Data application of the month: Mapping land cover

GLC-SHARE: Global land cover map based on data from multiple sources launched in 2014

What are land cover maps used for?

Land cover information is important for many applications like flood modeling, observation of agricultural drought, climate change modeling, and monitoring of environmental changes including vegetation phenology, flooding, fire occurrence, and monitoring of carbon emission due to deforestation and forest degradation.

Land cover is the observed (bio)physical cover of the Earth’s surface. Typical land cover classes include, for example, forest, crops, grassland, water surfaces, or artificial cover. Land cover interacts strongly with the water cycle and the climate system. The type of land cover determines the behavior of precipitation on the Earth. This includes the infiltration rate, runoff, and evapotranspiration of water. The spatial distribution of different land cover types is therefore a decisive factor for water and climate models. In hydrology, a so-called curve number is used to predict runoff and infiltration. The value of the curve number depends on the soil and land cover type. Land cover maps are needed to determine the spatial distribution of different curve number values within a catchment. A methodology how to do this step-by-step using GIS software is recommended by the University of Purdue (cf. recommended practice). By manipulating the land cover map of a catchment to match different land cover change options (for example industrial development, housing, or deforestation), this methodology can also be used to simulate floods for different land use planning decisions and to support informed decision making.

Another example of using land cover information for disaster risk reduction is agricultural drought monitoring. For example, the Food and Agriculture Organization (FAO) extracts a crop mask from land cover information to calculate the Agricultural Stress Index for cropland. UN-SPIDER’s Regional Support Office in Iran, the Iranian Space Agency, recommends a similar approach. Detailed step-by-step procedures on this recommended practice will be available soon here on the UN-SPIDER Knowledge Portal.

 

How is land cover mapped from space?

Current global land cover maps are based on medium resolution optical satellite imagery, for example AVHRR, MODIS, SPOT-Vegetation, and MERIS. High resolution optical imagery, for example Landsat and SPOT-4/5, IRS LISS III and RapidEye is used for regional land cover maps as well as for most recent global products. Basically, the approach to classify satellite imagery is to group together similar pixels and to assign class names to each group of pixels thus transforming data into information. Different types of classification algorithms can be applied. In supervised classification approaches, the class names are defined a priori and known pixels are used to train the unknown ones. Object-based approaches make use of more than just the spectral information of single pixels; they take into account, for example, the relation to neighboring objects, the shape, homogeneity, size and color of the object. Multi-temporal approaches are used to separate very similar classes, for example different crop types, by looking at different plant growth patterns. Multi-sensor fusion of optical and synthetic aperture radar (SAR) data is applied to increase the number of acquisitions within a phenological cycle and to improve the classification accuracy. The choice of the most suitable classification algorithm depends on different factors like data availability or the number and type of land cover classes.

The drawback of having so many different classification algorithms in place is that available land cover maps are often not comparable. The international remote sensing community including GEO and GTOS has worked on the harmonization of gloabl land cover products and standardization of classification systems to improve interoperability of different land cover maps. The Land Cover Classification System (latest version: LCML – LCCS v.3; ISO 19144-1) aims at providing a globally applicable classification scheme. Progress is also made in the development of validation standards, for example by the CEOS Cal/Val Working Group.

 

How can I access land cover maps?

Many land cover datasets are available for free. This includes global datasets based on medium to coarse resolution satellite imagery, for example

  • MODIS land cover products: years 2001-2012; 500m-5,600m resolution (link to the data)
  • Land cover products from ESA's climate change initiative: years 1998-2012; 300m resolution (link to the data)
  • Global land cover by national mapping organisations - GLCNMO: years 2003-2008; resolution 500-1,000m (link to the data)
  • Global land cover GLC2000 by the European Commission's Joint Research Center (link to the data)
  • ESA's Globcover map: year 2009; 300m resolution: year 2000; 1,000m resolution (link to the data)

With the GLC-SHARE map, FAO provides a "best of" global land cover map. Through harmonization and standardization of classification systems, FAO has fused best available land cover maps for all regions to produce this global product, which was released in October 2014. (link to the data)

China has released in summer 2014 the first global land cover map based on Landsat data at 30m resolution. (link to the data)

For many countries, national land cover maps are available based on high resolution Landsat or SPOT-5 imagery.  National land cover data are included in the UN-SPIDER database on data sources, which can be filtered by countries (link to the data). For example, the Space and Upper Atmosphere Research Commission (SUPARCO), Regional Support Office of UN-SPIDER in Pakistan, has published two land cover atlases in Pakistan based on 5m resolution SPOT-5 imagery, which are available for download in pdf format (link to the data). For many African countries, land cover data are available from FAO's Africover project including Sudan, Kenya, Senegal, Libya, Egypt, Eritrea, and DR Congo (link to the data).

 

Further reading: A recent review on the current status and future trends of global land cover maps has been published by Brice Mora et al. (2014) (dx.doi.org/10.1007/978-94-007-7969-3_2).

 

Land cover datasets are included in the UN-SPIDER database on data sources. If you are missing a dataset or if you have any suggestions regarding the database, contact us via the the contact button on the top right of this page, please.

 

Zircon - This is a contributing Drupal Theme
Design by WeebPal.