Global Data Hound

We live in the Holocene era, a postglacial period that began about 10,000 years ago.  In this period, human activity has literally changed the face of the Earth.  Human settlement and the development of agriculture drastically reshaped the original forest cover. (Billington et al. 1996).  William Ruddiman has argued that human land use has had a significant effect on climate since as early as 8000 years ago when newly cultivated rice paddies in East Asia began to inject significant quantities of methane into the atmosphere (Ruddiman 2003). More recently, industrialization and urbanization have caused even more rapid transformations, to the point where the “night lights” of human technological activity are readily visible from space.

As the Earth’s land cover has changed, humans have tracked the changes.  As soon as land is converted to agriculture, the need arises to know the borders and extent of the new land use.  Crops must be planned, resources must be allocated fairly, and intruders must be discouraged.

Broadly speaking, the history of land cover and land use mapping follows the history of cartography. The earliest thematic land use maps date to the 1600s and were the province of surveyors and atlas-makers.  Of course, there were no land use maps in prehistoric times, but historians and climate scientists have worked to fill the gaps in our historical understanding of land use and land cover change with efforts such as the HYDE Historical Land Use and Land Cover (LU/LC) data base (Klein Goldewijk 2001). In the modern era, traditional mapping techniques have been supplemented by satellite sensors such as LANDSAT, AVHRR, MODIS, and MERIS that have provided the basis for consistent, synoptic assessments of global land use and land cover.

It is almost impossible to overstate how important  global LU/LC data sets are to the scientific study of climate change. Land cover change affects albedo, which determines in part the amount of solar radiance reflected by the Earth’s surface back to space: a crucial factor in global warming.  Change in vegetative land cover affects the rate of carbon uptake by the biosphere, another key factor in reducing the impact of net CO2 emissions.  Land cover changes also affect the availability of vital ecosystem services such as fresh water and biodiversity.  For more on this, see  the Working Group I report on the Physical Science Basis of Climate Change (full text PDFs) in the Fourth Assessment Report (FAR) of the Intergovernmental Panel on Climate Change (IPCC), especially sections 2.5 and 7.3.3.1.5 and Box 11.4.

Moreover, land cover or land use type is a parameter for any gridded global data set or model that needs to know what type of terrain covers a given portion of the Earth’s surface. Because such data sets and models are global, it is essential that the LU/LC inputs should also be truly global. There have been three major generations of satellite-based global LU/LC data sets at 1-km resolution:  Global Land Cover Characteristics (GLCC), based on AVHRR data and morphed into 8 different views of the globe;  Global Land Cover 2000 by the Joint Research Centre (JRC) using the VEGA sensor aboard SPOT4; and the MODIS Land Cover product.

So it was a significant milestone when in October 2008, the European Space Agency released a new 300 m resolution, global land use map called GLOBCOVER. GLOBCOVER relies on classification of data from the MERIS multispectral sensor, which has 15 bands in the visible and near infrared range between 0.415 and 0.9 GHz. GLOBCOVER is available via ESA’s Ionia portal. The complete data set is quite large (6 GB for the land cover classification, 140 GB for the MERIS reflectance data set); a BitTorrent client is required for the reflectance data set.

For some examples of what the improved resolution means, compare these images of the area north of the Aswan dams in GLOBCOVER, MODIS Land Cover, and NASA’s “Blue Marble” (l, c, r).

. 

Both GLOBCOVER and MODIS provide more insight into the structure of the “oasis” in the middle of the image than the Blue Marble view from space.

And the Nile Delta (GLOBCOVER, MODIS, the NASA blue Marble):

Note how the layout of the smaller cities in the delta is much more plausible in GLOBCOVER than in MODIS.

GLOBCOVER provides 22 land cover classes. Seventeen of those classes provide fractional classifications (e.g. “Closed (>40%) broadleaved deciduous forest”).  An important trend in recent work with global land cover / land use data sets has been a push towards providing fractional values on a scale from 0 to 1 for each type of interest rather than categorical classifications (the pixel must be classified as type A, B, or C). An example is the fractional tree cover data set created by using MODIS data. Fractional classifications are an improvement over categorical classifications “because they yield improved depictions of spatially complex landscapes,” and provide better insight into trends and change (Hansen, 2003).

GLOBCOVER is exciting because it offers the opportunity for more highly refined understanding and analysis of the changing face of the Earth.  To date, there have only been a few of these”global snapshots” of land cover and land use, taken at several-year intervals and according to different standards. For a sobering comparison of the differing results of land cover data sets, see MacCallum 2006.

The MODIS land cover product is planned in theory to be repeated every year, but it has not actually happened that regularly.  The more basic problem is that the civil Earth observation system is fragile and defined as a “research” effort, so that there is no firm national or global commitment to ongoing updates of satellite land cover and land use.  Yet, with the ever-increasing importance of understanding the impact of anthropogenic activity, I am confident that eventually this problem will be resolved, and eventually, in the big picture, we will see GLOBCOVER as one more step along the way to regular observations of global land cover and land use in an ever more “self-aware” Earth as it is modified by anthropogenic activity.

On that note — one final aperitif for the land cover/land use reader — check out this fascinating new  paper by Ellis and Ramankutty describing the “anthropogenic biomes” of Earth.  IMHO, a landmark study that provides an indispensable new view of land cover and land use.  We will get into this product and maps of population in our next blog.