Using R for time series analysis and spatial-temporal distribution - - PowerPoint PPT Presentation

▶
using r for time series analysis and spatial temporal
SMART_READER_LITE
LIVE PREVIEW

Using R for time series analysis and spatial-temporal distribution - - PowerPoint PPT Presentation

Feb 17, 2023 384 likes •687 views

Outline Problem R vs GIS Methods Further works Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product Jedrzej Bojanowski* C esar Carmona-Moreno European Commission - Joint Research

slide-1
SLIDE 1

Outline Problem R vs GIS Methods Further works

Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

Jedrzej Bojanowski* C´ esar Carmona-Moreno

European Commission - Joint Research Centre Institute for Environment and Sustainability Global Environment Monitoring Unit, Ispra, Italy * Institute of Geodesy and Cartography, Remote Sensing Department, Warsaw, Poland

The R User Conference 2008, August 12-14 Technische Universitt Dortmund, Germany

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-2
SLIDE 2

Outline Problem R vs GIS Methods Further works

Problem R vs GIS Methods Further works

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-3
SLIDE 3

Outline Problem R vs GIS Methods Further works

The need for global burnt area product

Fires: a significant component of global ecosystem Influence on climate, carbon cycle, pollution... Climate change?

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-4
SLIDE 4

Outline Problem R vs GIS Methods Further works

The need for global burnt area product

Fires: a significant component of global ecosystem Influence on climate, carbon cycle, pollution... Climate change?

PROBLEM

Lack of an exhaustive base of past fires activities!

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-5
SLIDE 5

Outline Problem R vs GIS Methods Further works

The need for global burnt area product

Fires: a significant component of global ecosystem Influence on climate, carbon cycle, pollution... Climate change?

PROBLEM

Lack of an exhaustive base of past fires activities!

TO DO Concatenation of two existing databases: GBS and L3JRC

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-6
SLIDE 6

Outline Problem R vs GIS Methods Further works

GBS and L3JRC

GBS L3JRC time range 1982–1999 2000–2007 input data NOAA/AVHRR SPOT VEGETATION temporal resolution 1 week 1 day spatial resolution approx 8 km approx 1 km advantages seasonality! area estimates!

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-7
SLIDE 7

Outline Problem R vs GIS Methods Further works

Why R and not GIS?

Wide functionality Import of all data formats Easy data manipulation Statistical and geostatistical analysis Graph plotting Map plotting Results into LaTeX code

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-8
SLIDE 8

Outline Problem R vs GIS Methods Further works

Why R and not GIS?

Wide functionality Import of all data formats Easy data manipulation Statistical and geostatistical analysis Graph plotting Map plotting Results into LaTeX code

AUTOMATION!

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-9
SLIDE 9

Outline Problem R vs GIS Methods Further works

Methods

◮ Data import ◮ Data manipulation ◮ Time series analysis ◮ Regression modeling ◮ Principal components analysis & 3D visualization ◮ Spatial temporal distribution visualization technique

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-10
SLIDE 10

Outline Problem R vs GIS Methods Further works

Data import and storage

rgdal rNetCDF data manipulation NetCDF Analyses GeoTiff Vizualization GeoTiff

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-11
SLIDE 11

Outline Problem R vs GIS Methods Further works

Generalization

GBS

8 km weekly

L3JRC

1 km daily

GENERALIZATION

0.5 degree monthly

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-12
SLIDE 12

Outline Problem R vs GIS Methods Further works

Time series

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-13
SLIDE 13

Outline Problem R vs GIS Methods Further works

Seasonality shift

Time series for Kazahstan

Time Burnt area in [ km2 ] 1985 1990 1995 2000 2005 10000 20000 30000 1 2 3 4 5 6 7 8 9 10 11 12

Area burnt in months in GBS

Months Burnt area in [ km2 ] 50000 150000 1 2 3 4 5 6 7 8 9 10 11 12

Area burnt in months in L3JRC

Months 50000 150000

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-14
SLIDE 14

Outline Problem R vs GIS Methods Further works

Area estimation

1000 2000 3000 4000 −90 −60 −30 30 60 90 Yearly mean burnt area [ km2 ] latitude 1982−1999 (GBS) 2000−2006 (L3JRC)

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-15
SLIDE 15

Outline Problem R vs GIS Methods Further works

Probability map

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-16
SLIDE 16

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-17
SLIDE 17

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-18
SLIDE 18

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-19
SLIDE 19

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-20
SLIDE 20

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-21
SLIDE 21

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-22
SLIDE 22

Outline Problem R vs GIS Methods Further works

Probability extension algorithm

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-23
SLIDE 23

Outline Problem R vs GIS Methods Further works

Probability map after extension

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-24
SLIDE 24

Outline Problem R vs GIS Methods Further works

Principal Components & 3D interactive visualization

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-25
SLIDE 25

Outline Problem R vs GIS Methods Further works

Principal Components & 3D interactive visualization

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-26
SLIDE 26

Outline Problem R vs GIS Methods Further works

Spatial-temporal distribution

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-27
SLIDE 27

Outline Problem R vs GIS Methods Further works

Conclusion

◮ Probability extension algorithm ◮ PCA with 3D interactive visualization ◮ Map of spatial-temporal distribution of global data

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-28
SLIDE 28

Outline Problem R vs GIS Methods Further works

Conclusion

◮ Probability extension algorithm ◮ PCA with 3D interactive visualization ◮ Map of spatial-temporal distribution of global data

rgdal data import RNetCDF data storage zoo time series analysis rgl 3D interactive plots spatial interpolation PET image rotation fields raster maps plotting

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product

slide-29
SLIDE 29

Outline Problem R vs GIS Methods Further works

THANK YOU FOR YOUR ATTENTION!

Jedrzej Bojanowski*, C´ esar Carmona-Moreno EC Joint Research Centre Using R for time series analysis and spatial-temporal distribution of global burnt surface multi-year product