Estimation of Livestock Densities in Europe

Sarah Mubareka, Carlo Lavalle

European Commission

Joint Research Centre

Institute for Environment and Sustainability

TP 267

Via Enrico Fermi, 2749

21027 Ispra Italy

+39 0332 78 6741

Datasets in package:

FAO raw maps: “ *\Livestock_densities\RAW_FAO_dataset_EU\(livestock category name)05_1km ” + manual

Corrected FAO maps: “ *\Livestock_densities\FINAL_mosaics_etrs\mos_(livestock category name)05_corr.tif ”

CoCo regional statistics: “ *\Livestock_densities\FAO_CoCo_Regional_statistics\ (countryFIPS)_(animalcategory).csv “ In this data series we also compare the CoCo data with the FAO data on a REGIONAL basis for 2 years: 2000 and 2005 (Table 1)

Table 1: Example of regional statistics provided with datasets

Region / Cattle00_CAPRI / Cattle00_FAO / Cattle05_CAPRI / Cattle05_FAO
DE11 / 386410 / 307924 / 337770 / 273818
DE12 / 133860 / 192502 / 138250 / 171204
DE13 / 282700 / 378367 / 249840 / 336950
DE14 / 449730 / 468778 / 371740 / 416856
DE21 / 965170 / 1179030 / 849930 / 1048990
DE22 / 517340 / 455843 / 461560 / 405553
DE23 / 435700 / 254743 / 399090 / 226520
DE24 / 274800 / 176763 / 254490 / 157183
DE25 / 346020 / 250081 / 315950 / 222382
DE26 / 205920 / 200183 / 193350 / 178010
DE27 / 690840 / 706626 / 629490 / 628559
DE40 / 793700 / 0 / 689930 / 0
DE71 / 164820 / 169774 / 151060 / 150970
DE72 / 161770 / 125054 / 147980 / 111203
DE73 / 264090 / 140319 / 232360 / 124777
DE80 / 749120 / 919326 / 646320 / 817562
DE91 / 198790 / 222871 / 213620 / 198186
DE92 / 338690 / 329080 / 316760 / 292631
DE93 / 848510 / 1104220 / 704650 / 981917
DE94 / 1144490 / 1456320 / 933960 / 1295250
DEA1 / 241490 / 310022 / 221680 / 276017
DEA2 / 256000 / 402972 / 236450 / 358574
DEA3 / 408790 / 543806 / 360810 / 484131
DEA4 / 259640 / 265364 / 238990 / 235972
DEA5 / 230900 / 295055 / 211230 / 262374
DEB1 / 189860 / 218477 / 173140 / 194279
DEB2 / 192420 / 204725 / 175620 / 182059
DEB3 / 122560 / 187225 / 117510 / 166524
DEC0 / 63460 / 87447 / 84350 / 77800
DED0 / 612490 / 394566 / 548520 / 350890
DEE0 / 559370 / 455665 / 480840 / 405196
DEF0 / 1123680 / 1267190 / 1015600 / 1126580
DEG0 / 470240 / 244084 / 432770 / 217049

Methodology

The map series was created using the Food and Agriculture Organization of the United Nations (FAO) livestock density maps[1] for 2005, actual livestock figures for 2005 as given by the Complete and Consistent database (CoCo) made available through the Common Agricultural Policy Regionalized Impact Modeling System[2] (CAPRI), and a series of water requirements per livestock type data from the literature.

Mapping livestock densities

The FAO livestock density maps are a 1 kilometer-resolution gridded product with global coverage, detailed in terms of livestock decomposition. In this study, we use the livestock density maps for 2005 estimates on sheep, goats, cattle, poultry and pigs.

The CoCo data is given at aggregates of regions we call “CAPRI-regions”. These regions correspond for the most part to NUTS-2 regions (Nomenclature of territorial units for statistics[3]at level 2) with few exceptions in Germany, the UK and Italy for example where the regions correspond to NUTS-1delimitations. For non EU countries only, the CAPRI-regions correspond to nationallevel boundaries. Most of the data within CoCo are derived from EUROSTAT (Farm and market balances, economic indicators, acreages, herd sizes and national input output coefficients). The datasets used in CoCo, and the derivation of secondary data and imputation techniques are described in detail in Britz and Witzke (2008), p.14-26.

The CoCo data is more refined in terms of livestock typologies, but was aggregated to match this FAO classification and the two datasets were compared.The different cattle activities in the CoCo database correspond well with the FAO summed heads of cattle for the same regions, there is little discrepancy between the two datasets. The CoCo dataset assumes newborn calves to be born on the 1st of January which enter immediately into the category of heifers and bulls for fattening a year after their birth (January 1st of successive year). Thus the animals are counted on a yearly basis. The correspondences between FAO and CoCo data in terms of sums of heads per CAPRI region were evaluated at two levels: for EU 27 and for EU27 + Norway, Kosovo, Serbia, Bosnia and Herzegovina, Montenegro, Croatia, Macedonia, Albania and Turkey in order to begin to understand the reliability of the CoCo data with respect to the FAO data. It was found that the correspondences were higher between FAO and CoCo at EU27 level for cattle and sheep than for all CAPRI-regions together. This was not true however for pigs or poultry. Given this mixed response, an analysis was made on country-level data. The degree of uncertainty for the FAO dataset, using CoCo as a referenceshowed which countries were compatible between the datasets. Figure 1 highlights the discrepancies between the two datasets, re-ordering the output to show the order of countries for which the datasets are most compatible to those which are least compatible. In some graphs, the countries with very high incompatibilities were removed in order to facilitate the readability of the figures.

a.

b.

c.

d.

Figure 1.Countries in order of uncertainty (0=no uncertainty) for FAO density map sums per CAPRI region for 2005, using CoCo data for 2005 as a reference.

(a)Cattle (NO, KO, MK, BA values exceed 50 and are not included in the graph to facilitate legibility)

(b)Pigs

(c)Poultry

(d)Sheep and Goats (TR values exceed 1795 and are not included in the graph to facilitate legibility)

For pigs, poultry, sheep and goats, the correction coefficient was applied at regional level. This means that the sums given by FAO are rescaled to match the sums given by CoCo. The rescaling is done equally for all cells within the region. The original livestock density maps from the FAO were corrected using the correction coefficient for each CAPRI region only if, based on the exercise described above, the correction coefficient did not surpass a threshold of ten. For those regions whose correction coefficient was above the cut-off point, the FAO data was plugged in without any correction.

For the cattle, the situation is more complicated given the large discrepancies in water requirements within this category. First, detailed density maps per cattle category were created based on the cattle layer from the FAO. These were computed using the CoCo data by redistributing the proportion of each of the cattle type evenly over each region. To do this, each CAPRI region cattle number is subdivided into its proportion of sub-categories using the CoCo data as a reference. Equation 1 shows the contributing sub-categories to the “Cattle” category.

CATA = ∑ DCOH,DCOL,SCOW,HEIR,HEIH,HEIL,BULH,BULL,CAMR,CAFR,CAMF,CAFF (eq. 1)

where

CATA=All cattle activities

DCOH=Dairy cows high yield

DCOL=Dairy cows low yield

SCOW=Other cows

HEIR=Heifers breeding

HEIH= Heifers fattening low weight

HEIL=Heifers fattening high weight

BULH=Male adult cattle high weight

BULL= Male adult cattle low weight

CAMR= male calves for raising

CAFR= female calves for raising

CAMF=male calves for fattening

CAFF= female calves for fattening

Each of the above classes consumes a different amount of water. The statistics available in the literature are not detailed enough to assign a water consumption value to each of the above classes so these are aggregated as follows, based on the available water uptake statistics (Equations 2-7):

HEI = ∑ HEIR,HEIH,HEIL (eq. 2)

BUL = ∑ BULH,BULL(eq. 3)

DCOL = DCOL (eq. 4)

DCOH = DCOH (eq. 5)

CAL = ∑ CAMR,CAFR,CAMF,CAFF(eq. 6)

SCOW = SCOW (eq. 7)

In total, a series of nine livestock density maps were created: Sheep and goats, poultry, pigs, heifers, bulls, calves, high-yield dairy cows, low-yield dairy cows and “other”cows. A summary of the methodology for weighting the correction coefficients for all livestock classes is shown in Figure 2.

Figure 2. Correction coefficients based on CoCo data for 2005 are applied to the FAO livestock density maps for 2005.

Figure 3 shows an example of a resulting map (for high yield dairy cows). Grey regions are those whose correction coefficient was not implemented because it surpassed the correction coefficient. The correction coefficient could only be applied to the regions for which there was data in CoCo. Following the correction, the maps were combined with the FAO dataset in a mosaic procedure in order to ensure the full coverage of the area.

Figure 3. An example of a FAO Livestock density map for 2005, corrected for the true number of heads according to the CoCo database for 2005 (high-yield dairy cows).

The maps were compared to the statistics found in the literature. For example, the dairy cows map was visually compared to the table given in the IPTS report of February 20 2009 and reproduced below. Table 2shows that the biggest population of dairy cows in 2008, according to EUROSTAT, is in Bretagne (France, FR52) and Lombardia (Italy, ITC4), followed by Mazowieckie (Poland, PL12).

Table 2. Top 16 regions for dairy cows in EU 27 (source: IPTS 2009).

Figure 3 shows the overlay of the regions cited in Table 1 with the sum of the high yield dairy cow population per region.

Figure 3. The sums of high yield dairy cows for the regions in the EU27, derived from the simulated livestock density maps. The red polygons are the regions which, according to the data given by Eurostat in 2008, are those with the highest number of dairy cows.

Software

ESRI ArcMap 10.0 software was used throughout the project for the visualization of the maps and for extracting the initial subset of the FAO dataset of interest to us. The reading of the CoCo data; derivation of the correction coefficient for the FAO livestock density maps; and the actual computation of the water requirements’ series using the air temperature time series per livestock type was done using IDL 7.1.2. The maps were resampled to a 5km grid using ArcMap 10.0 and converted to ASCII grids. PC RASTER was used in successive steps to prepare the maps for entry into the hydrological model LISFLOOD.

1

[1] consulted February 23 2012

[2] consulted February 23 2012

[3]