The field of data visualization is currently undergoing a significant shift in its understanding of color theory and its practical application in scientific communication. For decades, the "rainbow colormap"—a spectral progression of hues typically ranging from blue to red—has been a subject of intense scrutiny and frequent condemnation by visualization experts. Critics have long argued that these colormaps are perceptually misleading, mathematically non-uniform, and exclusionary to those with color vision deficiencies. However, a landmark paper published in the 2023 edition of Computer Graphics & Applications, titled "Rainbow Colormaps Are Not All Bad," is challenging this long-standing orthodoxy. Authored by a collaborative team of veteran and contemporary researchers, the study suggests that while the rainbow colormap has documented flaws, it remains a highly effective tool for specific analytical tasks, particularly in meteorology and geographical data analysis.
The Historical Context of Color Mapping Critique
To understand the current debate, one must look at the history of color in digital displays. In the early days of computer graphics, the "Jet" or "Rainbow" colormap became the default standard in many software packages, including MATLAB and various weather forecasting tools. Its popularity stemmed from its high contrast and the intuitive association of red with "hot" and blue with "cold."
However, by the mid-2000s, a vocal movement within the academic community began to highlight the "harmful" nature of these scales. The seminal 2007 paper by David Borland and Russell M. Taylor II, "Rainbow Color Map (Still) Considered Harmful," laid out the primary grievances that would define the next fifteen years of visualization best practices. The critique focused on three main areas: the lack of perceptual uniformity, the introduction of "false contours," and the lack of a natural ordering.
Perceptual uniformity refers to the idea that a change in the data value should correspond to a perceived change in the visual representation. In a rainbow scale, the luminance (perceived brightness) fluctuates wildly. For instance, the yellow and green sections of a rainbow often appear much brighter than the red or blue sections, even if they represent mid-range values. This can lead the human eye to perceive "boundaries" or "edges" where none exist in the underlying data—a phenomenon known as false contouring. Despite these academic warnings, the rainbow colormap remained the public face of data, most notably in the daily temperature maps produced by the National Oceanic and Atmospheric Administration (NOAA) and various global news outlets.
A Chronology of the Shift Toward Perceptual Standards
The movement to replace the rainbow colormap gained significant momentum in the 2010s. In 2015, the development of the "Viridis" colormap for the Matplotlib library represented a turning point. Viridis was designed to be perceptually uniform and accessible to those with colorblindness, moving from dark purple to bright yellow in a consistent increase of luminance. This prompted many scientific journals and software developers to ban or discourage the use of traditional rainbow scales.
By 2020, the consensus in the data science community was almost entirely anti-rainbow. However, practitioners in fields like oceanography and meteorology continued to use spectral scales, citing their utility in "value-reading" tasks. This disconnect between academic theory and professional practice prompted the 2023 study by Colin Ware, Maureen Stone, and Danielle Albers Szafir. Their research sought to reconcile why a "scientifically inferior" tool remained so pervasive and effective in the field.
Analysis of the 2023 Findings: Where Rainbows Excel
The 2023 paper, "Rainbow Colormaps Are Not All Bad," provides a nuanced analysis of how different colormaps perform depending on the user’s objective. The authors categorize visualization tasks into two primary types: "shape-perception" tasks and "value-reading" tasks.
Supporting data from the study indicates that while rainbow colormaps are indeed poor for perceiving the three-dimensional shape of a surface, they are superior for reading specific values off a map using a legend. In a maximum temperature map of the United States, for example, a viewer is often trying to determine the specific temperature in a particular county. Because the rainbow scale uses distinct hues (red, orange, yellow, green, blue, purple), the viewer can more easily match a color on the map to a specific point on the key. In contrast, a monochromatic or perceptually uniform scale like Viridis may make it difficult to distinguish between 85 degrees and 90 degrees because the subtle change in shade is less "nameable" than the transition from yellow to orange.
Furthermore, the research highlights that rainbow scales are effective at drawing attention to global structures and the relationships between different data distributions. The high contrast provided by hue changes allows the human brain to categorize regions of data quickly, which is essential in time-sensitive fields like emergency weather management.
Addressing the "Strawman" of Shape Perception
A significant portion of the paper addresses the common criticism that rainbow colormaps distort the perception of "shape." Critics often use the example of a 3D-rendered object, such as an apple or a portrait, to show how a rainbow scale destroys the sense of depth and form. Because the human visual system uses luminance to understand 3D shapes (shading and highlights), the fluctuating luminance of a rainbow scale creates a confusing visual signal.
However, the authors argue that this critique is often a "strawman" in the context of 2D data visualization. Most maps, such as the NOAA temperature maps, are not intended to represent 3D topography or complex surfaces. They are 2D spatial distributions where the primary goal is identifying intensity and location. In these instances, the loss of shape perception is a negligible cost compared to the benefit of high-resolution value identification. By recognizing where shape perception is relevant—and where it is not—designers can make more informed choices about when a rainbow scale might actually be the optimal tool.
Technical Data and Perceptual Uniformity
The study acknowledges the mathematical deficiencies of the standard "Jet" colormap, which has been the primary target of criticism. The "Jet" scale has a notorious luminance spike in the yellow region, which creates a sharp visual break. However, the authors point out that modern "optimized" rainbows, such as Google’s "Turbo" colormap, have addressed many of these issues.
Turbo was designed specifically to provide a rainbow-like progression while smoothing out the luminance transitions and ensuring that the scale is more accessible to those with certain types of color vision deficiency. Data shows that Turbo maintains the high-contrast benefits of the rainbow while minimizing the false contours that plagued earlier versions. This evolution suggests that the problem is not the "rainbow" concept itself, but rather the specific mathematical implementation of the colors.
Implications for Scientific Communication and Policy
The implications of this research are far-reaching for the fields of cartography, medical imaging, and climate science. If the "rainbow is always bad" mantra is retired in favor of a task-based approach, it could lead to a more sophisticated set of standards for data visualization.
- Contextual Standards: Instead of a blanket ban, organizations like the IEEE or the American Meteorological Society may move toward "task-specific" recommendations. For instance, sequential scales for 3D surfaces and spectral scales for 2D value-lookup maps.
- Accessibility Integration: The push for better rainbows will likely lead to increased adoption of scales like "Turbo" or "Magma," which provide the hue variety users crave while maintaining the accessibility standards required for public-facing information.
- Software Defaults: Software developers may begin to offer "smart defaults" that change based on the type of data being plotted (e.g., discrete vs. continuous, 2D vs. 3D).
Official Perspectives and Industry Reaction
The authors of the study represent a powerful cross-section of the visualization community. Maureen Stone, a former manager at Tableau Research and an expert in color science at Xerox PARC, brings decades of industrial experience to the findings. Colin Ware is a leading figure in the perception of oceanographic data, and Danielle Albers Szafir represents the new generation of researchers focusing on the intersection of human-computer interaction and data science.
Their collective "call to action" at the end of the paper suggests a pivot in the research community. They argue that instead of spending more effort dismissing rainbows, researchers should focus on finding out "when and why they are good." This sentiment has resonated with many practitioners who felt that the academic "war on rainbows" had become disconnected from the practical needs of users who find multi-hue scales more engaging and easier to interpret for everyday tasks.
Conclusion: A Balanced Future for Data Coloration
The 2023 reevaluation of rainbow colormaps marks a maturing of the data visualization field. It signifies a move away from dogmatic "rules" toward a more nuanced, evidence-based understanding of human perception. While the traditional, unoptimized rainbow scale will likely never return to favor in high-precision scientific modeling due to its inherent distortions, its refined successors are proving to be indispensable.
As data becomes increasingly central to public discourse—from tracking global temperature rises to visualizing pandemic spread—the ability to communicate complex information clearly and accurately is paramount. The research by Ware, Stone, and Szafir provides a roadmap for using one of the most popular, yet controversial, tools in the visual arsenal with greater intent and effectiveness. By acknowledging the strengths of the rainbow colormap for value-reading and global structure identification, the visualization community can move toward a future where color is used not just for aesthetic appeal, but as a finely tuned instrument for human understanding.