Science

Creating a Global Digital Soil Map

by Madeline Fisher

On the wall of Alfred Hartemink’s office at the University of Wisconsin–Madison hangs a framed picture from 1960. The World Congress of Soil Science was in town that year, and a large group of soil scientists posed for a photo along the campus shoreline of Lake Mendota—the same lake Hartemink now sees every day from his office window. the world,” he says. “The only one we The historical connection clearly have at the moment.” pleases the Dutch soil scientist and That may be true now, but a second SSSA member, who came to the uni- map is on the way. For the past five versity just last year after living and years, in fact, Hartemink and dozens working all over the world—most of other soil scientists worldwide have quite an endeavor to produce a global recently at ISRIC-World Soil Informa- been working not only on another digital soil map,” says University of tion in Wageningen, the Netherlands. global soil map, but also a digital one Florida soil scientist and ASA and But the real significance of the 1960 this time: a fine-resolution spatial SSSA member Sabine Grunwald, who meeting wasn’t in the setting, but in repository of data on pH, organic leads an ASA Community focused on what the group, known as the Inter- carbon levels, texture, and other soil the project. properties that can be readily and national Society of Soil Science (ISSS), But the map is also desperately continually updated. decided to do. needed. Existing soil maps—like Despite the rifts created by the Named GlobalSoilMap.net, the effort the FAO-UNESCO one—are often Cold War and Suez Crisis at the time, is somewhat like going from a 1980s years or even decades out of date. “They said, ‘Let’s do something bold. printed road atlas to Google Earth in More critically, the soil class infor- Let’s make a soil map of the world,’” one giant leap, and there are major mation they contain can’t easily be recalls Hartemink, who is secretary scientific and technological hurdles used by researchers in other fields, general of the successor organization to overcome. Add the challenges of making it difficult to fully integrate to ISSS: the IUSS, or International reaching consensus on methods, coor- soil science with hydrology, ecology, Union of Soil Sciences. And eventually dinating dozens of institutions around atmospheric science, economics, and they did, although it took 25 years. the globe, and surmounting geopoliti- other disciplines. Instead, scientists in “That’s the FAO-UNESCO soil map of cal obstacles to data sharing, and “it’s these fields often produce their own Image courtesy of iStockphoto/Studio-Pro

4 CSA News October 2012 Creating a Global Digital Soil Map

Backdrop image courtesy of T.J. Purakayastha. Computer-related insets courtesy of (from l to r): David Hopkins, Ohio Department of Natural Resources, Utah State University, and USDA-NRCS.

estimates of soil properties “and that finer scales is actually important in the or to prove to other researchers that scares us,” says Hartemink, who led behavior of ecosystems.” soils are important. GlobalSoilMap.net from the Nether- While this is frustrating, he adds, “The real aim, and maybe we’re be- lands until last year. the real issue is how the lack of digital ing idealistic, is to tackle those larger “So often we see [in studies] that information hinders our ability to global issues,” McBratney says. soils are either not treated at all or know the impact of soils on global are treated as a 50-kilometer average climate change or predict the water- across large chunks of space,” agrees holding capacity—and thus, crop- From Soil Classes to Soil Alex McBratney, a University of growing potential—of soils world- Properties Sydney soil scientist and the project’s wide. In other words, the group isn’t To grasp the significance of the overall scientific adviser. “But we making the map merely because it can project, which officially launched in know that the variability of soils at

October 2012 CSA News 5 Science

New York in 2009, it helps to know organic carbon content,” Hartemink weren’t advanced enough for many something about traditional soil map- says. “And there are a lot of people years to produce a digital map on a ping. After surveying soils or study- who are mainly interested in soil very large scale, but in 2006, the IUSS ing aerial photographs, soil scientists properties, but not classes.” working group thought the methodol- draw boundaries on a map to desig- This is why the IUSS formed a ogy might be ready. Near the end of nate areas, called “polygons,” where working group nearly a decade ago to a meeting in Rio de Janeiro that year, soils differ across a region. Based on study and develop methods for digital the group talked about attempting a the dominant soil types and proper- soil mapping—which is not the same digital map of the available water ca- ties within the polygons, each is then as digitizing existing soil maps. A true pacity of the world’s soils. One or two assigned to a soil class and named ac- digital map doesn’t employ polygons, weeks later, they discussed the pos- cording to a taxonomic system similar for one; instead, it’s broken into a fine sibility again at the World Congress of to the one for plants and animals. For grid of cells, or rasters, which typi- Soil Science in Philadelphia. And by example, just as the potato’s scientific cally range in size from 5 by 5 to 90 by year’s end, they’d organized the first name is Solanum tuberosum, soil in 90 meters. Rather than a class name, official meeting on the global digital what’s commonly called the Antigo each cell has an associated set of soil soil map, chaired by ASA and SSSA series is classified as a “Haplic Glos- property data. And these soil values Fellow Pedro Sanchez at Columbia sudalf.” are often predicted from other types University. Plants and animals are distinct of data, such as climate, relief, and The meeting resulted in a proposal, beings, however, whereas different geology, rather than being directly which in turn attracted $1 million in soils blend gradually into one another measured. seed money from the Bill and Melinda across the landscape, Hartemink Most importantly, when the global Gates Foundation. (The foundation says, making the polygon boundaries digital soil map is complete, it will also gave another $18 million to map between them somewhat arbitrary. A provide a measure of the uncertainty the continent of Africa, where soil bigger issue is that although soil maps associated with each soil property es- information is especially sparse.) The and class names are invaluable to soil timate, Hartemink says. For example, seed funds ran out before most of the scientists, land managers, and farm- scientists will be able to say that the scientific work began. But the money ers, they are less useful to modelers pH at a certain spot is 6.5 plus or mi- did allow the creation of a consor- and other researchers who need quan- nus 0.3, or “we’re 80% sure,” he says. tium and network of administrative titative, spatial data on pH, organic “We’re very used to having a prob- “nodes” that now covers nearly the carbon, bulk density, and other soil ability to the weather: A 40% chance entire planet, says Neil McKenzie of properties. of showers. But we haven’t done that Australia’s Commonwealth Scientific “It’s quite difficult to get that a lot in soil science.” and Industrial Research Organisation (CSIRO), which assumed leadership information from the class. The name Much of the holdup was that

Photos courtesy of (clockwise): iStockphoto/Bryngelzon, The Forest Watershed Research Center at the University of New Brunswick, and Jim Bockheim Research Watershed Photos courtesy of (clockwise): iStockphoto/Bryngelzon, The Forest of the project after Hartemink left tells you relatively little about pH or mapping and computing techniques ISRIC in 2011.

6 CSA News October 2012 McKenzie and CSIRO, for example, lead the Oceania node; Jon Hempel of USDA-NRCS in Lincoln, NE, over- sees the North American node; and Lou Mendonça Santos of EMBRAPA in Brazil directs the Latin American node. There is also a network of “sup- porting” institutes (see map) and doz- ens of partner organizations. In North America alone, for instance, more than 10 universities and agencies are involved, including Agriculture and Agri-Food Canada and the National Institute of Statistics, Geography, and Informatics in Mexico. Indeed, in its scope and ambition, the venture is not unlike the Human Genome Project, whose comple- GlobalSoilMap.net is organized into eight “nodes” around the world, led by sci- tion also took the combined work of entists at the following institutions: USDA-NRCS in North America; EMBRAPA scientists and research institutions (Brazilian Agricultural Research Corporation) in Latin America/Carribean; JRC worldwide. Yet, while that project was (Joint Research Centre of the European Commission) in Eurasia; Jordan’s Min- fueled by vast sums of money—not istry of Agriculture CUMERC in North Africa/West Asia; TSBF-CIAT (Tropical Soil to mention high-octane personalities Biology and Fertility Institute of the International Center for Tropical Agriculture) like Craig Venter—GlobalSoilMap.net in Africa; ISSAS (Chinese Academy of Sciences) in East Asia; and CSIRO (Com- has so far been driven mainly by the monwealth Scientific and Industrial Research Organisation) in Oceania. A node commitment of the scientists, Hempel leader for South Asia is pending. says. Several other institutions also help to coordinate the project: the Earth Institute “It’s amazing to me that we could of Columbia University in the United States; the University of Sydney in Austra- pull together a consortium of people lia; ISRIC in the Netherlands; and INRA (French National Institute for Agricultural from around the world and stay Research), which recently replaced IRD (International Relief and Development) together on the idea, considering the as a supporting institution of GlobalSoilMap.net in Europe. distance and communication issues,” says the SSSA member. “I mean, all of the nodes have remained engaged and Another key decision was which actual numbers from soil analyses, are pretty dedicated to the concept.” soil properties to focus on: They class information from traditional soil include organic carbon, pH, texture, maps, and data on climate, relief, and Sorting Out the Details available water capacity, and depth other environmental factors. to bedrock or restrictive layer. Others, The second question is especially So what has the team been do- such as cation exchange capacity, can critical given the project’s aim of ing? “Although it probably seems then be predicted by using what are using a century’s worth of “legacy” dry, we’ve spent several years getting called pedotransfer functions. “That’s soil information to the fullest extent agreement on the technical specifica- why those particular properties were possible. McKenzie explains that tions of the project,” McKenzie says. chosen,” Hempel says. “They give us countries like Denmark and South “We’re essentially building a very the most flexibility in modeling other Korea have sampled and analyzed large system for providing web-based properties.” their soils in exquisite detail, provid- services relating to soil properties. So, If agreeing to all this on paper has ing a rich supply of “point” data that the size of the dataset and how we been a challenge, however, it’s noth- are relatively easy to put into digital generate the estimates of the proper- ing compared with actually making format. But in other countries, such ties is quite a large thing to sort out.” the map. New scientific questions as the United States, most soils data For example, the consortium has and problems have emerged con- must be extracted from the polygon settled on a spatial resolution for the stantly, McBratney says, such as how information in traditional soil maps. map of 100 by 100 meters, and it de- to model soil properties with depth This involves converting the polygons cided to offer data down to 2 meters, (they’ve come up with an innovative into grids of cells, deducing from at six depth intervals: 0 to 5 centime- solution, called the spline technique) soil class information which soils are ters, 5 to 15 centimeters, and so on. and how to utilize and mesh together present within each polygon, and then different types of data, including

October 2012 CSA News 7 Science

assigning soil properties to the cells uncertainty values are fairly easy to adds McKenzie. “It’s quite a challeng- individually. calculate with standard statistical ing time to be getting large grants In North America, much of the techniques. But when soil properties internationally as I’m sure people continent’s existing soil class, or poly- are predicted from polygon infor- can well appreciate.” Still, he notes, gon information, has been converted mation through the disaggregation CSIRO is close to securing funding to a grid format. However, until the process, not only is the uncertainty for a core technical team that will polygons are “disaggregated” into dif- naturally higher, but computing it is coordinate the implementation of the ferent soil types, the cells created from tougher, as well. “So at our current project worldwide. And, meanwhile, individual polygons will all contain stage, we are pretty good with the the transition in leadership that began the exact same soil properties, Hempel point data, but we’re still figuring out when Hartemink stepped down as explains; predicting how properties how to deal with the disaggregation director is now complete. vary from cell to cell will come later. model,” McBratney says. One aspect of the global digital soil But before the data are publicly map project that many don’t appreci- released, there’s another hurdle to Building Collaborations ate is how much time is spent in ne- surmount: calculating the uncertainty gotiations and diplomacy, McKenzie associated with each soil property Asked where the entire project adds. Soils, he explains, are sovereign value. This is tricky, McBratney says, stands, and McBratney falls back on a resources that play a huge part in each because in addition to the uncertainty familiar line. “Let’s say what all scien- country’s capacity to grow food and that comes from the inherent variabil- tists say: In five years’ time,” he says maintain clean water supplies. So, for ity of the soil, uncertainty arises from with a laugh. “That’s far enough away this reason, a system where countries the quality and density of the data that it’s not tomorrow.” hand over their soils information to itself. In regions where soils have been “Everyone in the consortium a central organization won’t work. sampled and analyzed at high density, would love for it to be going faster,” Individual countries instead have to be responsible for maintaining their own data, he says, or, if they can’t, to be willing to collaborate with a part- “This is not a project that can be switched ner country that manages their soils information for them. And then coun- on overnight. There’s a lot of building of tries have to agree to share their data, which comes with its own sensitivities collaboration and coordination around the and complications.

Photos courtesy of (clockwise): Janis L. Boettinger, the CaliforniaLab at University of California–Davis, Soil Resource and iStockphoto Photos courtesy of (clockwise): Janis L. Boettinger, world to make it happen.”

8 CSA News October 2012 “So this is not a project that can be switched on overnight,” McKenzie For More on This Topic.... says. “There’s a lot of building of col- laboration and coordination around There will be two symposia focused on the global digital soil map at the the world to make it happen.” Annual Meetings in October: Whenever it happens, he and the • Global Soil Mapping In a Changing World: I (part of ASA’s Global others expect the impact to be sub- Agronomy program): http://scisoc.confex.com/scisoc/2012am/ stantial. With the Anthropocene—the webprogram/Session9682.html Age of Humans—upon us, Grunwald • Global Soil Mapping in a Changing World (part of SSSA’s Pedology believes the map will be a critical tool program): http://scisoc.confex.com/scisoc/2012am/webprogram/ for knowing how soils are forming, Session10052.html changing, and degrading in response Also, check out ASA’s Global Digital Soil Map Community at to human activities. The collabora- www.agronomy.org/membership/communities/global-digital-soil-map. tions forged by the map between soil science and other disciplines should foster wide understanding of soil’s role in global issues like food security, the data with additional soil survey their countries’ data fit into the global biodiversity, drought mitigation, and and sampling work. system, whether they’re from Chile or climate change, Hempel says—as well Even before it’s finished, in fact, Mozambique or Europe or elsewhere as help address them. And because the global digital soil map seems to in the world,” McKenzie says. “That’s soils information is scant in many be making an impression. “We feel very heartening for us to see.” regions, Hartemink thinks the uncer- that there is a whole new generation tainty associated with the map’s soil of very capable soil scientists who are M. Fisher, science communications property predictions will frequently picking up on this project and start- manager be high, sparking efforts to improve ing to figure out how they can make Canopy SunScan SS1 Canopy Analysis System • Direct display of Leaf Area Index (LAI) • Usable in cloudy, clear and changeable conditions • Measures incident and transmitted PAR in plant canopies

Delta-T Devices Ltd www.delta-t.co.uk

October 2012 CSA News 9