Land cover
Historical land useThe update of HYDE described here includes several improvements compared to its predecessor: (i) the HYDE 2 version used a Boolean approach with a 30 minute degree resolution, while HYDE 3 uses fractional land use on a 5 minute resolution; (ii) more and better sub-national (population) data (Klein Goldewijk, 2005) to improve the historical (urban and rural) population maps as a basis for allocation of land cover; (iii) updated historical land-cover data for the period 10,000 B.C.-2,000 A.D.; (iv) implementation of different allocation algorithms with time-dependent weighting maps for cropland and grassland.
The HYDE 3 historical land-cover inventory is based on a number of general concepts for allocating land cover. We argue that early settlers or pioneers in large parts of the world (e.g. U.S.A. in early 18th century, South America, South Africa and Australia) were quite limited in their decision on where to settle and start agricultural activity. Accessibility was limited in many regions, either by the nature of the surrounding landscape (swamps, mountains, dense forests, poor soils, unfavourable climate) or the hostile indigenous population. In addition, large parts of the world were not accessible due to lack of infrastructure. This limited the early spreading of agriculture considerably. Human population growth can be regarded as the main driving force of global change over time. Therefore, it is crucial to get a good insight of the demographic developments of the past. Historical population numbers of McEvedy & Jones (1978), Livi-Bacci (2007), and Maddison (2001) form the basis of our national historical population estimates. Supplemented with the sub-national population numbers of Populstat (Lahmeyer, 2004) and many other sources, time series were constructed for each province or state of every country of the world. For simplicity reasons, current administrative units were kept constant over time, and every historical source was adjusted to match the current boundaries of HYDE 3.1 (e.g. by taking fractions of former larger empires). Spatial patterns were obtained by using weighing maps based on the population density map patterns of Landscan (2006) for current time periods, and gradually replacing them with weighing maps based on proxies such as distance to water and soil suitability when going back in time. See for a full description of the methodology in Klein Goldewijk et al (2009). Global population increased from 2 to 6,145 million people from 10,000 B.C. to 2000 A.D., resulting in a global population density increase of < 0.1 cap km-2 to almost 46 cap km-2 and a urban built-up area evolving from almost zero to 0.5 million km2, still only <0.5% of the total global land surface (Klein Goldewijk et al., 2009). It is clear that this demand of food, services, and building materials has had profound impact on the Earth’s environment through deforestation and conversion of land cover. Input data for land use from historical statistics Country totals for cropland and pasture Starting point are the country totals for cropland and pasture from the FAO (2008), who present data for the post-1961 period on a country basis. Divided by the country population it yields a per capita use of cropland and pasture. For the pre-1960 period a following approach was used. We assumed that the per capita values for cropland and pasture are not constant, but slightly increase or decrease over time. The 1960 value is rather on the low side for many countries since population numbers have exploded after 1950 and have lowered the per capita numbers considerable. However, when going further back in time, population numbers were lower which increased the cropland and pasture areas per capita again, but they are ceiled by the lack of technology and thus limited the maximum amount of land that a subsistent farmer could handle. By estimating country by country the per capita use of cropland and pasture we derived the historical pathways of agricultural areas, see table 2. Satellite maps For the satellite maps we use the 5 by 5 minute resolution current global cropland and grassland maps developed by Klein Goldewijk et al. (2006) These maps were based on satellite data from the DISCover version 2 data using the IGBP classification map (Loveland et al., 2000) and the Global Land Cover (GLC) based on the Global Land Cover 2000 VEGA2000 data (Bartholome et al., 2002), combined with national land-use statistics (FAO, 2007) and sub-national land-use data for U.S.A. (USDA, 2006) and China (China National Bureau of Statistics, 2006; China National Bureau of Statistics, 2006). Allocation of land use CroplandThe method to allocate historical cropland is carried out for each grid cell of 5’ by 5’ grid cell (ca 85 km2 around the equator). For allocating historical cropland six major assumptions were made: (i) in urban builtup areas (Uarea) no allocation was allowed (no space left for agriculture); (ii) in areas with population density (Wpopd) lower than 0.1 cap km-2 no allocation was allowed (no need for agriculture); (iii) land with highest soil suitability for crops is colonized first (Wsuit); (iv) coastal areas and river plains are more favorable for early settlement as being easily accessible (Wriver); (v) steep terrain with high slopes are less attractive for settlement and agriculture (Wslope); (vi) below the threshold an annual temperature of 0 oC no agricultural activity is assumed (Wtemp_crop). These assumptions result in weighing maps that were normalized between 0 and 1 and multiplied to construct a final, unique weighing map for each time step. (Sub-)national crop area statistics are allocated to grid cells according to a mix of two weighing maps; a current one which was constructed from a satellite map of 2000 A.D. for cropland (Wcrop_satellite) (Klein Goldewijk et al., 2006), and a historical one, which is constructed according to the six rules as described in the former paragraph. The influence of the satellite map increases gradually from 10,000 B.C. to 2,000 A.D until it completely dominated the historical weighing map, i.e. until the cropland distribution equals the satellite map distribution (present situation). Cropland is allocated by combining historical cropland area statistics from HYDE with the various weighing maps: Current weighing map for allocation: Wcrop2000 = Wsatellite2000 (1) Historical weighing maps for allocation: Wcrop,t = Warea,t * Wpop,t * Wsuit * Wriver * Wslope * Wtemp_crop (2) where Warea,t = [Garea – Uarea,t] / Gareamax (3) Garea is the total land area (no ice and snow), Uareat the urban builtup area for year t, GAREAmax is the maximum area of a 5’ grid cell. See Supplementary Figure SF1 for the cropland allocation scheme. Note that only Warea and Wpopd is changing over time. Pasture The method to allocate historical pasture is carried out for each grid cell of 5’ by 5’ grid cell.For the allocation of pasture the following assumptions were made: (i) in urban areas (Uarea)and areas already occupied by cropland (Carea)no allocation was allowed; (ii) in areas with population density lower than 0.1 cap km-2 no allocation allowed (Wpopd); (iii) natural herbaceous areas as defined by the BIOME model (Prentice et al., 1992) are more attractive for use of livestock/pastoral activities then other land cover classes (Wbiome); (iv) below the of -10o C for the annual air temperature no year-round pastoral activity is assumed to happen (Wtemp_pasture). (Sub-)national pasture area statistics are allocated to grid cells according to a mix of two weighing maps; a current one which was constructed from a satellite map of 2000 A.D. for pasture (Klein Goldewijk et al., 2006), and a historical one, which is constructed according to the four rules as described in the former paragraph. The influence of the satellite map increases gradually from 10,000 B.C. to 2,000 A.D until it completely dominated the historical weighing map, i.e. until the pasture distribution equals the satellite map distribution (present situation) So, Wpasture,t = Warea,t * Wpopd,t * Wbiome * Wtemp_pasture (4) where Warea,t = [Garea,t – Uarea,t – Carea,t] /Gareamax (5) and Careat, is the area occupied by cropland on year t. See Supplementary Figure SF2 for the pasture allocation scheme. Summary tables 1 to 5 of historical estimates of areas of cropland ans pasture can be found here. References |
