What we can see from space; and how to link it to data and statistics

Transcription

1 What we can see from space; and how to link it to data and statistics Mohammed Said 1, Shem Kifugo 1, Madelene Ostwald 2, Gert Nyberg 3, and Lance Robinson 1 1 International Livestock Research Institute, 2 Chalmers University of Technology, Swedish University of Agricultural Sciences

2 Remote sensing platform Satellite Airplane Ground survey Photos: TAWIRI, ILRI, NASA

3 N Data and scale of information Satellite water resource Nairobi City Location Map Ngong Rongai Nairobi National Park Kenya Nairobi District Kiserian Athi River Kitengela Lukenya Tanzania Kajiado District INDIAN OCEAN Legend Wildebeest Observed Airplane land cover Olooloitikoshi Isinya Konza Towns Major Rivers Fences Nairobi National Park N 0 20 Kilo me te rs Kajiado Ground survey wildebeest distribution Data: NASA, DRSRS, ILRI

4 1858: First aerial pictures by French photographer and balloonist Nada over Paris, France. 1959: Explorer 6 takes first satellite image of earth. 1972: Birth of Landsat and spaceborne MSI. 1972: Blue Marble from Apollo 17. History 2007: Worldview-1 has current lowest commercial 0.5m for a satellite.

5 Use of Satellite images in landscape mapping and monitoring Land-cover (land-cover change detection) Crop growth stage length of growing season, tree growth Normalized Vegetation Index, Fire detection Wildlife wildebeest in the Mara,.. Deforestation monitoring Soil Brightness Index (SBI)

6

7 Visible Wavelength Sensors Sensor Launch Date GSD [m] Country Orbview 2,3 1997, 2003 USA Kompsat Korea Ikonos USA Quickbird USA FORMOSAT Taiwan Kompsat Korea Worldview USA GeoEye-1 (Orbview-5) USA

8 Houston, TX 10/07 via Worldview-1

9 Houston, TX 10/07 via Worldview-1

10 Land cover change detection Masai Mara Tarangire Simanjiro Ecosystem

11 Land cover changes based on Landsat and Landsat TM Serneels, Said and Lambin 2001

12 Land cover analysis of the agricultural farms north limits of Serengeti- Mara Ecosystem Kilometers

13

14 Habitat use and land use changes Msoffe et al. 2011

15

16 Length of growing period

17 NDVI time series product AVHRR NDVI3g SPOT VEGETATION Data range July 1981 December 2011 April 1998 July 2012 (now) Time step 15 days (2 per month) 10 days (3 per month) Spatial resolution 8 km 1 km Data size 1.3 GB 38 GB Vrieling et al. 2013

18 Link NDVI crop development Credits: Curt Reynolds, USDA

19 a) b) Length of the growing period Length of Growing Period (days) N Kilometers Vrieling et al. 2013

20 Zoom on Kenya: average LGP long rains short rains large maps: SPOT VGT smaller maps: AVHRR Vrieling et al. 2013

21 T L U per km2 Vegetation index trends over that last 30 years 5 year y = x R 2 = Norm alis ed NDVI Said et al., in prep

22

23 West Pokot Satellite Images SPOT Landsat AVHRR Geoeye

24 High resolution images - SPOT

25 FALSE COLOR COMPOSITE VEGETATION 1973 Landsat Multispectral Scanner (MSS) 1-3

26 FALSE COLOR COMPOSITE VEGETATION 1994 Landsat Thematic Mapper (TM) 4-5 Chepareria Town

27 NDVI changes over time AVHRR data

28 GOOGLE EARTH IMAGES VISUALIZATION OF THE STUDY SITE Chepareria Town

29 SPOT 5 with ground pixel resolution of 5m by 5m was used to capture edges of Maasai bomas in Narok

30 GeoEye with 1m ground resolution is able to capture greater details of information far much better. Note potential used to count livestock

31

32 Temperature trend

33 Research opportunities

34 Conclusion from this Any RS assessment of change in vegetation in West Pokot will be novel! Due to the phenomenon we want to describe (enclosures, trees vs bushes, mixed vegetation etc. ) fine/high resolution (<15-30 m) RS would be favourable Due to high degree of bare soil, alternative indexes that account for soil fractions such as EVI should be tested in combination with conventional NDVI A RS time series over the area would most likely include products from multiple-sensors By including climate variables (account for variability and influence of rainfall, evapotranspiration and/or temperature) a more reliable picture of human induced change will be found (this could possibly be done on large scale with low resolution data such as AVHRR, LandSat and/or MODIS and anomalies in climate variable could than be used for high resolution assessments)

35 Way forward for the RS part of Triple L 1. Get a clear picture of veg change over time incl. natural fluctuations (climate variability) (do we have access to climate data? What type?) 1. Large scale (e.g. MODIS start 2001 or eq.) to get landscape dynamics 2. Small scale (aerial photos, ASTER or eq.) to get phenomenon of enclosures, vegetation types, land use classes etc. 2. Correlate these physical parameters at large and small scale with 1. Demographical information e.g. size, location, pastoralis vs. sedentary (data availability?) 2. Cattle information e.g. type, population, location (data availability?) 3. Or other incidents, phenomenon, processes that the Triple L researchers might find. A plan for these 4 distinct research questions/packages/articles could be the base for a remote sensing based project proposal?

36 Biophysical NRM Socio- Economic Policies, Institutions & Governance

37 Research Tools Biophysical characterization Socio-economic assessment Institutional/governance assessments Participatory Systems Analysis Data Gathering and Analysis M & E Synthesis Action Decision Support Participatory landscape modeling Scenario tools Planning Research on NRM Processes What approach to NRM is being used What in this NRM system/initiative is and is not working well

38 Thank You