GIS and Remote Sensing Unit at FEM-CRI

MODIS NDVI/EVI analysis

From MODIS (flown on Terra and Aqua satellites), Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI) maps are available. They are aggregated to 16 days to minimize cloud contamination. We prefer to use EVI in our studies as it tends to perform better than NDVI. EVI is less prone to saturation as well as less sensitive to haze due to the inclusion of the blue channel (Huete et al., 2002). This is of interest in mountainous regions where valleys are often relatively hazy.

Ecological indicators from MODIS Vegetation indices

EVI permits to detect inter-annual seasonal vegetation differences (pixel-wise map subtraction), spring/autumn detection (EVI thresholding) and the calculation of the growing season length (period with EVI over certain value). In Trentino, we observe for example that even over short distances the seasons are slightly shifted due to effect of valley orientation and exposition.

(click to enlarge)

Landuse/Landcover and EVI

The detailed identification of zones relevant to a particular ecological problem can be extracted from the combined analysis of LULC (e.g., the European CORINE maps) and MODIS EVI. The screenshot shows CORINE polygon boundaries over MODIS EVI from June 2003 (Trento, Italy, with Molveno and Caldonazzo lakes in grey).

(click to enlarge)

Comparison of NDVI and EVI performance

Both NDVI and EVI maps are colored with identical color table (MODIS/Terra scene MOD13, composite of 21 March - 5 April 2000, Calabria, Southern Italy). EVI is less prone to atmospheric distortion. Both maps have been processed according to the list shown below.

Calabria, Southern Italy: NDVI and EVI (composite of 21 Mar-5 Apr 2000)

GIS processing of MODIS NDVI/EVI (MOD13/MYD13)

The following preprocessing steps are required to obtain usable NDVI/EVI maps from MOD13:

• reprojection from SIN to a commonly used projection (e.g., UTM)
• application of the quality maps (requires pixelwise bitpattern analysis)
• division by 10000.0 to get back the range from -1.0 .. +1.0
• elimination of pixel with value outside of that range
• application of color table

VI Details: QA maps

The filtering of MODIS VI maps with the QA map bitpatterns is a bit tricky. Here a list of relevant bit combinations:

Bit position	15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0	Notes
Integer	32768	16384	8192	4096	2048	1024	512	256	128	64	32	16	8	4	2	1
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	VI produced, Good Quality
1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	VI produced, but check other QA
2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	Pixel produced, but most probably cloudy
3	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	1	Pixel not produced due to other reasons than clouds
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	VI, Highest quality
4	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	VI, Good quality
8	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	VI, Acceptable quality
12	0	0	0	0	0	0	0	0	0	0	0	0	1	1	0	0	VI, Fair quality
16	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	VI, Intermediate quality
20	0	0	0	0	0	0	0	0	0	0	0	1	0	1	0	0	VI, Below intermediate quality
24	0	0	0	0	0	0	0	0	0	0	0	1	1	0	0	0	VI, Average quality
28	0	0	0	0	0	0	0	0	0	0	0	1	1	1	0	0	VI, Below average quality
32	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	VI, Questionable quality
36	0	0	0	0	0	0	0	0	0	0	1	0	0	1	0	0	VI, Above marginal quality
40	0	0	0	0	0	0	0	0	0	0	1	0	1	0	0	0	VI, Marginal quality
48	0	0	0	0	0	0	0	0	0	0	1	1	0	0	0	0	VI, Low quality
52	0	0	0	0	0	0	0	0	0	0	1	1	0	1	0	0	VI, Quality so low that it is not useful
56	0	0	0	0	0	0	0	0	0	0	1	1	1	0	0	0	L1B data faulty
60	0	0	0	0	0	0	0	0	0	0	1	1	1	1	0	0	Not useful for any other reason/not processed
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Climatology
64	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	Aerosol, Low
128	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	Aerosol, Average
192	0	0	0	0	0	0	0	0	1	1	0	0	0	0	0	0	Aerosol, High
256	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	Adjacent cloud detected, yes
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Adjacent cloud detected, no
512	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	Atmosphere BRDF correction performed, yes
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Atmosphere BRDF correction performed, no
1024	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	Mixed Clouds, yes
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Mixed Clouds, no
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Shallow ocean
2048	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	Land (Nothing else but land)
4096	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	Ocean coastlines and lake shorelines
6144	0	0	0	1	1	0	0	0	0	0	0	0	0	0	0	0	Shallow inland water
8192	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	Ephemeral water
10240	0	0	1	0	1	0	0	0	0	0	0	0	0	0	0	0	Deep inland water
12288	0	0	1	1	0	0	0	0	0	0	0	0	0	0	0	0	Moderate or continental ocean
14336	0	0	1	1	1	0	0	0	0	0	0	0	0	0	0	0	Deep ocean
16384	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Snow/ice, yes
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Snow/ice, no
32768	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Shadow, yes
0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Shadow, no

The color indications are recommendations for the QC pixel values (red: VI pixel to be rejected; orange: VI pixel probably acceptable; green: VI pixel ok). The decision to accept or reject orange indicated values depends on you. Find here a decimal to binary converter.

Reference

M. Neteler, 2008: Free GIS Software meets zoonotic diseases: From raw data to ecological indicators, Proc. FOSS4G 2008, 29 Sept-3 Oct 2008, Cape Town, South Africa [ Abstract ]

---