the proposed solution that is implemented in this project. Geospatial data can be provided from local storage or from remote servers through the Internet using WMS. The software tool can aggregate files from local storage that conforms to widely used file formats, such as: GeoTIFF files, ESRI shape files, KML, GeoJSON, and LASer files (Figure1). Big Data Cogn. Comput. 2018, 2, 9 4 of 14 Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16

Figure 1. Various file formats and protocols used by geospatial data vendors/suppliers. Figure 1. Various ﬁle formats and protocols used by geospatial data vendors/suppliers.

The level of integration required would be high and may be beyond the capabilities of open- sourceThe suppliers. level of integration WMS and required WFS have would been be created high and to may aid bein beyondtransferring the capabilities geospatial of data open-source over the suppliers.Internet in WMS PlaniSphere. and WFS The have design been of created modern to aidma inp servers, transferring such geospatialas GeoServer data and over MapServer, the Internet is inbased PlaniSphere. on service-oriented The design architecture of modern (SOA). map servers, NASA suchWorld as Wind GeoServer [46] can and access MapServer, remote map is based servers on service-orientedusing WMS version architecture 1.3, and (SOA).also parses NASA files World that Windreside [ 46in ]local can accessstorage remote and conforms map servers to the using following WMS versionformats: 1.3, GeoTIFF, and also ESRI parses shape files files, that resideand Keyhole in local Markup storage andLanguage conforms (KML). to the WMS following is used formats: as the GeoTIFF,underlying ESRI protocol shape files,that supports and Keyhole SOA. Markup Also, WMS Language supports (KML). the WMS interoperability is used as the between underlying map protocolservers and that clients. supports The SOA. end Also,result WMS of WMS supports is to produce the interoperability a map that is between represented map serversby an image and clients. stored Theas a end png, result gif or of jpg WMS format is to produce[47]. These a map analyzing that is representedtechniques and by an the image increasing stored availability as a png, gif of or geo- jpg formatreferenced [47]. data These facilitates analyzing an techniques effective andway the to increasingmanage spatial availability information of geo-referenced by providing data large-scale facilitates anstorage, effective multidimensional way to manage spatialdata management, information byand providing OLAP querying large-scale capabilities, storage, multidimensional together in one datasystem management, [48–50]. WMS and OLAPis a standa queryingrd widely capabilities, used togetherfor accessing in one and system retrieving [48–50]. maps WMS isfrom a standard remote widelyservers used made for available accessing over and the retrieving Internet. maps WMS from can be remote used serversto retrieve made maps available from public over the and Internet. private WMSservers can via be the used Internet. to retrieve WMS maps versions from 1.1 public and and 1.3 are private widely servers supported via the by Internet. commercial WMS vendors versions and 1.1 andopen-source 1.3 are widely suppliers, supported but WMS by commercialversion 1.3 is vendors not backwards and open-source compatible suppliers, with WMS but version WMS version1.1 [42]. 1.3WMS is not can backwards retrieve geospatial compatible data with from WMS a remote version server 1.1 [ 42without]. WMS knowing can retrieve the originating geospatial datafile(s) from and atheir remote type(s)/format(s), server without but knowing it requires the originatinga server-side file(s) implementation. and their type(s)/format(s), When a user retrieves but it requiresgeospatial a server-sidedata from local implementation. storage, they When may not a user have retrieves a server-side geospatial implementation data from local of WMS, storage, and, they as maya result, not haveWMS a cannot server-side be used. implementation of WMS, and, as a result, WMS cannot be used. 2.2. Graphic User Interface Design 2.2. Graphic User Interface Design PlaniSphere is a desktop application with graphical capabilities. Its functionality is available to the PlaniSphere is a desktop application with graphical capabilities. Its functionality is available to user through a graphical user interface (GUI) (Figure2). Because the NASA World Wind API [ 46] lacks the user through a graphical user interface (GUI) (Figure 2). Because the NASA World Wind API [46] export capabilities and a plug-in infrastructure, the functionality and design of the GUI for PlaniSphere lacks export capabilities and a plug-in infrastructure, the functionality and design of the GUI for is extremely important. It uses a similar GUI design as MS Office 2007 and newer applications. At the PlaniSphere is extremely important. It uses a similar GUI design as MS Office 2007 and newer heart of the plug-in infrastructure is the Plug-In Manager (Figure3). applications. At the heart of the plug-in infrastructure is the Plug-In Manager (Figure 3). PlaniSphere revolves around a main window with a ribbon toolbar. The toolbar presents the core PlaniSphere revolves around a main window with a ribbon toolbar. The toolbar presents the features to the user. Under the ribbon, there is a working area where a map or 3D virtual globe image core features to the user. Under the ribbon, there is a working area where a map or 3D virtual globe is displayed. The optional capability to manipulate any layers within the map is in a tabbed pane to image is displayed. The optional capability to manipulate any layers within the map is in a tabbed the left of the working area. See Figure2 for the general layout of the main window. pane to the left of the working area. See Figure 2 for the general layout of the main window. The ribbon toolbar in PlaniSphere allows for users to load any resources available from any map The ribbon toolbar in PlaniSphere allows for users to load any resources available from any map servers that support the WMS 1.3 protocol. PlaniSphere can connect to multiple servers provided the servers that support the WMS 1.3 protocol. PlaniSphere can connect to multiple servers provided the URL of the server is known. The software loads any geospatial data resource represented by files, such as an ESRI shape file, GeoTIFF, and KML. Each file represents a single layer that can be rendered Big Data Cogn. Comput. 2018, 2, 9 5 of 14

URL of the server is known. The software loads any geospatial data resource represented by files, such asBig anData ESRI Cogn. Comput. shape 2018 file,, 2 GeoTIFF,, x FOR PEER and REVIEW KML. Each file represents a single layer that can be 2 rendered of 16 on theBig map. Data Cogn. A ribbon Comput. 2018 band, 2, x is FOR dedicated PEER REVIEW to“Custom Map Sources” support. It should be noted 2 of 16 that on the map. A ribbon band is dedicated to “Custom Map Sources” support. It should be noted that each time a connection to a known map server is required the user needs to click on the “Add WMS oneach the time map. a connection A ribbon band to a knownis dedicated map serverto “Custom is required Map Sources”the user needs support. to click It should on the be “Add noted WMS that Server” button in the ribbon band. Each file type is then represented by a button. eachServer” time button a connection in the ribbon to a known band. Eachmap serverfile type is isre quiredthen represented the user needs by a button.to click on the “Add WMS Server” button in the ribbon band. Each file type is then represented by a button.

Figure 2. Main graphical user interface (GUI) window. Figure 2. Main graphical user interface (GUI) window. Figure 2. Main graphical user interface (GUI) window.

Figure 3. PlaniSphere with the Plug-in Manager and Sample Plug-ins (JRE SysInfo and NMEA Parser). Figure 3. PlaniSphere with the Plug-in Manager and Sample Plug-ins (JRE SysInfo and NMEA Parser). FigurePlaniSphere 3. PlaniSphere is built with from the Plug-in the ground Manager up and Samplepossesses Plug-ins several (JRE advantages SysInfo and over NMEA existing Parser). GIS packages.PlaniSphere Thus, PlaniSphereis built from supports the ground many up well-known and possesses open several standards, advantages such as over WMS, existing WFS, andGIS PlaniSpherepackages.file formats Thus, (KML, is builtPlaniSphere GeoTIFF, from theGeoJSON,supports ground many LAS, up andwell-known and ESRI possesses Shape open Files). severalstandards, Table advantages 1such lists asbasic WMS, over files WFS, and existing dataand GIS packages.filesources formats Thus, that (KML, PlaniSphereare provided GeoTIFF, supports by GeoJSON, PlaniSphere. many LAS, well-known The and secondESRI Shape openadvantage Files). standards, Tableis that 1 such PlaniSpherelists asbasic WMS, files is WFS,andplatform data and file formatssourcesindependent. (KML, that GeoTIFF, are It is provided implemented GeoJSON, by PlaniSphere. in LAS, Java and and ESRI runsThe Shapesecondon any Files).desktopadvantage Table (Linux, is1 thatlists macOS, PlaniSphere basic MS files Windows) and is platform data that sources independent.supports Java Itversion is implemented 8 or higher. in The Java third and advantageruns on any is desktopthat PlaniSphere (Linux, macOS, renders MS GIS Windows) data using that the that are provided by PlaniSphere. The second advantage is that PlaniSphere is platform independent. supports Java version 8 or higher. The third advantage is that PlaniSphere renders GIS data using the It is implemented in Java and runs on any desktop (Linux, macOS, MS Windows) that supports Java Big Data Cogn. Comput. 2018, 2, 9 6 of 14 version 8 or higher. The third advantage is that PlaniSphere renders GIS data using the World Geodetic System 1984 (WGS 84) with projections (Round/Spherical, Mercator, Flat-Modified Sinusoidal).

Table 1. Files and Data Sources supported by PlaniSphere.

Native Support (Out of Box Support by Planisphere) 3rd Party Support Open Sources Closed Sources Network/Internet Local Files Proprietary Files and Services WMS GeoJSON SQL (may be supported by creating a plugin) WFS (currently an experimental feature) KML Proprietary ﬁles (may be supported by creating a plugin) Raster (TIFF, GeoTIFF) Custom Web Services (may be supported by creating a plugin) ESRI Shape Files LiDAR (LAS)

2.3. Implementation PlaniSphere allows for users to build their own applications by providing a geographic rendering engine for powering a wide range of projects, from satellite tracking systems to traffic simulators. PlaniSphere is designed for interoperability. It consumes and renders data from any major spatial data source using open standards. Using GeoServer becomes an easy method of connecting existing information and data to virtual globes, such as Earth and NASA World Wind as well as to web-based maps, such as OpenLayers, OSM, Google Maps, and Bing Maps. GeoServer functions as the reference implementation of the Open Geospatial Consortium Web Feature Service standard. WMS and WFS are used as standard protocol for serving (over the Internet) georeferenced map images that a map server generates using data from a GIS database and also implements the Web Coverage Service and Web Processing Service specifications. PlaniSphere will aggregate various map services, as WMS and WFS are a standard that is widely used for accessing maps from various servers and can be used to retrieve maps from public and private servers that are available over the Internet. Geospatial information that resides on local storage (GeoTiff, ESRI shape files, KML files, las files, etc.) may also be used (Figure1). A 3D virtual globe desktop application that provides a rich set of features for displaying and interacting with geospatial data is created based on NASA WorldWind for Java. The 3D virtual globe has the following characteristics:

• a Java SE 1.8 application that can run on any operating system; • the 3D virtual globe is an interactive application that has a low learning curve; • open-standard interfaces to GIS services and databases • capable of rendering in 3D/2D: ESRI shapeﬁles, GeoTIFF, KML, LiDAR; • a plugin framework that can be used to expand PlaniSphere by any third party (Figure4); and, • capable of rendering in 3D/2D high-resolution imagery, terrain and geospatial information from any source using WMS 1.3.

Each geospatial data source (WMS endpoint or ESRI shapefile, GeoTIFF and KML file) creates its own appropriate layer. A layer is a direct or indirect instance of RenderableLayer or TiledImageLayer types. The algorithm is useful in aggregating geospatial data of a graphical nature. The files that compose the geospatial data are maps or content that can be converted into an image. GeoTIFFs are images with geotags; ESRI shapefiles are simplified 2D maps; KML files are XML files that contain information that can be drawn on top of a map. Algorithm 1 may be used interactively by a user in their viewing session. The user can select in real time the area of interest and the layers to be aggregated. The resulting image is generated almost in real time. The user may interactively reorder the layers as needed. Each time the layers are reordered, a new image is generated. Figure5 exhibits an example of aggregating layers in a viewing session. Out of all the layers that are available, only two are visible and used in the aggregation. Only visible layers are aggregated using this algorithm by PlaniSphere. The two visible Big Data Cogn. Comput. 2018, 2, 9 7 of 14

Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16 layers that are involved in the aggregation and fusion are “Bing imagery” and “planet_osm_line”. “Bingimagery” imagery” and is “planet_osm_line”. a base layer, while “Bing “planet_osm_line” imagery” is a base is an layer, overlay while layer. “planet_osm_line” is an overlay layer.

AlgorithmAlgorithm 1: Algorithm 1: Algorithm for Graphical for Graphical Aggregation Aggregation and and Fusion Fusion of of Geospatial Geospatial Data Data refers refers to to Java Javaand and NASA WorldNASA Wind TypesWorld suchWind asTypes RenderableLayer, such as RenderableLayer, TiledImageLayer, TiledImageLayer, etc. [46, 51etc.]. [46,51]. 1. Identify the type of data source, then: 1. Identify the type of data source, then: i. If it is a known file (ESRI shapefile, GeoTIFF, or KML), parse it, and load its content i. If it isinto a known a single file instance (ESRI shapefile, of an appropriate GeoTIFF, layer or KML), type. parse Geospatial it, and data load consumed its content from into a single instancean ESRI of an shapefile, appropriate GeoTIFF, layer type. or GeospatialKML file may data consumedbe represented from anby ESRI instances shapefile, of GeoTIFF,RenderableLayer or KML file maytype. be represented by instances of RenderableLayer type. ii. If it is a WMS endpoint, download the geospatial data into an instance of an ii. If it is a WMS endpoint, download the geospatial data into an instance of an appropriate layer appropriate layer type. Geospatial data produced by WMS endpoints may be type. Geospatial data produced by WMS endpoints may be represented by an instance of represented by an instance of TiledImageLayer type. TiledImageLayer type. 2. Identify categories of layers (base layers, overlay layers). Layer categories can be 2. Identifydetermined categories from of layers the geospatial (base layers, data overlay represented. layers). Base Layer layers categories have a cansolid be background determined or from the geospatialare a datasolid represented. image without Base any layers transparent have a solid sections. background Overlay or arelayers a solid contain image transparent without any transparentsections. sections. Overlay layers contain transparent sections. 3. Order3. Order layers layers so that so overlay that overlay layers layers are always are always on top on of top base of layers. base layers. User intervention User intervention may be may required in this step.be Note:required since in base this layers step. have Note: a solid since background base layers (no transparencies)have a solid onlybackground one base layer(no will be visibletransparencies) after the completion only one of thisbase algorithm. layer will be visible after the completion of this algorithm. 4. Order4. Order layers layers so that so overlay that overlay layers layers are always are always on top on of top base of layers. base layers. User intervention User intervention may be may required in this step.be Note:required since in base this layers step. have Note: a solid since background base layers (no transparencies)have a solid onlybackground one base layer(no will be visibletransparencies) after the completion only one of thisbase algorithm. layer will be visible after the completion of this algorithm.

5. Render5. Render all layers all layers on the on screen the screen based based on the on order the order created created in previous in previous step. step. Always Always render render background layersbackground ﬁrst followed layers by overlay first followed layers. A by consequence overlay layers. of this A consequence step is that any of overlappingthis step is that geospatial any data will beoverlapping overwritten geospatial with the geospatial data will data be fromoverwr theitten currently with renderedthe geospatial layer. data from the currently rendered layer. 6. An image is created as an end result of completing the above steps. 6. An image is created as an end result of completing the above steps.

(a) (b)

Figure 4. Fusion of multiple layers from different WMS servers: (a) LandSat7 photograph of Toronto, (b) OpenStreetMap roadmap, (c) OpenStreetMap map superimposed over a LandSat photograph, and (d) After analyzing using Planisphere. Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16

Figure 4. Fusion of multiple layers from different WMS servers: (a) LandSat7 photograph of Toronto, Big Data Cogn.(b) Comput.OpenStreetMap2018, 2, 9 roadmap, (c) OpenStreetMap map superimposed over a LandSat photograph, 8 of 14 and (d) After analyzing using Planisphere.

Figure 5. A PlaniSphere plug-in demonstrating Online Analytical Processing (OLAP) capabilities by Figure 5. A PlaniSphere plug-in demonstrating Online Analytical Processing (OLAP) capabilities by identifying areas of lower population concentration. Note the blue lines represent an extrapolated identifyingmunicipal areas border of lower generated population by analyzing concentration. population concentration Note the blue on lineseither representside of the anborder. extrapolated municipal border generated by analyzing population concentration on either side of the border. 3. Results and Discussion 3. ResultsAs and a multi-function Discussion platform, PlaniSphere is well supported by data management software such Asas databases a multi-function and GIS platform,platforms through PlaniSphere the NASA is well World supported Wind API by dataframework management and is compatible software such as databaseswith the OGCs and GIS and platforms WMS. Unlike through Google the Earth, NASA Plan Worldisphere Windprovides API a plug-in framework function and for is users compatible to build their own applications. Therefore, it can aggregate data and metadata in one single file using a with the OGCs and WMS. Unlike Google Earth, Planisphere provides a plug-in function for users text format. Plug-in window function allows for users to develop further self-describing data formats to buildfor storage, their own transfer applications. and aggregation Therefore, of 3D spatia it canl data aggregate and metadata data and information metadata using in oneJavaScript single file usingon a demand. text format. Online Plug-in web data window processing function unusually allows allows for for users utilizing to develop with one further or several self-describing datasets data formatsin different for formats, storage, which transfer are assessible and aggregation locally or of online 3D spatial for user data interaction and metadata and create information a web app. using JavaScriptThis enables on demand. interacting Online and web interchanging data processing data with unusually several allows datasets for in utilizing different with data one formats, or several datasetsincluding in different plain text, formats, markup which languages are assessible and binary locally files orand online to easily for create userinteraction interactive visualization and create a web app.of This 3D enablesglobe, map interacting and conventional and interchanging descriptions data of geographical with several features, datasets vector, in different coverages, data and formats, includingsensor plain data. text,By using markup KML, languages the semantic and meaning binary filesof the and data to can easily be createextracted interactive from the visualizationfile itself making it suitable to optimally represent the metadata. The platform enables visualization and of 3D globe, map and conventional descriptions of geographical features, vector, coverages, and mapping of 3D spatial data and interactive mapping applications, such as hovering over a graph, sensor data. By using KML, the semantic meaning of the data can be extracted from the file itself zoom in/out on a particular portion of a map/graph through MapServer and GeoServer. Thus, makingPlaniSphere it suitable may to be optimally used for simple represent purposes, the such metadata. as identifying The platform camp sites enables and bicycle visualization paths, and and mappingfor more of 3D complicated spatial data purposes, and interactive such as mapping vehicle tracking applications, and monitoring such as hovering weather over patterns. a graph, The zoom in/outfollowing on a particular illustrates portion PlaniSphere of a map/graph using five exampl throughes of MapServer landscape and and environmental GeoServer. Thus, applications PlaniSphere maythat be used are being for simplestudied purposes,and/or researched: such as identifying camp sites and bicycle paths, and for more complicatedFigure purposes, 4 shows a such fusion as of vehicle multiple tracking layers from and different monitoring WMS servers. weather First, patterns. Bing Map The provides following illustratesa background PlaniSphere layer using using fivea LandSat examples 7 image of landscape with buildings and environmental and landscape applicationsdetails (Figure that 4a). are An being studiedOpenStreetMap and/or researched: (OSM) of the same area with roads and names WMS layer represents the road network that was used for the second layer (Figure 4b) and it can be easily updated. The geospatial Figure4 shows a fusion of multiple layers from different WMS servers. First, Bing Map provides data used in Figure 4a (LandSat) and Figure 4b (Open Street Map) are provided by two independent a background layer using a LandSat 7 image with buildings and landscape details (Figure4a). An OpenStreetMap (OSM) of the same area with roads and names WMS layer represents the road network that was used for the second layer (Figure4b) and it can be easily updated. The geospatial data used in Figure4a (LandSat) and Figure4b (Open Street Map) are provided by two independent and distinct sources from NASA and a community of mappers, respectively. Then, the user can generate an enhanced map of the same area containing all of the details from the Figure4a,b (buildings, landscape, roads, and names). The aggregation and fusion of the photograph and the road map is the Big Data Cogn. Comput. 2018, 2, 9 9 of 14

Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16 rendered result, as depicted in Figure4c. The third layer is metadata from Wikipedia, which provides a restand web distinct service sources [52], where from NASA points and of interesta community may beof retrievedmappers, forrespective an arealy. ofThen, interest. the user In thiscan case, the areagenerate of interest an enhanced is a neighborhood map of the in same downtown area containing Toronto. all Wikipedia of the details provides from spatial the Figure data using4a,b the Wikipedia(buildings, GeoData landscape, extension roads, [52 and], whichnames). is The a proprietary aggregation standard.and fusion Aof pluginthe photograph was created and the to road consume suchmap a proprietary is the rendered standard. result, This as depic pluginted isin the Figure Wikipedia 4c. The Pluginthird laye Extractor,r is metadata which from is exhibitedWikipedia, at the which provides a rest web service [52], where points of interest may be retrieved for an area of bottom left of Figure4d. This is a unique demonstration of aggregation of map data of different map interest. In this case, the area of interest is a neighborhood in downtown Toronto. Wikipedia provides formats. As far as OGC compliance concerns, the WMS services have made the OSM maps one of the spatial data using the Wikipedia GeoData extension [52], which is a proprietary standard. A plugin manywas layers created in the to consume employed such GIS a proprietary application. standard. Thus, the This multiplicity plugin is the of Wikipedia layers in Plugin WMS allowsExtractor, for the user towhich focus is onexhibited the required at the bottom one (e.g., left buildings,of Figure 4d. names, This is ora unique transportation demonstration here). of In aggregation terms of the of data adaptation,map data OSM of different data is map projected format ins. As Toronto far as OGC transport compliance Grid concerns, so it can the be WMS easily services integrated have tomade the rest of thethe map OSM sources maps inone the of system,the many either layers in in raster the employed or in vector.A GIS application. PlaniSphere Thus, Plug-in, the multiplicity demonstrating of OLAPlayers capabilities in WMS by allows identifying for the boundaries user to focus of municipalitieson the required based one on(e.g., concentrations buildings, names, of points or of interesttransportation (Figure5). here). Municipality In terms of boundaries the data adaptati mayon, be determinedOSM data is projected by the absence in Toronto of transport points of Grid interest. Noteso that it can the be blue easily border integrated represents to the therest extrapolatedof the map sources boundary in the system, of themunicipality. either in raster Theor in OLAP vector.A plugin PlaniSphere Plug-in, demonstrating OLAP capabilities by identifying boundaries of municipalities uses two factors to identify areas of high population concentration. The first factor is the number of based on concentrations of points of interest (Figure 5). Municipality boundaries may be determined points of interest retrieved per square kilometer. The second factor is the distance between points by the absence of points of interest. Note that the blue border represents the extrapolated boundary of interestof the withinmunicipality. the square The kilometerOLAP plugin in question. uses two Thefactors higher to identify the number areas ofof points high population of interest, the closerconcentration. the distance The between first factor neighboring is the number points of points and the of interest higher retrieved the population per square concentration kilometer. The of the area.second Using factor the OLAP is the distance process, between areas ofpoints lower of in concentrationterest within the can square be identified kilometer in question. Figure5. The OLAP enableshigher the end-usersthe number to of perform points of other interest, ad hocthe clos analysiser the ofdistance data in between multiple neighboring dimensions, points through and the Plugin Function,higher such the aspopulation LULC [53 concentration], thereby providing of the ar theea. insightUsing the and OLAP understanding process, areas that theyof lower need for betterconcentration decision-making. can be identified in Figure 5. OLAP enables the end-users to perform other ad hoc Whenanalysis map of data in of othermultiple planetary dimensions, bodies through exists Plugin from Jet Function, Propulsion such Laboratory as LULC [53], and thereby is available providing the insight and understanding that they need for better decision-making. by standard mechanisms, such as WMS, PlaniSphere may consume these maps and render them. When map data of other planetary bodies exists from Jet Propulsion Laboratory and is available Figure6a,b exhibit map data of our earth, while Figure6c,d render an overview of the Moon. It was by standard mechanisms, such as WMS, PlaniSphere may consume these maps and render them. createdFigure by a6a,b single exhibit layer map from data the of our Jet Propulsionearth, while LaboratoryFigure 6c,d render WMS an server. overview Figure of the6a,b Moon. show It thewas Blue Marblecreated Next by Generation a single layer imagery from the from Jet Propulsion NASA WMS. Laboratory It should WMS be server. noted Figure that 6a,b each show figure the rendersBlue a map imageryMarble Next using Generation a different imagery projection. from NASA Although WMS. all It ofshould the figuresbe noted render that each the figure same maprenders imagery, a Figuremap6a rendersimagery Blueusing Marblea different Next projection. Generation Although imagery all of displayedthe figures render using Flat-Modifiedthe same map imagery, Sinusoidal ProjectionFigure and 6a renders Figure6 Blueb renders Marble Blue Next Marble Generation Next imagery Generation displayed imagery using displayed Flat-Modified using Sinusoidal the Mercator Projection.Projection Figure and Figure6c shows 6b renders a moon. Blue InMarble Figure Next6d, Generation several detail imagery points displayed of interest using the of Mercator the Moon are exhibited.Projection. ThisFigure image 6c shows is composed a moon. In of Figure a Clementine 6d, several Elevation detail points map of andinterest a superimposed of the Moon are Lunar exhibited. This image is composed of a Clementine Elevation map and a superimposed Lunar Nomenclature layer. Both of the map layers are made available from the Jet Propulsion Laboratory Nomenclature layer. Both of the map layers are made available from the Jet Propulsion Laboratory WMS and use the same WMS server and layers for rendering imagery. WMS and use the same WMS server and layers for rendering imagery.

(a) (b)

Figure 6. Cont. Big Data Cogn. Comput. 2018, 2, 9 10 of 14

Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16 Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16

(c) (d) (c) (d) Figure 6. Blue Marble Next Generation imagery and the Moon from the Jet Propulsion Laboratory FigureFigure 6. Blue 6. Blue Marble Marble Next Next Generation Generation imagery imagery and the the Moon Moon from from the the Jet JetPropulsion Propulsion Laboratory Laboratory WMS server: (a) Flat-Modified Sinusoidal Projection, (b) Mercator Projection, (c) Lunar Orbital WMSWMS server: server: (a) Flat-Modiﬁed (a) Flat-Modified Sinusoidal Sinusoidal Projection, Projection, (b) Mercator(b) Mercator Projection, Projection, (c) Lunar(c) Lunar Orbital Orbital Mosaic, Mosaic, Colorized and Shaded Elevation by a single layer, and (d) A detail view of several points of Mosaic, Colorized and Shaded Elevation by a single layer, and (d) A detail view of several points of Colorizedinterest. and Shaded Elevation by a single layer, and (d) A detail view of several points of interest. interest. Figure 7 shows the capability of aggregating 2D maps into a 3D geospatial data set. The FigureFigure7 shows 7 shows the capability the capability of aggregating of aggregating 2D maps 2D into maps a 3D into geospatial a 3D geospatial data set. Thedata background set. The is background is created by using Bing Aerial photographs with a resolution of 30 m and is projected createdbackground by using Bingis created Aerial by photographsusing Bing Aerial with photog a resolutionraphs with of 30 a mresolution and is projected of 30 m and on ais 3Dprojected space with on a 3D space with the aid of the Shuttle Radar Topography Mission (SRTM) elevation data [54]. The the aidon ofa 3D the space Shuttle with Radar the aid Topography of the Shuttle Mission Radar Topography (SRTM) elevation Mission (SRTM) data [54 elevation]. The foreground data [54]. The shows foreground shows LiDAR [55] cloud points of a neighborhood in North Vancouver. Figure 8 shows foreground shows LiDAR [55] cloud points of a neighborhood in North Vancouver. Figure 8 shows LiDARa 3D [55 visualization] cloud points with of some a neighborhood hills and farm land in North superimposed Vancouver. over Figure a Bing8 aerial shows 2D alandscape 3D visualization map a 3D visualization with some hills and farm land superimposed over a Bing aerial 2D landscape map with someand data hills at and an farmarea landnear superimposedCochrane, Alberta over (51.1830, a Bing aerial−114.4742). 2D landscape A LandSat map photograph and data atwith an area and data at an area near Cochrane, Alberta (51.1830, −114.4742). A LandSat photograph with near Cochrane,landscape details Alberta and (51.1830, an elevation−114.4742). map of the A LandSatsame area photograph are provided with by landscapetwo independent details and and an landscape details and an elevation map of the same area are provided by two independent and elevationdistinct map sources. of the The same end area result are is providedthe aggregation by two and independent the fusion of the and photograph distinct sources. and the Thelandscape end result distinct sources. The end result is the aggregation and the fusion of the photograph and the landscape is themap. aggregation and the fusion of the photograph and the landscape map. map.

Figure 7. A LiDAR cloud point map superimposed over a Bing aerial view of North Vancouver. Figure 7. A LiDAR cloud point map superimposed over a Bing aerial view of North Vancouver. Figure 7. A LiDAR cloud point map superimposed over a Bing aerial view of North Vancouver.

Geospatial information plays an important role in environmental modelling, resource management, business operations, government policy, and in enhancing the quality of life. Currently, there is a greater quantity of geospatial information available from remote sensing, monitoring networks, survey, and censuses than in any time in history. However, difficulties arise from the fact that there may be little or no commonality between the formats of these geospatial databases at the spatial and temporal scales. Image information and quality with respect to land cover patches of varying sizes and shapes are determined by its spatial resolution and the image processing used. Most multiscale approaches use labels from only one level of the categorical scale, i.e., they are categorically monoscale. A patch can be labeled as general class only when, at a certain spatial scale, it is judged to Big Data Cogn. Comput. 2018, 2, 9 11 of 14 contain sufficient information likely to correspond to patches of more than one specific class. The term “categorical scale” (also called categorical resolution by other authors) refers to the level of detail in the categories used in classification. Big Data Cogn. Comput. 2018, 2, x FOR PEER REVIEW 2 of 16

Figure 8. A three-dimensional (3D) visualization with some hills and farm land superimposed over a Figure 8.BingA aerial three-dimensional two-dimensional (3D) (2D) visualizationmap and data at with an area some near hillsCochrane, and farmAlberta land (51.1830, superimposed −114.4742). over a Bing aerial two-dimensional (2D) map and data at an area near Cochrane, Alberta (51.1830, −114.4742). Geospatial information plays an important role in environmental modelling, resource management, business operations, government policy, and in enhancing the quality of life. Currently, In order to utilize different data for any new and innovative applications, many disparate data there is a greater quantity of geospatial information available from remote sensing, monitoring sourcesnetworks, must be aggregatedsurvey, and censuses to make than use ofin any the largetime in body history. of geospatialHowever, difficulties information arise available. from the fact However, some visualizationthat there may tools,be little such or no as commonality Google map, between have the no formats functions of these for geospatial users to databases perform at the their own modellingspatial and and data temporal aggregation. scales. Image Unlike information Google and map, quality PlaniSphere with respect provides to land cover plug-in patches functionality, of which allowsvarying forsizes users and shapes to analyze are determined and fuse by spatial its spatial data resolution through and modelling the image algorithms. processing used. The plugin functionalityMost multiscale of the system approaches shows use that labels this from platform only canone level be adapted of the categorical for different scale, applications, i.e., they are such as categorically monoscale. A patch can be labeled as general class only when, at a certain spatial scale, environmental spatial analysis or distance education. This topic is of interests in the scope and priority it is judged to contain sufficient information likely to correspond to patches of more than one specific of Informationclass. The Visualization. term “categorical For scale” example, (also ascalled large-scale categorical measurements resolution by other are time-consuming authors) refers to the and costly, many projectslevel of detail are restricted in the categories to single used point in classification. or local data available for the RS model calibration. As a result, land classificationIn order to utilize from different remote data sensing for any images new and can innovative be difficult applications, because many of lack disparate of data data for model calibration.sources In themust plugin be aggregated window, to a make remote use sensing of the large (RS) body image of ofgeospatial LULC caninformation be classified available. through its However, some visualization tools, such as Google map, have no functions for users to perform their colour bands [53]. The image of RS raw data was stored as the coordinate of each point post-processed own modelling and data aggregation. Unlike Google map, PlaniSphere provides plug-in to producefunctionality, several per-pointwhich allows attributes, for users to including analyze and the fuse elevation, spatial data RGB through color, modelling and classification algorithms. codes of the point.The The plugin latter functionality are used toof labelthe system points shows as belonging that this platform to buildings, can be bare adapted earth, for crops, different water, etc. Furtherapplications, post-processing such as allowed environmental for comparing spatial analysis with itsor categorydistance education. of national This census topic is data of interests in this region, typicallyin involvingthe scope and the priority integration of Information of other Visualization. data format, For such example, as census as large-scale data. The measurements national census are data was usedtime-consuming for validation and and costly, calibration many projects of classification are restricted to of single LULC point of remoteor local data sensing. available Classification for the of RS model calibration. As a result, land classification from remote sensing images can be difficult LULC of remote sensing image can be performed in regional scale since many censuses are available because of lack of data for model calibration. In the plugin window, a remote sensing (RS) image of at the nationalLULC can scale. be classified This provided through anits approachcolour bands to improve[53]. The image the accuracy of RS raw of data LULC was classifications stored as the using conventionalcoordinate approaches of each point of limited post-processed field labeling to prod anduce sampling several per-point at small attributes, scale. Therefore, including PaniSpherethe provideselevation, not only RGB a visualization color, and classification platform, codes but also of the an point. analyzed The latter window are used for variousto label users.points as belonging to buildings, bare earth, crops, water, etc. Further post-processing allowed for comparing 4. Conclusionswith its category of national census data in this region, typically involving the integration of other data format, such as census data. The national census data was used for validation and calibration of A platformclassification framework of LULC of named remote PlaniSpheresensing. Classi hasfication been of developedLULC of remo forte the sensing aggregation image can and be fusion of variousperformed geospatial in regional data scale and since raw data.many censuses The framework are available uses at the NASA national World scale. Wind This provided to access an remote map serversapproach using to improve WMS. WMSthe accuracy has been of LULC used classifi to aidcations in transferring using conventional geospatial approaches data over of limited the Internet as the underlying protocol that supports service-oriented architecture (SOA). We have illustrated the potential applications for the PlaniSphere platform in aggregation of 2D spatial data and maps Big Data Cogn. Comput. 2018, 2, 9 12 of 14

into 3D visualization using six examples of open street maps, earth, moon, and land landscape with hills and farmland. The results show that PlaniSphere can aggregate and parses ﬁles that reside in local storage and conforms to the following formats: GeoTIFF, ESRI shape ﬁles, KML, GeoJSON, and LiDAR(LASer). Internet spatial data can create geospatial data sets (map data) from multiple sources regardless of who the data providers are. These analyzing techniques and the increasing availability of geo-referenced data, could provide an effective way to manage spatial information by providing large-scale storage, multidimensional data management and OLAP querying capabilities together in one system. Because of the rich data resource with various data formats, this can offer a new approach for remote sensing map aggregation. The plug-in function of this framework illustrates that the platform can be expanded for greater uses, such as special data formats in environmental spatial analysis and distance education by providing customizations to serve future uses, which existing GIS software is not available. Aggregation of different data sources will generate a new map with a higher level of informational detail that is not currently provided by any individual vendors.

Acknowledgments: Mihal Miu would like to thank James R. Miller, Department of Electrical Engineering and Computer Science, The University of Kansas for providing access to Lidar Data Visual Analysis and Editing. http: //people.eecs.ku.edu/~miller/; M.M. would like to thank the City of Vancouver for LiDAR data collected in 2013 with coverage up to extents of City of Vancouver’s legal jurisdiction. http://data.vancouver.ca/datacatalogue/ LiDAR2013.htm. Author Contributions: Mihal Miu developed idea and software, implemented the program and paper writing; Xiaokun Zhang involved in drafting or critically revisiting the manuscript; M. Ali Akeber Dewan involved in data acquisition, image processing and critically revisiting; Junye Wang developed concept, interpreted the results and structured the manuscript. All authors have read and approved the final manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Chi, M.; Plaza, A.; Benediktsson, J.A.; Sun, Z.; Shen, J.; Zhu, Y. Big Data for Remote Sensing: Challenges and Opportunities. Proc. IEEE 2016, 104, 2207–2219. [CrossRef] 2. Ma, Y.; Wang, L.; Liu, P.; Ranjan, R. Towards building a data-intensive index for big data computing—A case study of Remote Sensing data processing. Inf. Sci. 2015, 319, 171–188. [CrossRef] 3. Guo, H.; Wang, L.; Chen, F.; Liang, D. Scientiﬁc big data and Digital Earth. Chin. Sci. Bull. 2014, 59, 5066–5073. [CrossRef] 4. Yu, L.; Liang, L.; Wang, J.; Zhao, Y.; Cheng, Q.; Hu, L.; Liu, S.; Yu, L.; Wang, X.; Zhu, P.; et al. Meta-discoveries from a synthesis of satellite-based land-cover mapping research. Int. J. Remote Sens. 2014, 35, 4573–4588. [CrossRef] 5. United Nations; European Commission; Food and Agriculture Organization of the United Nations; International Monetary Fund; Organisation for Economic Cooperation and Development; World Bank. System of Environmental-Economic Accounting 2012: Central Framework, Prepublication (White Cover). Available online: http://unstats.un.org/unsd/envaccounting/White_cover.pdf (accessed on 28 January 2018). 6. Wang, J.; Cardenas, L.M.; Misselbrook, T.H.; Gilhespy, S. Development and application of a detailed inventory framework of nitrous oxide and methane emissions from agriculture. Atmos. Environ. 2011, 45, 1454–1463. [CrossRef] 7. Shrestha, N.K.; Du, X.; Wang, J. Assessing climate change impacts on fresh water resources of the Athabasca River Basin. Sci. Total Environ. 2017, 601–602, 425–440. [CrossRef][PubMed] 8. Lokers, R.; Knapen, R.; Janssen, S.; van Randen, Y.; Jansen, J. Analysis of Big Data technologies for use in agro-environmental science. Environ. Model. Softw. 2016, 84, 494–504. [CrossRef] 9. Vitolo, C.; Elkhatib, Y.; Reusser, D.; Macleod, C.J.A.; Buytaert, W. Web technologies for environmental Big Data. Environ. Model. Softw. 2015, 63, 185–198. [CrossRef] 10. Wolfert, S.; Ge, L.; Verdouwa, C.; Bogaardt, M.-J. Big Data in Smart Farming—A review. Agric. Syst. 2017, 153, 69–80. [CrossRef] 11. Daniel, B.K. Reimaging Research Methodology as Data Science. Big Data Cogn. Comput. 2018, 2, 4. [CrossRef] 12. Hogland, J.; Anderson, N. Function Modeling Improves the Efﬁciency of Spatial Modeling Using Big Data from Remote Sensing. Big Data Cogn. Comput. 2017, 1, 3. [CrossRef] Big Data Cogn. Comput. 2018, 2, 9 13 of 14

13. Cavallaroa, G.; Riedel, M.; Bodenstein, C.; Glock, P.; Richerzhagen, M.; Goetz, M.; Benediktsson, J.A. Scalable developments for big data analytics in remote sensing. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, Milan, Italy, 26–31 July 2015. 14. Kussul, N.; Shelestov, A.; Lavreniuk, M.; Butko, I.; Skakun, S. Deep learning approach for large scale land cover mapping based on remote sensing data fusion. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, Beijing, China, 26–31 July 2016. 15. Yang, C.; Yu, M.; Hu, F.; Jiang, Y.; Li, Y. Utilizing cloud computing to address big geospatial data challenges. Comput. Environ. Urban Syst. 2017, 61, 120–128. [CrossRef] 16. Fritz, S.; See, L.; Perger, C.; McCallum, I.; Schill, C.; Schepaschenko, D.; Duerauer, M.; Karner, M.; Dresel, C.; Laso-Bayas, J.-C.; et al. A global dataset of crowdsourced land cover and land use reference data. Sci. Data 2017, 4, 170075. [CrossRef][PubMed] 17. Li, J. Assessing spatial predictive models in the environmental sciences: Accuracy measures, data variation and variance explained. Environ. Model Softw. 2016, 80, 1–8. [CrossRef] 18. Granell, C.; Havlik, D.; Schade, S.; Sabeur, Z.; Delaney, C.; Pielorz, J.; Usländer, T.; Mazzetti, P.; Schleidt, K.; Kobernus, M.; et al. Future Internet technologies for environmental applications. Environ. Model Softw. 2015, 78, 1–15. [CrossRef] 19. Kuhn, M.; Johnson, K. Applied Predictive Modeling; Springer: New York, NY, USA, 2013. 20. Shrestha, N.K.; Wang, J. Predicting sediment yield and transport dynamics of a cold climate region watershed in changing climate. Sci. Total Environ. 2018, 625, 1030–1045. [CrossRef] 21. Maas-Hebner, K.G.; Harte, M.J.; Molina, N.; Hughes, R.M.; Schreck, C.; Yeakley, J.A. Combining and aggregating environmental data for status and trend assessments: Challenges and approaches. Environ. Monit. Assess. 2015, 187, 278. [CrossRef][PubMed] 22. Elliott, P.; Wartenberg, D. Spatial Epidemiology: Current Approaches and Future Challenges. Environ. Health Perspect. 2004, 112, 998–1006. [CrossRef][PubMed] 23. Cromley, E.K.; Cromley, R.G. An analysis of alternative classification schemes for medical atlas mapping. Eur. J. Cancer 1996, 32, 1551–1559. [CrossRef] 24. Soulard, C.E.; Acevedo, W.; Cohen, W.B.; Yang, Z.; Stehman, S.V.; Taylor, J.L. Harmonization of forest disturbance data sets of the conterminous USA from 1986 to 2011. Environ. Monit. Assess. 2017, 189, 170. [CrossRef][PubMed] 25. Villarreal, M.L.; van Riper, C., III; Petrakis, R.E. Conflation and aggregation of spatial data improve predictive models for species with limited habitats: A case of the threatened yellow-billed cuckoo in Arizona, USA. Appl. Geogr. 2014, 47, 57–69. [CrossRef] 26. Brewer, C.A.; Pickle, L. Evaluation of Methods for Classifying Epidemiological Data on Choropleth Maps in Series. Ann. Assoc. Am. Geogr. 2002, 92, 662–681. [CrossRef] 27. Taylor, J.R.; Lovell, S.T. Mapping public and private spaces of urban agriculture in Chicago through the analysis of high-resolution aerial images in Google Earth. Landsc. Urban Plan. 2012, 108, 57–70. [CrossRef] 28. Kamadjeu, R. Tracking the polio virus down the Congo River: A case study on the use of Google Earth in public health planning and mapping. Int. J. Health Geogr. 2009, 8, 4. [CrossRef][PubMed] 29. Lozano-Fuentes, S.; Elizondo-Quiroga, D.; Farfan-Ale, J.A.; Fernandez-Salas, I.; Beaty, B.J.; Eisen, L. Use of Google Earth to Facilitate GIS Based Decision Support Systems for Arthropod-Borne Diseases. Adv. Dis. Surveill. 2007, 4, 91. 30. Xiong, J.; Thenkabail, P.S.; Gumma, M.K.; Teluguntla, P.; Poehnelt, J.; Congalton, R.G.; Yadav, K.; Thau, D. Automated cropland mapping of continental Africa using Google Earth Engine cloud computing. ISPRS J. Photogramm. Remote Sens. 2017, 126, 225–244. [CrossRef] 31. Gorelick, N.; Hancher, M.; Dixon, M.; Ilyushchenko, S.; Thau, D.; Moore, R. Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sens. Environ. 2017, 202, 18–27. [CrossRef] 32. ESRI Team. ArcGIS Software. Available online: https://www.arcgis.com/features/ (accessed on 28 June 2016). 33. Bhatt, G.; Kumar, M.; Duffy, C.J. A tightly coupled GIS and distributed hydrologic modeling framework. Environ. Model. Softw. 2014, 62, 70–84. [CrossRef] 34. Miller, M.; Odobasic, D.; Medak, D. An Efficient Web-GIS Solution based on Open Source Technologies: A Case-Study of Urban Planning and Management of the City of Zagreb, Croatia. In Proceedings of the FIG Congress Facing the Challenges—Building the Capacity, Sydney, Australia, 11–16 April 2010. Big Data Cogn. Comput. 2018, 2, 9 14 of 14

35. Viswanathan, G.; Schneider, M. User-Centric Spatial Data Warehousing: A Survey of Requirements & Approaches. Int. J. Data Min. Model. Manag. 2014, 6, 369–390. 36. Horsburgh, J.S.; Tarboton, D.G.; Piasecki, M.; Maidment, D.R.; Zaslavsky, I.; Valentine, D.; Whitenack, T. An integrated system for publishing environmental observations data. Environ. Modell. Softw. 2009, 28, 879–888. [CrossRef] 37. Shu, Y.; Liu, Q.; Taylor, K. Semantic validation of environmental observations data. Environ. Modell. Softw. 2016, 79, 10–21. [CrossRef] 38. Hallgren, W.; Beaumont, L.; Bowness, A.; Chambers, L.; Graham, E.; Holewa, H.; Laffan, S.; Mackey, B.; Nix, H.; Price, J.; et al. The Biodiversity and Climate Change Virtual Laboratory: Where ecology meets big data. Environ. Modell. Softw. 2016, 76, 182–186. [CrossRef] 39. PlaniSphere. Available online: www.planisphere.biz (accessed on 28 January 2018). 40. GeoServer. An Open Source Server for Sharing Geospatial Data. Available online: http://geoserver.org/ (accessed on 28 January 2018). 41. MapServer. Open Source Platform for Publishing Spatial Data and Interactive Mapping Applications to the Web. Available online: http://mapserver.org/ (accessed on 28 January 2018). 42. Web Map Service. Open Geospatial Consortium. Available online: http://www.opengeospatial.org/ standards/wms (accessed on 28 January 2018). 43. OGC. OpenGIS® White Paper. Available online: http://www.opengeospatial.org/docs/whitepapers (accessed on 28 January 2018). 44. Gomez, L.; Haesevoets, S.; Kuijpers, B.; Vaisman, A.A. Spatial aggregation: Data model and implementation. Inf. Syst. 2009, 34, 551–576. [CrossRef] 45. Kimball, R.; Ross, M. The Data Warehouse Toolkit: The Complete Guide to Dimensional Modeling, 2nd ed.; Wiley & Sons: New York, NY, USA, 2002. 46. NASA. World Wind. Available online: http://worldwind.arc.nasa.gov/features.html (accessed on 28 January 2018). 47. Pourabdollah, A. OSM-GB: Using Open Source Geospatial Tools to Create OSM Web Services for Great Britain, FOSS4G, Nottingham. Available online: http://www.academia.edu/4591884/OSM-GB_Using_ Open_Source_Geospatial_Tools_to_Create_OSM_Web_Services_for_Great_Britain (accessed on 28 January 2018). 48. Herbreteau, V.; Salem, G.; Souris, M.; Hugot, J.P.; Gonzalez, J.P. Thirty years of use and improvement of remote sensing, applied to epidemiology: From early promises to lasting frustration. Health Place 2007, 13, 400–403. [CrossRef][PubMed] 49. Boulil, K.; Bimonte, S.; Pinet, F. Conceptual model for spatial data cubes: A UML proﬁle and its automatic implementation. Comput. Stand. Interfaces 2015, 38, 113–132. [CrossRef] 50. Sarwat, M. Interactive and Scalable Exploration of Big Spatial Data—A Data Management Perspective, Mobile Data Management (MDM). In Proceedings of the 2015 16th IEEE International Conference on Mobile Data Management (MDM), Pittsburgh, PA, USA, 15–18 June 2015; pp. 263–270. 51. Miu, M. Aggregation of Map (Geospatial) Data. Master’s Thesis, Athabasca University, Athabasca, AB, Canada, 2016. 52. Wikipedia. Available online: https://www.mediawiki.org/wiki/Extension:GeoData (accessed on 28 January 2018). 53. Miu, M.; Zhang, X.; Dewan, M.A.A.; Wang, J. Aggregation and visualization of spatial data with application to classiﬁcation of land use and land cover. Geoinform. Geostat. 2017, 5, 1–9. [CrossRef] 54. Shuttle Radar Topography Mission (SRTM). The Mission to Map the World, Jet Propulsion Laboratory, California Institute of Technology. Available online: http://www2.jpl.nasa.gov/srtm/ (accessed on 28 January 2018). 55. Lidar. City of Vancouver. Available online: http://data.vancouver.ca/datacatalogue/LiDAR2013.htm (accessed on 28 January 2018).

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).