Eective high-quality science graphics from s-Ink.org

A standard internet image search on a scienti c topic is common practice, and o ers a plethora of images. However, the suggested images provide neither a guarantee in the accuracy of the science content being portrayed, nor in their clarity or artistic elegance. Being curated by both scientists and artists, the online s-Ink collection (https://s-Ink.org) is, in contrast, an exchange platform for high quality, science-related graphical products. The collection is highly structured and e ectively searchable via text, keyword, or category. As necessary, its hosted and linked content can be regularly updated to maintain pace with advances in both scienti c understanding and graphicrelated techniques. Key related information is provided alongside the graphical products, including an explanatory caption, the given license guiding further use, the names of their creators to value their work, and any related publication for proper citing and further in-depth information. By being open for contributions, the s-Ink collection facilitates access to open-access, community-driven, visually engaging, and science-proof graphics for free, while honouring the work that went into them.


Shortcomings of common science graphics
The image quality in a scienti c (i.e., not artistic) sense can easily be checked. Critical image pitfalls are counted for relevant key Earth Science topics on one of the most common science-images retrieval platforms (see Figure A. ) and the results (as of December ) presented in Figure . These are data misrepresentation (Figure a), bred by missing or faulty scales (e.g., colour bars with perceptually uneven colour gradients) obstructing the factual representation of data gradients, visual inaccessibility (Figure b), provoked by unintuitive colour orders or colour combinations excluding colourvision de cient readers (e.g., green and red combinations), and conceptual misrepresentation (Figure c), brought about by outdated and wrong scienti c content such as the likes of a "molten mantle" or the " st order geodynamic connection between mantle plumes and spreading ridges". The presence of those critical image pitfalls in a commonly accessible science image resource is striking. Universally accessible science graphics free of data misrepresentation and/or misleading concepts are rare. Furthermore, many images are available in low-resolution only, might not be suitable to variable backgrounds, and have limited or unclear copyright usage purposes. Conversely, a situation may arise in which someone has produced an image they are keen to share more widely but don't know what channels are available to increase its visibility.
In summary: open-accessible scienti c-style artwork that looks nice, but also represents scienti c and, in particular, geoscienti c content clearly and accurately is uncommon and hard to nd. It is in the interests of all scientists to share their research in an accessible, informative, aesthetic, and reproducible way (Makri, ); and this includes stand-alone graphical products.

Graphics categories
A multitude of graphics used in science exist and they may be grouped into one (or several) of the following categories.
• Conceptual illustrations are here considered "freehand" drawings to portray certain concepts qualitatively (see e.g., s-ink.org/ocean-plate-tectonics). Instead of directly representing data, they are visualising current knowledge or core concepts. It is therefore crucial that conceptual graphics are as accurate as possible. While sketches must be simple and details streamlined to a certain degree, they cannot be oversimpli ed and thereby misrepresent actual knowledge and lead to common misconceptions (for example, like a red, hot, and, therefore, molten Earth's mantle; the Earth's mantle is in fact a solid that deforms on geological timescales) that are di cult to counteract later on. Sketches are a particular type of conceptual illustration and, in particular, should be regularly revised and, if necessary, updated or removed.
• Data visualisations are data-based maps, charts, or graphs that quantify and relate the underlying data via graphical rulers, like an x-axis or a colour bar (see e.g., s-ink.org/oceanplate-age). For scienti c data visualisations, data needs to be queen: the data must be represented without visual distortion, so that the readers can access (and interpret it) themselves, and the data should, in most cases, be a visually dominant aspect of a gure. If a data eld is represented via colour and a colour bar, then the used colour map has to be perceptually uniform, meaning that the colour gradient between equally-spaced colour bar axis ticks must be equal (or change equally) to the eye all along the colour bar, as must the spacing between axis ticks be equal (or change equally) on a x-axis of a graph (Crameri, Shephard, and Heron, ). • Animations are here considered motion graphics that can either be data-based (quantitative; see s-ink.org/mobile-lidmantle-convection), or not (qualitative). If animations are quantitative, they must ful l the same quality standards as Data visualisations. If they are qualitative, they should be handled like Sketches. • Artistic impressions are here artworks that represent a ctitious view on, and provide a feeling of, scienti c concepts or di erent worlds (see e.g., s-ink.org/dynamic-planet-earth).
The decreased emphasis of scienti c accuracy di erentiates Artistic impressions from Conceptual Illustrations. For example, a drawing of a liquid water-covered Martian landscape would be considered an Artistic impression. • Icons are small, simple yet useful graphics that portray things beyond language borders. Several existing resources are available, one of them being the TheNounProject.com. • Logos are graphical identi ers for private persons, groups, institutions, and initiatives (see e.g., s-ink.org/s-ink-logo). In contrast to other graphics, logos must only ful l certain technical aspects like high resolution and are therefore open to unconditional artistic freedom. • Posters are here display graphics (for example, for print) for scienti c and/or public outreach, or simply science-related decoration (see e.g., s-ink.org/scienti c-colour-map-poster). As posters can contain multiple panels constituting almost any type of content, they should undergo the same quality demands as outlined in the respective paragraphs. The poster itself, can, however, be freely designed, and should only consider e ective visual knowledge transfer (e.g., by considering the betterposter approach; Morrison, ), if applicable. • Graphical templates are the basic structures to a nished product enabling their time e ective and graphically-guided creation. An example could be a presentation display (see e.g., s-ink.org/betterposter-poster-template) or x-y data template. As long as the graphical templates allow to ful l all scienceproof standards mentioned in other paragraphs, they are open to unconditional artistic freedom.
by Undertone.design E ective high-quality science graphics from s-Ink.org / Figure : Abundant awed images in internet searches. Critical image pitfalls (in reddish colours) present in openly accessible images on key geoscience topics. Shown are counts for (a) perceptually distorted or even missing colour scales for the search 'Mars topography', (b) colourvision de ciency unreadable colour maps or unintuitive colour orders orders for the search 'Seismic hazard map', and (c) faulty, misleading concepts for a molten mantle or the direct connection of rising mantle plumes and spreading ridges for the search 'Mantle convection'. Individual counting panels are shown in order of appearance on search-results page with o -topic image results being marked not applicable (light grey). Results are based on an image search in December as shown in Figure A. . by Undertone.design E ective high-quality science graphics from s-Ink.org / • Instructions are here guidelines for graphic designers to produce science-quality graphical products. A ow chart indicating what colour map type to use for a given data set would be such an example (see s-ink.org/colour-map-guideline). • Graphical tools are code, algorithms, software, software libraries, and graphical parts thereof to create a nished product (see e.g., s-ink.org/delight-colour-inversion). The graphical tools in the collection must also (allow to) ful l the quality standards outlined in the other paragraphs depending on their purpose. Graphical tools may require additional and detailed README-like documentation to ensure user ease. • Photographs are here real-world images taken of scientically relevant content. Great existing repositories like EGU imaggeo exist for photos and videos. Photographs should ful l relevant graphical aspects as, for example, be clear in terms of the spatial scales portrayed. • Videos are here real-world motion pictures taken of scienti cally relevant content. As with photos, videos can be uploaded to and shared via repositories like EGU imaggeo. Videos should ful l the same quality standards as the Photographs.

Graphical quality measures
should be regularly updated, with the updates being communicated, or else, if outdated, be deleted. While a particular graphic product might not convey information in the clearest way, high artistic quality tends to motivate the reader to spend more time looking at it (see e.g., Figure a). Whilst beauty is to some degree subjective, and in itself is hard to grasp scienti cally, artists are well aware of the advantages of some of the nature-based mathematical patterns, like the golden ratio based on the Fibonacci sequence that produces eye-pleasing proportions. Similarly, various graphic design principles exist to make graphics more pleasing to look at. The colour theory, for example, o ers a guideline for using multiple colours by de ning primary, secondary, complementary, analogous, triadic, and other colour combinations. In today's ood of science output, unattractive graphics might simply not be looked at and are thereby rendered ine ective.
Artistically sound design is advantageous to attract readers, but graphics should also be readable to the general audience. Graphically, information is conveyed via our visual apparatus (i.e., the eye lens, the optical receptors named cone cells, the optical nerve, and the optical cortex), which is not uniform across all viewers (see e.g., Crameri, Shephard, and Heron, ). There is important variations in the perception of colours amongst human beings, with some colour-vision de ciencies (CVDs) disabling, for example, the di erentiation of red and green at similar luminosity. While total colour blindness is fortunately rare, printing in black and white is still common amongst analogue paper readers. Graphics that are CVD-friendly and readable in grey-scale o er therefore a huge gain in visual accessibility (e.g., Figure c

and Figures A. and A. ).
Images appearing online pro t from a short ALT-text (short for 'alternative text'), which is a concise description of the image in the event it cannot be viewed, and is important for accessibility. Further graphic accessibility comes from a clear visual layout, following design principles such as hierarchy (guiding the reader visually), balance (evenly laying out elements across the graphic), and contrast (di erentiating individual elements). A good credo to follow is that when in doubt, simplicity should be chosen over complexity: Everything one can leave away in a visual (including text) should be taken away until only the essentials remain (e.g., Murchie and Diomede, ). The choice of font and typeface should also be based on both maximum clarity and aesthetics keeping in mind that individual fonts also transmit certain tones; common suggestions of (clear and serious) fonts in scienti c graphic design include Helvetica, Futura, or Arial, but less common, but suitable options, like Open Sans, exist too.
In addition to a universally accessible design, scienti c visualisation must ensure representing concepts and data in a clear and data-fair manner, preventing ambiguity and bias. While this seems obvious, the science community (editors and science communicators included) ran head-front into a literal visualisation crisis during the last couple of years: Data visualisation was predomiby Undertone.design E ective high-quality science graphics from s-Ink.org / nantly created with faulty colour axes leading to data misrepresentation and conceptual bias (see e.g., Figure A. ). Colour axes (a.k.a. colour bars) with uneven colour gradients (a.k.a. colour maps) introduce visual distortion and error to the underlying data of more than % of the displayed data range (Crameri, a). Atypical for the science community, the warning call by a myriad of continuous scienti c publications and visualisation experts (e.g., ) has been widely ignored by non-experts, or even confronted with scienti cally unfounded arguments, until recently.
A high standard for data representation is achieved, for example, by implementing the use of scienti cally-derived colour palettes following the suggestions given in Crameri, Shephard, and Heron ( ) and the references therein. While representation (and not only interpretation) is an absolute necessity in scienti c data visualisation, the use of outdated scienti c concepts in graphical illustrations is crucial to avoid. Representing accurate concepts (as is done in Figure A. ) helps to eradicate common misconceptions such as the ones about a molten mantle or the direct dynamic connection between whole-mantle upwelling and spreading ridges at the surface (e.g., Crameri et al., ; Crameri, Shephard, and Conrad, ). High quality graphics are a crucial tool to narrow the currently widening gap between science and society (Makri, ). Below, we therefore outline powerful capacities of scienti c graphic design represented by various, widely applicable examples.
. Specimen : Ensuring accessibility Firstly, we provide an example based on publicly-available data and re-plotted as an original gure. For this scenario, we use the age of the Earth's oceanic sea oor (ocean-plate age; Müller et al., ). To-date, this dataset has rarely been represented fairly and inclusively. To the contrary, by being commonly displayed with a rainbow-like colour chart, it contains visual artefacts and is un t for colour-vision de ciencies. The prevalent use of rainbow for this dataset, and others, becomes clear through an online image search (similar to Figure A. ) or by using current versions of widelydistributed geoscience software.
Figure a is an openly available data visualisation that provides the ocean-plate age free of unfair and inaccessible colour combinations. The colour coded ocean-plate age data is readable for people with colour vision de ciency ( . Specimen : Broadening applicability Secondly, we provide an example of an original, visually attracting gure. Time is synonymous with geology and there exists a multitude of artistic versions of the geologic time scale. While the table-like representation of geologic times and occurrences is certainly useful and clear to an interested viewer, it tends to appear rather unpleasing to a non-enthusiast's eye. Figure is an original, artistic version of the geologic time scale that may not be the most accessible version (due to tilted text and non-linear time axis) compared to more standard rectangular versions (compare to Figures A. and A. ), but, thanks to its artistic avour, it tends to also draw the non-enthusiast in for closer visual inspection of its content (see Figure a). Importantly, the gure also illustrates how information-based graphics can provide exibility to use (Figure b-d) and accessibility to read (Figure e-g). This original graphic is provided as a so-called bare-bone layer too, providing only the key visual detail (here the time scale) to enable the users to alter and adjust the content to their speci c intention. In some cases, the time axis might be more insightful when numbered as Million years ago from the present (Figure b). In other cases, labelling time as Million years since the start of something might be more suitable (Figure c). The scienti c content displayed onto the bare-bone layer may be varied too (i.e., replaced to not overload the graphics; Figure d and e). And nally, other aspects like language or font-size may be changed, again to extend the graphic's usage.
. Specimen : Adding variability Closely related to the previous example, by broadening graphic applicability, the following graphic varies an image's appearance. Avoiding to bias our scienti c understanding by a certain constantly repeated -and hence deeply ingrained -perspective on certain content should be avoided in science to reduce unnecessary mental barriers. World maps are a good and often discussed example as they introduced widespread misconception about the relative size or geometry of landmasses.
A multitude of map projections exist (Figure ). Map projections can be divided into di erent families, which are cylindrical (e.g., Mercator; Figure c), conic (no example shown here), or azimuthal (e.g., Vertical Perspective Azimuthal; Figure b). Furthermore, map projections can be divided based on the properties they preserve, such as shape, distance, direction, scale, and area, and also based to the distortion to be minimised. Map projections are called equal-area (or previously homolographic), when they portray areas across it proportionally to the natural counterparts they represent (e.g., Mollweide; by Undertone.design E ective high-quality science graphics from s-Ink.org / conformal (previously orthomorphic) when the shape is preserved locally on a map (e.g., Mercator; Figure c). This is the case when the scale is the same in any direction and meridians (lines of longitude) and parallels (lines of latitude) cross themselves at right angles. No map can be both conformal and equal-area. A map projection is called equidistant, if distances from the centre (and from the centre only) of the projection to all other places on the map are preserved (e.g., Cassini; Figure e). In addition, maps are also described as equidistant when the separation between parallels is uniform. However, no map projection maintains distance proportionality in all directions from any arbitrary point.
Depending on the scienti c task at hand, one map projection is likely more optimal than the other. In addition, for certain tasks, only some geographic areas are relevant, such as when only oceanic parts have to be shown (as in Figure ). Custom maps like One Ocean (Figure d) might therefore be useful. Providing multiple map projection families and types of the same underlying data set therefore makes such graphics more versatile.
The other reason to provide multiple map versions is to prevent deeply ingrained perceptions of distorted geometries. For example, the relative size of Greenland with respect to Europe is widely misjudged, due to the widespread use of non-Equal-area map projections. Providing additional data set translations or orientations on a map does prevent perceptual and, possibly, conceptual bias risen from a certain appearance. .

Specimen : Facilitating creativity
The next scenario is a graphic template to facilitate creating original graphics. Most academics attend large conferences to communicate their research. While digital and hybrid conference formats are becoming more common, the good old printed poster is still omnipresent. Large meetings with hundreds or thousands of attendants and presentations impede in-depth discussions. It is therefore as crucial to seek relevant scienti c ndings e ectively, as it is to promote a scienti c nding to as many interested peers as possible. Moreover, busy academic schedules often leave little room for carefully designing every single poster. The 'betterposter' design (Morrison, ) o ers a quick-tocreate design solution to e ectively communicate research ndings at large conferences with busy attendees. Powerpoint and Keynote templates (see Figure ) enable (re-)creating a poster in the given design. The visually prominent focus area allows to directly access the main nding discussed on the given poster from a distance. Details are arranged on the side of the poster area to guide interested viewers coming up-close through more in-depth aspects like the methods.
. Specimen : Spreading capability The fth example is a scenario of a gure-editing software program. Such a graphic tool is factually packaging graphic skills (if only in form of a tool) and makes them readily available to the wider science community.
Science communication includes sharing scienti c work in different formats. Original publications are often reformatted into, for example, oral presentations, news articles, or twitter posts. Doing so, often demands to re-use certain graphics. However, most graphics are intended for, and work best with, the speci c canvas they have been produced for. An e ective way to adjust graphics for di erent purposes is therefore not only helpful to the communicator, but ultimately improves science communication itself.
Since the perception of foreground colours strongly depends on the background colour (see e.g., Crameri, Shephard, and Heron, ), image colour inversion is such a critical adjustment to be undertaken for displaying a certain graphic e ectively on a variety of di erent canvases. For example, the image colour inversion tool deLight (Crameri, ) enables science communicators to adjust the colours of an existing image le from working e ectively with a light background to work e ectively with a dark background (Figure ). Apart from some other functionality, the MatLab tool o ers to either invert all colours based on their lightness or, else, to invert only the grey-scale image portions. While using MatLab itself necessitates a licence, the software code is open source (i.e., readable) and therefore reproducible, extendable, and free to use.
After creating science-grade graphics, the next, important step is to make them readily available. All above examples of science graphics, and many more, are now freely and readily available via a novel online collection.

The s-Ink collection
Here, we introduce s-Ink (From Source To Ink, s-Ink.org, a pun on the geological method of source-to-sink), an online collection to enable easy access to openly-licensed, visually engaging, and science-quality graphics that are of wide use. The collection o ers a convenient and time-saving work ow in form of a user-friendly online platform (Figure ) to nd suitable graphics for conveying scienti c aspects in, for example, presentations, proposals, publications, editorial pieces, and news and Wikipedia articles. Moreover, s-Ink supports artistically-inclined scientists that often tend to end up spending a signi cant amount of their work-time crafting graphics for their peers; their e orts are now visible and shared with the entire community under a clearly stated license, so that their individual contribution is acknowledged. The content is downloadable either directly from the s-Ink website or via a link to an external site or repository (for example, in the case of large or pre-existing content).
Apart from being grouped into categories as outlined in Section , products o ered via s-Ink are tagged with several keywords to facilitate nding graphical content within the collection. Both keywords and categories are easily accessible and searchable from the webpage front, accompanying an actual text search tool ( Before publication on s-Ink, suggested graphics will be initially reviewed regarding their usability to a wide audience and on the basis of s-Ink's artistic and scienti c quality standards. The latter are conveniently provided in form of a submission check-list (e.g., on s-ink.org/contribute; see also Section ). The moderators may request the submitter to make changes to the contribution before it appears in the s-Ink collection. .

Community feedback
The s-Ink graphics being accessible for everyone opens a truly powerful avenue: feedback by the entire education, science, and art communities is enabled. After being exhibited in the s-Ink collection, any graphical content in the online collection is open to continuous community feedback, which acts as a reliable and unbiased quality control mechanism. Feedback is exchanged via the comment option at the bottom of each individual graphic's webpage and necessitates a simple login, to prevent spamming. Following constructive comments, graphic products are supposed to be revised by the individual creators, if needed and possible (e.g., a data visualisation based on a numerical simulation is possibly more challenging to change content-wise than a sketch is). A user can therefore be con dent of the good scienti c and graphic quality and wide visual accessibility of the s-Ink content.
. Accessing collection content The s-Ink collection brings together widely useful, key (geo-)scienti c visual products and makes them readily and freely accessible via the centralised online web platform s-Ink.org. The easy access to content is one of its key pillars. It manifests itself by a clear and streamlined web design with an optimised search function, and by the absence of pay walls.
The s-Ink web platform aims to provide a self-explanatory user interface that o ers a familiar environment already upon its rst use, both on desktop computers and mobile devices. At the centre of the interface design is the top (or side) menu bar that includes links to all main pages and access to the search functionality. The search engine -the most important element for a personalised access to the s-Ink collection -is, the most visually prominent feature on the front page (Figure a). It is accompanied by a tag cloud, o ering a topically assisted suggested search option, as well as a category list, to scan through a certain category of graphical products. Including the featured content, a 'latest by s-Ink' suggestion, and an entire gallery page (Figure b), an e cient access to speci c content is thereby guaranteed.
The page design is uid, and adjusts seamlessly, from a large desktop screen size to a mobile device screen size. Spanning all common screen sizes is an always-on 'page up' button giving easy access to the menu and the search functionality. The website's colour design is carefully crafted to be both visually appealing and e ective in presenting (and highlighting) the individual graphical by Undertone.design E ective high-quality science graphics from s-Ink.org / products, whether they may have a white, black, or transparent background. A large collection is only useful if the content can be accessed e ciently and e ortlessly. At the heart of s-Ink online platform is therefore a powerful search engine. It provides, for example, so-called 'fuzzy matching', that e ectively matches partial words, if complete words do not match. Moreover, search results are sorted in the order of relevance, and not date. The search engine can be readily accessed via the s-Ink Front page (Figure a), or the Gallery (Figure b). A page universal search panel is provided in the top menu for the convenience of the user.
. Using collection content One of the intentions of the s-Ink collection is to facilitate the use and creation of science-related visuals for academic purposes. These include for journal articles, books, proposals, teaching, presenting, outreach, and even creating new graphic tools. Therefore, all graphical products in the online collection will be openly accessible, but licensed individually to grant recognition of the time and work invested by their creators. Using s-Ink content will therefore be subject to speci c licensing. The most common licenses used (e.g., CC . , creativecommons.org/licenses/by/ . /, or MIT, opensource.org/licenses/MIT) necessitate to acknowledge the creator(s) in the same way it is common standard across all science disciplines for other academic products. To alter the graphic content of the individual products after download, it must be allowed by the given licence. Moreover, the s-Ink collection curation, storage, and free-of-charge access, should be acknowledged in addition (see Section . ).

. Collection curation and maintenance
The s-Ink collection will update, improve, and extend its content over time. Everyone can suggest high-quality (see Section ) graphics of wide interest to be added to the collection, and provide feedback on existing content. To minimise the storage needs on the ever-growing online collection itself, graphical content may also be linked from external repositories. Artists are also encouraged to put widely-used graphics on permanent repositories like Zenodo (zenodo.org), which additionally facilitates versioning and copyright attribution. The costs accompanying the online platform maintenance are currently covered by an institutional contribution, and led under a community contribution. On the long term, however, the s-Ink collection will rely on sponsoring by academic institutions and/or donations (s-ink.org/donate).

Closing
Science-grade graphics are uncommon, but unmistakably an invaluable tool to scientists. By upholding key graphic design principles, the From-Source-To-Ink (s-Ink) initiative provides an e ective means to distribute (geo-)scienti c knowledge visually amongst peers, but also to the general public. Thereby, it enhances visual science communication for both upcoming research and outreach. Importantly, providing the collection content under a clear creator's license is a crucial step forward towards granting the academic recognition scienti c artists deserve. As an ultimate result, we believe, s-Ink will put the ink back to the source, the pen back into the hand of the (geo-)scientist, and the scienti c knowledge back to the minds of the general public.   by Undertone.design E ective high-quality science graphics from s-Ink.org / Figure A. : Earth-system processes. A schematic highlighting some of the most relevant Earth processes occurring in and on the dynamic, presentday-style Earth, which generates and erases geologic records of its transforming states and is now experiencing unprecedented environmental change. An accurate, up-to-date depiction of features, like a solid (and not molten) mantle, and processes, like mantle convection with spreading ridges not being directly linked to upwelling mantle plumes, is critical to science outreach. Not all displayed individual features are to scale. Specimen available at s-ink.org/earth-processes.
by Undertone.design E ective high-quality science graphics from s-Ink.org / by Undertone.design E ective high-quality science graphics from s-Ink.org /