Time series – spatial information ESPON workshop, 6 th May 2010 Oscar Gomez, EEA.

Time series – spatial information

ESPON workshop, 6th May 2010Oscar Gomez, EEA

My understanding of time series and GIS

• Traditional GIS does not consider time dimension miss dynamics of some phenomena

• Time dimension is important in:– Administrative boundaries– Land cover/land use– Population– Hydrology– Vegetation/crops– Wildlife

• On the EEA side, this is secured in land cover with land cover changes layers

Example: hydro systems• Our context:

ECRINS hydro database

• Drainage doesn’t change with our time scale (*)

• Continental water dynamics

(*) except in the case of canals

Spatial information and time• Our context: CORINE Land Cover• 1:100.000, EEA MSs + collaborating countries geographic

extents, with exceptions• Homogeneous across participating countries• 1990, 2000, 2006, with exceptions• 39 countries CLC2000-2006• LC snapshots + LC changes• Many datasets derived spatial disaggregation generally

depends on CLC + something else• PHARE 1975 – 1990 LC Bulgaria, Hungary, Romania,

Slovakia, Czech Republic• LACOAST 1975 – 1990 coastal LC: EU15 (except UK, LU)

Exchange of data

CORINE Land Cover• Co-ownership EEA/Communities and Member States,

data flow control• They are modified only to harmonize borders• Problem: time span in 2010 we have 2006 data• GLOBCorine (ESA): from GLOBCover (medium

resolution), trained with CORINE GlobCorine– Better geographic coverage– Nowcasting– Less classes– Less certainty about individual changes– 2006 done, 2009 before summer (ESA)

GlobCORINE

•Urban and associated areas•Rainfed cropland•Irrigated cropland•Forest•Heathland and sclerophyllous vegetation•Grassland•Sparsely vegetated area•Vegetated low-lying areas on regularly flooded soil•Bare areas•Complex cropland•Mosaic cropland / natural vegetation•Mosaic of natural (herbaceous, shrub, tree) vegetation•Water bodies•Permanent snow and ice•No data (burnt areas, clouds,…)

Land cover and HANTS(EVI 2006)Salting (Bosplaat, Terschelling)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Urban (Amsterdam)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Grassland (Friesland)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Deciduous forest (Harderbos, Flevopolder)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Drifting sand (Veluwe)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Pine forest (Veluwe)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Agriculture (Flevoland)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Grain cultivation (Dollard, Groningen)

0

0.2

0.4

0.6

0.8

1

1 7 13 19

16-days-NDVI-composites

ND

VI

original

HANTS fitted

Red; medium, Green; medium, Blue; high

Red; medium, Green; high, Blue; medium

Red; high, Green; low, Blue; low

Red; low, Green; low, Blue; lowRed; high, Green; high, Blue; low

Red; medium, Green; high, Blue; low

Red; high, Green; low, Blue; medium

Red; low, Green; medium, Blue; low

Red = average NDVIGreen = Annual AmplitudeBlue = Six months Amplitude

Water quantity

• Run-off data• No data flow established• Based on the good willingness of MSs• The data flow is being defined

Spatial dimension

Land cover – and derived

• Neighbourhood: yes, on analysis; for example, green landscape outside cities

• Spatial disaggregation: Not directly us (by now), mainly the JRC:– Population density– Agro-land use: AFOLU crops, livestock

• Spatial aggregation: used constantly; OLAP cubes

Livestock density EU27

Water drainage (ECRINS)

• Interpolation of climate data (rainfall, temperature) “spatial disaggregation”

• Spatial aggregation: to the catchment level

Time dimension

Time dimension

• Water: climate data is always spatially interpolated data; extrapolation: IPCC scenarios, forecast

• Land cover: no interpolation; extrapolation: land use modelling (JRC: MOLAND, LUMOCAP)

• Need for time-dimensioned GIS layers: i.e. transport networks, protected areas, ... start_date, end_date fields!

Trends in the coast: 1975 to 2006 (30 years of changes)

•Artificialisation has a constant growth rate: 0.5% relative increase each year

•Water bodies were created in 1975-2000

•Agriculture shows a constant decline•Wetlands and forest and semi-natural decreased heavily (around 10%) in 1975-1990; it has slowed down

Trends 1990 – 2000 – 2006 (*)

(*) 100% = status in 1990; the lines show the relative increase (trend) for the 2 periods, 1990-2000, 2000-2006

• Urbanisation: same trend, above 0.5% yearly increase• Forest and semi-natural are stable• Wetlands don’t disappear as quickly as in the previous period; strong trend change (from 0.22% yearly loss to 0.06% yearly loss)• Water bodies are created at a slower pace (0.19% yearly increase to 0.08%)