Use of FORMOSAT images over the Gourma site (Mali)

Size: px

Start display at page:

Download "Use of FORMOSAT images over the Gourma site (Mali)"

Adrian Lucas
5 years ago
Views:

1 Use of FORMOSAT images over the Gourma site (Mali) E. Mougin, V. Demarez, P. Hiernaux, L. Kergoat, V. Le Dantec, M. Grippa, Y. Auda, F. Timouk Photo: Doug Parker

2 Content The study site FORMOSAT data Field data Applications

The AMMA Gourma site A Sahelian climate Anomalies of annual rainfall, Hombori, 1920-2009 500 400 mean = 373 mm 300 200 100 0-100 -200-300 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975

3 The AMMA Gourma site A Sahelian climate Anomalies of annual rainfall, Hombori, mean = 373 mm Short rainy season : JJAS High rainfall variability High aerosol loading A pastoral Sahelian site. Characterized by large homogeneous surfaces. Dedicated to satellite product validation Soil moisture - albedo, Radiation, Ts - LAI, FAPAR, NPP

4 The Hombori site (15.3 N, 1.6 W) AERONET photometer Radiometer / PAR sensor

5 The Gourma land units LAI : Shrub Savannah LAI : 2-4 LAI : 2-4 Flooded plains Open forest

6 The Gourma land units LAI : < 0.1 Erosion surface Pond LAI : Tiger bush Millet field

7 Erosion surface Monitoring site Ponds Sand dunes Agoufou Open forest 60 km Tiger bush SPOT- HRV image

8 A strong seasonal dynamics and a high inter annual variability of vegetation cover June 06 May 29 June 08 July 17 July 15 July 16 August 19 August 20 August 23

9 Seasonal dynamics of acacia forests and millet fields July 05 July 28 August 26 July 02 August 27 September 16

10 Field data : LAI, FCover, FAPAR Use of hemispherical photographs

11 Field data : LAI, FCover, FAPAR (trees) Open Acacia forest : 500 m transect Isolated trees WinScanopy software

12 Field data : LAI, FCover, FAPAR (grass) Evaluation of Can-Eye derived LAI (destructive measurements) and FAPAR (SunScan) Derivation of LAI/FCover and LAI/FAPAR relationships at quadrat, ESU and km scale

13 Seasonal variation of LAI (grass) Monitoring of 8 sites every 10 days during the rainy season

14 FORMOSAT-2 data (2007) No SPOT data during the rainy season in 2007! (15 images in 2006) Period: June -November Number: 29 images View angle: 53 Size: 58 x 24 km AOT images Landsat Aerial photos SPOT

15 Dry season AOT image (2007_06_09) Bare sand dune fire Agoufou pond Agoufou Bare sand dune Clay-silt plain Rocky outcrops

16 Wet season AOT image 2007_08_04 Agoufou - Begining of herbaceous vegetation growth - High spatial variability

17 Wet season AOT image 2007_08_24 Maximum of herbaceous vegetation LAI

18 Example of atmospheric contamination 29 images 08_20 24 with clear sky among which 15 were acquired during the rainy season : 08_12 June : 4 July : 4 August : 5 September 6 DoY

19 What do you mean by aerosols?

20 Land Cover

21 Formosat view of an agro-pastoral area (wet season) Sand dune covered by a grass layer Bare loamy soils Acacia forest Millet fields Temporary ponds Road Field with high organic content Isolated Trees July 27

22 Comparison of HR images Landsat (30m) SPOT-5 (10m) Formosat (8m) September, 2007 FORMOSAT offer a better discrimination of cultivated areas

July 27 September 29 November 04 A large contrast between

23 Time series of Formosat images over an agro-pastoral area Burnt area Dry season Wet season Wet season Dry season June 1 July 27 September 29 November 04 A large contrast between fields and grasslands during the rainy season (phenological differences)

24 Land cover classification (work in progress) 6 identified classes: LF, LJ, LN SF, SJ, SN NDVI Green J: fallow F: manured fields N: non manured fields L: loamy sands S: Sand dune September 29 Red NIR Only multi-date images enable LC to be discriminated July 27 Hiernaux, Auda,

25 Vegetation monitoring and LAI mapping

26 NDVI seasonal variation over the Agoufou site June July August Sept. Period of water stress well detected on NDVI images as well as regrowth of green vegetation Effect of straw/litter on NDVI

27 NDVI LAI relationships : VALERI methodology NDVI 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 NDVI en fonction des LAI vrais r² = 0.83 n = LAI Site17_3107 Site17_1508 Site18_0108 Site18_1608 Site19_0108 Site21b_0308 Site31_3107 Site31_1608 Site21b_1608 Site19_1608 Site17_2708 Série12 Site41_1708 site41_0208 sol nu ESU scale : 10m x 10m Lacombe, 2008; Larouziere, 2009

NDVI LAI relationships (Field) 0.60 Photo H planimètre 0.50 0.40 y = 1.4446x + 0.0788 R 2 = 0.9207 NDVI 0.

28 NDVI LAI relationships (Field) 0.60 Photo H planimètre y = x R 2 = NDVI _08_ y = x R 2 = LAI Le Dantec, 2009

29 Spatialization of LAI August 4 August 8 August 20 Demarez, Mougin, in preparation

30 Spatialization of LAI / Comparison with MODIS product 2 approaches are compared

31 Comparison with MODIS LAI Products 2 Site 17 ( ) MODIS LAI LAI in situ LAI (m 2 m -2 ) DAY of Y EAR (Since 2005/01/01) 2.5 All MODIS data 2.5 Principal algorithm MODIS LAI R² = 0.82 R² = 0.94 MODIS LAI in situ LAI N = 29 N = in situ LAI Mougin, Demarez, in preparation

32 Monitoring of Sahelian ponds : Ex of the Agoufou pond (Gardelle et al., in revision)

33 Comparison FORMOSAT/MODIS/SPOT MODIS-2007 SPOT-2005/06 Evaluation of a MODIS based methodology for surface water mapping

34 Concluding remarks High spatial and temporal resolution of Formosat data are found useful for : - Land cover mapping (change) of cultivated surfaces (small size) - Vegetation monitoring : detection of period of water stress - Mapping and monitoring of small ponds However, over sahel, data acquisition are hampered by aerosol and cloud contamination. As a consequence, no data was acquired during 2 weeks within the core of the growing season

Preparing for the exploitation of Sentinel-2 data for agriculture monitoring. JACQUES Damien, DEFOURNY Pierre UCL-Geomatics Lab 2 octobre 2013

Preparing for the exploitation of Sentinel-2 data for agriculture monitoring JACQUES Damien, DEFOURNY Pierre UCL-Geomatics Lab 2 octobre 2013 Agriculture monitoring, why? - Growing speculation on food