A NEW WEB-BASED SOFTWARE TOOL FOR ICESAT AND ICESAT-2 LASER ALTIMETRY DATA PROCESSING AND VISUALIZATION

Bruno Silva1, Luiz Guerreiro Lopes2, Pedro Campos2,3

1 Graduate Program in Informatics , University of , , 2 Faculty of Exact Sciences and Engineering, University of Madeira, Penteada Campus, 9020-105 Funchal, Madeira Is., Portugal 3 Interactive Technologies Institute (ITI/LARSyS), 9020-105 Funchal, Portugal

ABSTRACT using these data. After all, when dealing with satellite data, the data itself is only half of the equation, and having the abil- The purpose of this paper is to present and briefly describe ity to quickly and precisely identify the location and easily a web-based tool for processing and visualizing in a suitable characterize its surroundings according to the research objec- way the available data from satellite laser altimetry missions, tives is crucial for the interpretation of the results obtained in particular data from ICESat and ICESat-2. This new tool, from such data. called ICEComb, allows to access all data contained within each data granule of the datasets of both missions, visual- ize the data interactively on a geographic map, store data lo- 2. THE ICECOMB TOOL cally, and process and explore data in an efficient, detailed and meaningful way, thus providing an easy-to-use environment 2.1. Tool architecture, design and implementation for satellite laser altimetry data analysis and interpretation. In order to create a richer visual environment while accessing Index Terms— Laser altimetry, ICESat/GLAS, ICESat-2/ data from both the ICESat and ICESat-2 missions, a web in- ATLAS, software tool design, data visualization terface based on the graphical component of the Google Maps API was chosen. This mapping API was chosen since its a scalable and modern web platform that provides user access 1. INTRODUCTION to updated satellite imagery, high-resolution aerial photogra- phy and high-detail low altitude images, while offering high Handling the large volume of data from satellite laser altime- performance. try missions is challenging. A reference project for visualiz- In addition, the Google Maps API offers the possibility ing ICESat and ICESat-2 data is the OpenAltimetry platform to use different map view formats, such as Google Earth [1], which was created as a tool to provide altimetry-specific satellite images and a physical map based on terrain informa- data from ICESat and ICESat-2 missions with a web-based tion, and to access a set of other tools in the form of Client interactive interface. However, by only focusing on altimetric Libraries with APIs such as Distance Matrix and Elevation, laser data, that tool leaves out many other useful and impor- which allow, respectively, to calculate the distance between tant information present in the data products of both missions. geographic coordinates and provide elevation data for the In view of that, the aim of this paper is to present a new Earth’s surface and depth for the ocean floor. tool, called ICEComb, to process and visualize data from both The new web-based software tool has an architecture the ICESat and ICESat-2 missions in a meaningfully way on composed of two main components, a Data Server and a Web top of a highly detailed web mapping service, with basis on Engine that interfaces with the user through a Web Browser. the geographic location from where data was collected. This Client/Server approach, represented in a simplified way The ICEComb tool offers the possibility to browse the raw in Figure 1, permits sharing the processing demands and data contained in the datasets, represented using visual aids allows for better overall performance. like graphs, tables and data conversions, in a rich geographic The developed tool was built using standard languages related environment in order to facilitate data interpretation and well-known and well-documented web technologies in and discoverability. This approach simplifies knowledge cor- order to facilitate the incorporation of new functionalities and relation and the conception and validation of scientific models features to it and to extend its applicability to data from other satellite laser altimetry missions. This allows data mash-up This work was partially supported by the Portuguese Foundation for Sci- ence and Technology (FCT) through LARSyS – FCT Pluriannual funding and opens the possibility of expanding the context of studies 2020–2023. developed from the integration of these data. Figure 2, from an ongoing study using laser measure- ments on the world’s largest coastal lagoon, illustrates the client user interface, which is composed of two main areas, the map on the left and the input controls on the right.

Fig. 1. ICEComb tool architecture.

2.2. ICEComb data server Fig. 2. ICEComb data selection.

Data Server actions are limited to data file handling operations The map area is the main element of the user interface such as filtering, querying and gathering information, and all as it is where all data are presented to the user. Firstly, the other activities related to data presentation for the end user, real ground tracks are drawn, allowing the view of the pro- handling of data requests and responses, and other interface jection of the satellite’s data acquisition path on the Earth’s tasks are then offloaded to the client side. surface, and then the data points are marked along the tracks The Data Server was designed with the RESTful archi- using the coordinates of the locations from which the data tecture pattern in mind, so requests are made through the use was collected. Since ICESat/GLAS data was gathered at dif- of JavaScript URI libraries, payloads are formatted in JSON ferent rates, this causes different data point types, intervals (JavaScript Object Notation), and the server does not main- and available data by data point. tain the client’s state information. The area on the right of the user interface is separated in ICESat and ICESat-2 data files are stored using the HDF5 two tabs. The first one, visible on the right side of Figure 2, format, which was designed to manage very large numerical represents the options of the general data filtering system. data collections. However, applications using HDF (Hierar- Users start by selecting the data product to display (i.e., the chical Data Format) and HDF5 data manipulation APIs tend satellite mission and the dataset) and also may set a date in- to be hard to implement due to their low level of abstraction. terval filter. There is also a system for the management of data The API chosen for data access was the HDFql (Hierar- granules that allows users to set the allocation of the number chical Data Format query language), which is made available of data granules are processed each time, either all data files free of charge and royalty-free, and is the first data interface or by small chunks at a time. to allow HDF5 data access through a high-level language with Data quality can be set in the “Dataset Measured Parame- a syntax similar to SQL (Structured Query Language), that is ters” area, that uses QA (quality assessment) information spe- the standard for query languages and Web applications. cific to each dataset, obtained from the granule metadata files. These files contain the evaluation of the instruments and geo- 2.3. ICEComb client physical parameters expressed in the data, as well as specific test assessment flags. By having the user interface developed in JavaScript, the de- The second tab, visible on the right in Figure 3 (from the veloped solution is able to offload some computation to the same study in the Patos Lagoon), contains the general options client. This approach relegates all data access and processing for the ICEComb Client that are common to all datasets. In operations to the data providers, leaving the clients responsi- this area, the users can control the behavior of the user inter- ble to handle all user interface aspects, such as data requests face, the elements marked on the map, as well as to view and and aggregation, map display and data drawing. define various parameters related to data and the map. size, version, etc.), the Granule Measured Parameters (QA information from the metadata file) and the Granule Data Quality Analysis data (built-in imagery with data summary, history, QA analysis, and other parameters). After that, the gathered data is presented and organized by data rate. Figure 4 represents part of the data view from a granule of the GLAH10 (GLAS/ ICESat L2 Global Aerosol Vertical Structure Data) dataset and illustrates some mechanisms used to represent data in a most meaningful way. Data is converted to graphs, tables or simply shown as data lines, very long val- ues are converted to scientific notation or metric prefixes, and flag-type values are translated to their meaning, all this to fa- cilitate the data interpretation.

Fig. 3. ICEComb real ground tracks and coordinates points.

One part of the mechanism created to limit the amount of data sent by the ICEComb server, therefore the amount of data processed on the client, is called Map Bounds (or map bounding area). This works by defining two geographical lo- cations, the southwest and the northeast location points (with latitudinal and longitudinal values) that are used to instruct the ICEComb Server to filter any data point outside those two locations. The default geographic bound mechanism (“Map Bounds”) is based on the limits of the displayed map area, but users can set these bounds manually by changing the bound mechanism to “Shape Bounds” and draw a shape on the map Fig. 4. ICEComb data view. that defines the map area that data will be processed. The representation of ground tracks in ICEComb is de- The last tab on the ICEComb Client user interface holds fined by the coordinates of the granule data, rather than by the the Data Processor tab, illustrated in Figure 5. It allows the projection of the satellite’s reference orbit on the Earth’s sur- study of geographic areas using the altimetry data from the face. This means that the represented ground tracks are drawn ICESat and ICESat-2 missions and the build-in Elevation using the data location points and represent the real satellite Data Processor. track. Also, each track is represented by a different color, one color by each data granule. This approach was chosen so that the tool would have a data-centric approach on data represen- tation, meaning that users can rapidly identify which granule file produced the data presented on the map. The ground track can also indicate the satellite direction and be used to characterize the ground elevation underneath the track, as shown on the plot bellow the map in Figure 3. Data for this plot is provided by the Google Elevation Ser- vice that delivers elevation data for the surface of the Earth, including depth locations on the ocean floor. Coordinates Points are the geographic locations from which data was collected. These are marked on the map, us- ing specific shapes, on the latitude and longitude coordinates where data was collected, and offer access to the “Coordi- nates Point Information” window where data is displayed, as shown in Figure 3. This window shows some general infor- mation about the granule, such as Granule File Data (name, Fig. 5. ICEComb Data Processor. The selection of the Coordinates Points can be done in level rise, the rate of mass balance changes and the amount of two ways, by selecting points one by one (Single Point se- global biomass with high accuracy and at a global scale. lection) or by indicating a Coordinates Points range (Interval Moreover, the fact that data is directly coupled to the gath- selection). In this last mode, user just need to indicate two ered location information, by having a visual reference of the points of the desired track and the tool automatically selects satellite track and the possibility to evaluate the characteris- the points in between. Finally, the data for the selected points tics of that location, is a fundamental feature to understand is extracted and the Elevation with reference to the World and produce new scientific models. To create a solution that Geodetic System 1984 (WGS84) and the Earth Gravitational offers both ease of use and performance requires the develop- Model 2008 (EGM2008) datum is calculated. ment of tools that are based on well-known and user-friendly The extracted data is subjected to a set of statistical anal- platforms as well as the creation of processes based on pro- ysis and to an outlier detection algorithm. There are four dif- gramming models that allows some level of parallelism. ferent algorithms implemented in the ICEComb tool that can The proposed software tool is intended to be mainly used be used with both the ICEsat and the ICESat-2 data, namely by researchers and scientists to aid their work using ICESat the Outlier Interquartile Range (IQR), the 2-Sigma rejection and ICESat-2 data. It offers a ready-to-use system to rapidly criteria, the outlier residual and standard deviation fence, and access the raw collected data in a visually engaging way, with- the RANdom SAmple Consensus algorithm (RANSAC). out the need to have prior knowledge of the format, parame- ters or structure of the data products. Also, the ICEComb ar- chitecture was developed using standard languages, allowing 3. DISCUSSION AND CONCLUSIONS expandability and the integration of new scientific models. The need to create a new tool for the processing and visualiza- The built-in Elevation Data Processor is a valuable scien- tion of ICESat and ICESat-2 data products derives primarily tific tool as it allows the study of geographic areas using laser from the fact that the existing solutions only offer access to a altimetry data without recourse to other tools. limited amount of information contained in that datasets and Furthermore, the fact that the ICEComb solution has only because the feature set offered by the mapping tool used to two main elements—a server program and a web client— represent the world map play an important role in the inter- makes it can almost be seen as a self contained application pretation of data, since the characterization of the geographic that allows it to be easily deployed in a single computer for a location of the data collection points is often vital. quick access to data, without the need the allocation of high Tools like NASA’s Earthdata Search, Worldview and the performance server hardware, which is not always available, OpenAltimetry platform use a map as an interface for ICESat although the data storage requirements for storing the data and ICESat-2 data. The first tool is a web-based client for products of both missions is quite large. managing and distribute data from multiple Earth-observing In terms of efficiency, performance analysis shown that missions, having a focus on advanced data filtering and sub- the ICEComb tool can process 9 228 files of ICESat/GLAH01 setting, since its main objective is to allow the download of dataset (approximately 136 GB total data) in about 37 sec- data granules using a geographical map as a reference. In onds, when running the client and server in the same system. turn, Worldview is a tool that offers end users access to view Handling the same batch of files while running the client and and download remote sensing imagery [2]. Both tools only server in separated systems took about 26 seconds, an im- provide access to the spatial coverage of the two ICESat mis- provement of about 30%. The test systems were equipped sions but not to the data values. with 4 core 8 thread CPUs, 16 GB of RAM, SSD storage, and the measured execution time refers to the second test run. In contrast, the OpenAltimetry platform allows the spe- cific visualization of laser altimetry data from both missions by its geographical location [1] on a map-based interface us- 4. REFERENCES ing the OpenLayers mapping library. [1] S. J. S. Khalsa, A. Borsa, V. Nandigam, M. Phan, However, both the ICESat and ICESat-2 datasets contain K. Lin, C. Crosby, H. Fricker, C. Baru, and L. Lopez, more than laser elevation data. Data on the Earth’s atmo- “OpenAltimetry – rapid analysis and visualization of sphere, such as cloud layers, aerosol vertical structure and spaceborne altimeter data,” Earth Sci. Inform., 2020, planetary boundary layer are some examples of data that are doi:10.1007/s12145-020-00520-2. not possible to be viewed on the OpenAltimetry platform but can be visualized using the ICEComb tool. [2] K. J. Murphy, D. K. Davies, K. Michael, C. O. Justice, Both the ICESat and ICESat-2 missions have provided J. E. Schmaltz, R. Boller, B. D. McLemore, F. Ding, many important insights into the value of satellite laser al- B. Vollmer, and M. M. Wong, “LANCE, NASA’s Land, timetry and have demonstrated this to be a very useful tool for Atmosphere Near real-time Capability for EOS,” in Time- examining and understanding changes on the Earth’s surface. Sensitive Remote Sensing, C. D. Lippitt, D. A. Stow, and Data from these missions enabled scientists to quantify, e.g., L. L. Coulter, Eds., pp. 113–127. Springer, New York, the seasonal and annual contributions of the ice sheets to sea 2015.