Designing maps for the color vision impaired.

Cartography involves the drawing of maps, which in fact is both a science and an art. I’ve come to learn in the course of my years of study and professional experience that mapmaking is not as easy as it seems. Maps tell a story. And for maps to effectively tell that story, the mapmaker should think not only about the main subject of the map, but should also give due consideration to the various map elements such as their position and balance across the map, as well as the accuracy of the measurements, and the styling of these elements which involve, for example, the choice of colors, line weights, among other considerations.

To be honest, although I’ve been making maps for my entire professional career, I have never considered color vision impairment in the design of maps, at least those that I have made myself. (For instance, the green and red color scheme that I consistently use to depict forest loss and forest gain would be the same shade of gray for a person with color vision impairment.) I have always kept in mind the psychology of colors in maps to help get the message across to the users of these maps, thereby allowing them to better understand the story that the map intends to tell. But unfortunately, I have not put myself in the shoes of a person with color blindness and have given enough thought on how they would perceive and understand those maps. At least not yet.

The opportunity to rectify this finally came in the process of finalising a map figure for the final proof of our recently accepted paper for publication. A colleague of mine, G, suggested that we consider designing the color scheme of the figure to be more friendly for color blind people. It was a brilliant suggestion. I had not thought of it at first (yet again) and I welcomed it because it gave me the chance to learn and design maps for both people with normal vision and color vision impairments.

To implement this, luckily an app called Color Oracle existed, which is a free color blindness simulator that can be used across computer platforms. It shows what people with common color vision impairments would see in real time, thus allowing a person to select appropriate colors in designing the color scheme of a material, website, or publication. Color Oracle basically applies a color filter over the full screen of your computer regardless of the software in use.

For example, the screenshots of the same portion of the map figure below was the result of the Color Oracle app showing me what a person with normal vision would see and what a color vision impaired person would see. Through these color filters, I then selected colors that would ensure the images can be read by people with normal vision and color blindness. It adds another layer of consideration in designing maps, even making it more challenging, but this ensures that the map can be read by and be accessible to the widest possible audience.

Screenshot 2017-09-16 19.22.57

Screenshots of the same portion of a map figure showing the colors that people with normal color vision would see (leftmost image) and the colors that people with color vision impairments would see, particularly those with deuteranopia (2nd from left), protanopia (3rd from left), and tritanopia (rightmost image).

For additional reading, visit the Color Oracle website and read the papers by Jenny & Kelso (2007a, 2007b) to learn more about the design of information graphics and maps that are accessible for color-impaired readers.


Jenny, B., Kelso, N. V. (2007a). Designing maps for the color-vision impaired. Bulletin of the Society of Cartographers SoC, 41, 9-12. [PDF]

Jenny, B., Kelso, N. V. (2007b). Color design for the color vision impaired. Cartographic Perspectives, 58, 61-67. [PDF]


New paper on untangling the complex deforestation drivers in Myanmar.

I am happy to announce our recently accepted paper for publication in Conservation Biology journal. The paper entitled, Untangling the proximate causes and underlying drivers of deforestation and forest degradation in Myanmar, is the product of our research lab, and a study led by Ms Cheng Ling Lim and Dr Graham Prescott under the supervision of Prof Edward Webb (our lab’s lead principal investigator) and Prof Alan Ziegler (our collaborator and lead principal investigator of the Wet Lab at the NUS Department of Geography) [1]. The study is part of our ongoing project about understanding the proximate and underlying factors leading to deforestation and land use change in Myanmar, which is funded by the Ministry of Education, Singapore.

In the study, we conducted a systematic review of published papers regarding deforestation in Myanmar, and applied a system dynamics approach and causal loop network analysis to map and visualise the linkages among deforestation drivers. We constructed causal-network diagrams to comprehend the nexus between deforestation and forest degradation and their proximate causes, as well as the relationships among proximate causes and underlying drivers (see Geist & Lambin’s seminal 2002 paper regarding the proximate causes and underlying driving forces of tropical deforestation) [2].

Our study showed that proximate causes included infrastructure development, timber extraction, and agricultural expansion of rice and boom crops such as oil palm and rubber. These were facilitated mainly by formal agricultural, logging, mining, and hydropower concessions, economic investment, and social issues relating to civil war and land tenure. Reform of land laws, the link between natural resource extraction and civil war, and the allocation of agricultural concessions will influence the extent of Myanmar’s deforestation and forest degradation in the future.

As Myanmar transitions towards a more open economy, a complex array of pressures to convert its biologically diverse forests is expected to intensify over at least the next decade [1, 3-4]. Through the causal-network analysis, we were able to identify priority areas for policy interventions to reduce forest loss and degradation such as creating a public registry of land-concession holders to deter corruption in concession allocation. This analytical approach can also be applied to other countries, particularly those undergoing political transition, to inform policies aiming to reduce pressures to forests.

In the coming months, we will be presenting the results of our studies, particularly this causal network paper and an upcoming horizon scanning paper (more on the latter soon), to government leaders, policymakers, and development organisations in Myanmar to inform policy and actions in support of conserving its globally important forests and biodiversity. I hope to share about that presentation soon, so watch this space.

In the meantime, let me share one of the maps I’ve drawn for our new paper, which was in one of the earlier drafts but we eventually did not include in the final version. The map shows the proximate and some of the underlying causes of forest loss in Myanmar, together with the locations of case studies and the percentage of deforestation in each administrative unit.


map_ms-conserv-biol_v9 fig1a

Some of the proximate and underlying drivers of forest loss in Myanmar, together with the locations of case studies and the percentage of deforestation in each administrative unit.



[1] Lim, C.L., Prescott, G.W., De Alban, J.D.T., Ziegler, A.D. & Webb, E.L. (In press) Untangling the proximate causes and underlying drivers of deforestation and forest degradation in Myanmar. Conservation Biology. doi:10.1111/cobi.12984.

[2] Geist, H.J. & Lambin, E.F. (2002) Proximate causes and underlying driving forces of tropical deforestation. BioScience, 52, 143–150. doi:10.1641/0006-3568(2002)052[0143:PCAUDF]2.0.CO;2.

[3] Webb, E.L., Phelps, J., Friess, D.A., Rao, M.V. & Ziegler, A.D. (2012) Environment-friendly reform in Myanmar. Science, 336, 295–295. doi:10.1126/science.336.6079.295-a.

[4] Webb, E.L., Jachowski, N.R.A., Phelps, J., Friess, D.A., Than, M.M. & Ziegler, A.D. (2014) Deforestation in the Ayeyarwady Delta and the conservation implications of an internationally-engaged Myanmar. Global Environmental Change, 24, 321–333. doi:10.1016/j.gloenvcha.2013.10.007.

* In case you’re interested to read our paper but are unable to access it through a paywall, please send me a message and I can gladly provide you with a copy.


Learning Open Foris tools (Part 1).

Here’s another suite of software tools that land change scientists and geospatial analysts should have in their toolbox: Open Foris.

Screenshot 2017-08-29 13.24.52

The suite of Open Foris software tools for environmental monitoring.

Open Foris is a set of free and open-source software tools designed to facilitate flexible and efficient data collection, analysis, and reporting for environmental monitoring such as forest inventories, climate change reporting, socio-economic surveys, biodiversity assessments, land use/cover change assessments, among others [1]. This initiative, resulting from the collaborative efforts of numerous public and private institutions, is hosted by the Food and Agriculture Organisation of the United Nations.

At the moment, I am specifically interested in learning Collect Earth, one of the Open Foris tools that enables data collection through the Google Earth interface, to streamline my image classification and analysis workflow in Google Earth Engine. Used in conjunction with Collect, another tool for designing survey forms and managing survey data, bespoke data entry forms can be setup and streamlined with a user-friendly interface, say, for land use/cover change assessments. (Note that there are a few other tools included in the Open Foris suite such as Collect Mobile, Calc, and Geospatial Toolkit, which I have not yet explored but could still be useful and relevant in support of my research in the near future.)

In a nutshell, Collect Earth essentially facilitates the interpretation of high and medium spatial resolution imagery available in Google Earth, Bing Maps, and Google Earth Engine. It simplifies the data entry process in land use/cover change assessments (e.g., identifying regions of interest for classification and analysis), synchronises the view of each sampling point across all three platforms to streamline the process of reviewing satellite imagery, and seamlessly integrates this database of sampling points with Google Earth Engine for wall-to-wall image classification.

For this post, I will just share a few notes regarding the installation and setup of the Collect Earth system for Mac OSX. Basically I just followed the installation tutorial from the Collect Earth website. After installing Collect Earth and Google Earth Pro, I launched Collect Earth and went ahead with the setup, which involved indicating the operator name and tinkering with settings such as language and browser preferences.

Screenshot 2017-08-29 14.49.27

The Advanced tab of the Collect Earth Options dialog box for Mac OSX.

Collect Earth allows the user to modify optional settings, particularly found in the Tools > Properties > Advanced tab such as selecting a preferred browser (i.e., Firefox, Chrome). For my system, I opted to check the boxes to open Chrome browser windows with Google Earth Engine Playground as well as its timelapse and zoom into the plot area. I also opted to open Bing Maps. In a Mac OSX system, the paths to the Firefox and Chrome executables can be specified as follows:

/Applications/Google Chrome

(I did not tinker with the settings related to the survey data, particularly under the Sample Data, Plot Layout, Survey Definition tabs as I figured it will go together with using the Collect tool.) Once my settings were defined, I took Collect Earth for a spin to check whether the synchronised interfaces worked by using one of the project examples from their website.

However, after loading one of the project files and clicking at one of the plot areas, I observed that the HTML survey form designed for the project had popped out, but the synchronised browser interfaces failed to launch. After consulting the support community and the log file (Help > Open Application Log File), I learned that Collect Earth was unable to launch the Chrome/Firefox browsers despite having specified the correct pathnames to the browser applications.

To rectify this, I had to open the file found in /Library/Application Support/CollectEarth under my user account (or just go to Tools > Open Data Folder). I figured out that the pathnames to the browser applications in the file were incorrect despite having correctly entered them in the Collect Earth Options dialog box (perhaps it is a bug in the software). And so, I edited the file directly by replacing the following path lines:

chrome_exe_path=/Applications/Google Chrome

After a bit of troubleshooting and rectifying the erroneous path entries, I gave it another try until the synchronised interfaces finally worked. The following screenshots show the different synchronised browser interfaces that I selected.

This slideshow requires JavaScript.

The next post will be about designing bespoke survey forms for Collect Earth using the Collect tool. Watch this space.



[1] Bey, A., Sánchez-Paus Díaz, A., Maniatis, D., Marchi, G., Mollicone, D., Ricci, S., et al. (2016) Collect Earth: land use and land cover assessment through augmented visual interpretation. Remote Sensing, 8, 807. doi:10.3390/rs8100807