1. Introduction MODIS NDVI

2. Content MODIS NDVI Download/Pre-Processing MODIS
Paddy Area Extraction Using MODIS NDVI Data

3. MODIS Spatial Resolution 250 m 500 m 1000 m Spectral Resolution
36 spectral bands – Hyper-spectral
Radiometric Resolution
32 bit
Temporal Resolution
Daily
MODIS stand for Moderate Resolution Imaging Spectroradiometer
MODIS instrument provides high radiometric sensitivity in 36 spectral bands
MODIS are viewing the entire Earth's surface every 1 to 2 days.

4. The Terra Satellite (EOS AM) on board in 1999 to present
MODIS was launched into space aboard the NASA-EOS platform for studies of climate, vegetation, pollution, global and regional change, Environmental and Disaster Monitoring,
The Terra Satellite (EOS AM) on board in 1999 to present
The Aqua Satellite (EOS PM) on board in 2002 to present
Orbit
Terra: Orbit passes N  S Local equatorial crossing time
is approximately a.m.
Aqua: Orbit passes S  N Local equatorial crossing time
is approximately 1.30 p.m.
* Differences in Terra and Aqua Orbits result in different viewing
and cloud cover conditions for a given location.
Terra (EOS-AM1)
MODIS is a key instrument aboard the Terra (EOS AM) and Aqua (EOS PM) satellites.
Terra's orbit around the Earth is timed so that it passes from north to south across the equator in the morning,
while Aqua passes south to north over the equator in the afternoon.
Aqua (EOS-PM1)
NASA MODIS Website:

5. Terra Satellite Path
Sou1rce MODIS Rapid Response - LANCE (Feb,23,2012)
Terra Satellite flight direction is shown by red arrow. The red dots show the overpass ti me of each 5-min granule. Similarly as the yellow dots.

6. Aqua Satellite Path Sou1rce MODIS Rapid Response - LANCE (Feb,23,2012)
Aqua Satellite flight direction is shown by red arrow. The red dots show the overpass ti me of each 5-min.

7. 1.Download MODIS Goto: https://earthdata.nasa.gov/
Data >>> Near Real Time Data
Rapid Response >>> MODIS Subset >>> USDA Foreign Agricultural Service (FAS)

8. Download MODIS (Continued)
Select FAS_Indochina
Select Date
Select Image [NDVI] Terra, 250 m.

9. Download MODIS (Continued)
Download Geotiff file
Right Click > Save file as..
../Data/[leave file name as default]

10. 2. Download MODIS - FTP Things to know before further download
Data Version
MODIS Product
MODIS Tile
Another way for Download MODIS
Source:

11. Data Version Version 1 (V001)
Original, at-launch version. MODIS/Terra Snow Cover products originally processed as Version 001 (V001) were deleted from the NSIDC archive on 06 June 2003, due to the availability of reprocessed data with the improved V003 algorithm.
Version 3 (V003)
Contains processing refinements accommodating algorithm, instrument, and calibration stabilization.
Version 5 (V005)
Monthly snow products are available in a Climate Modeling Grid (CMG).
Fractional snow cover was added to MOD10_L2 and MOD10A1.
Sea Ice by Ice Surface Temperature (IST) and combined sea ice fields were removed from all sea ice products.
Browse images are available for all products.
V005 products use HDF compression making the file size much smaller.
Source:

12. MODIS Vegetation Products
MOD13Q1 Terra Vegetation Indices Tile 250m 16 day
MYD13C2 Aqua Vegetation Indices CMG 5600m Monthly
MOD13C2 Terra Vegetation Indices CMG 5600m Monthly
MYD13C1 Aqua Vegetation Indices CMG 5600m 16 day
MOD13C1 Terra Vegetation Indices CMG 5600m 16 day
MYD13A3 Aqua Vegetation Indices Tile 1000m Monthly
MOD13A3 Terra Vegetation Indices Tile 1000m Monthly
MYD13Q1 Aqua Vegetation Indices Tile 250m 16 day
MYD13A2 Aqua Vegetation Indices Tile 1000m 16 day
MOD13A2 Terra Vegetation Indices Tile 1000m 16 day
MYD13A1 Aqua Vegetation Indices Tile 500m 16 day
MOD13A1 Terra Vegetation Indices Tile 500m 16 day
MOD13Q1 of Terra and MYD13Q1 Aqua that used in this analysis.
The resolution is 250m and satellite passed the same location every 16 days.
Products that used in this analysis MOD13Q1, MYD13Q1
Source:

13. MODIS Tile Numbers MODIS Tile Number of Study Area - h28v06
Tiles are 10 degrees by 10 degrees at the equator.
The tile coordinate system starts at (0,0) (horizontal tile number, vertical tile number)
in the upper left corner and proceeds right (horizontal) and downward (vertical).
The tile number of study area is (28,06)
MODIS Tile Number of Study Area - h28v06

14. 2. Download MODIS - FTP Go to ftp://ladsftp.nascom.nasa.gov
Select allData
Select 5 [Version 5]
Select Data Product (MYD13Q1)
Select Year (2007)
Select Day (days of year; 201)
Search Tile (H28V06)
Save as *.hdf
Another way for Download MODIS
Source:

15. Download MODIS Data
Here Day is represented in Julian Date format – It is day count since the beginning of the year which is also known as Day of The Year (201)
ftp://ladsftp.nascom.nasa.gov/allData/5/MYD13Q1/2014/201/
/allData/[Version]/[Product Code]/[Year]/[Day of the Year]/
MODIS Tile Number of Study Area - h28v06

16. MODIS filenames MOD13Q1.A2007225.h28v06.005.2007245083237.hdf
-Product Short Name
-Julian Date of Acquisition (A-YYYYDDD)
-Tile Identifier (horizontal 29 vertical 07)
-Collection Version
-Julian Date of Production (YYYYDDDHHMMSS)
-Data Format (HDF-Hierarchical Data Format)
MODIS filenames follow a naming convention which gives useful information regarding the specific product.
For example, the filename indicates such as:

17. Download MODIS Data Downloaded Data are in Folder
1_Pre_Processing_MODIS
Input
Downloaded Data are in Folder
(Data\1_Pre_Processing_MODIS\Input)

18. Paddy Area Extraction Using MODIS NDVI Data
The Normalized Difference Vegetation Index (NDVI) is a simple graphical indicator
that can be used to analyze remote sensing measurements,
and assess whether the target being observed contains live green vegetation or not.

19. Different Variations of Greenness (NDVI) in Different Land Cover Types
Urban
Forest
Low NDVI, and positive values in Urban areas approximately 0.2 to 0.4 that Corresponding to Feature of grass
In water areas the NDVI values as near -1
Paddy Field
Water

20. NDVI Variation in Paddy Land
NDVI data has distinguished patterns in paddy field which correspond to variation of greenness of paddy fields. More elaborately, comparably low NDVI value at the Start of the Season (SOS) and high NDVI value in middle of the season. Because of this unique pattern, paddy fields can be extracted.
In Paddy Fields, NDVI  0.8
In Paddy Fields, the values of NDVI is
Start of the Season
Peak NDVI

21. Paddy Land Extraction– Crop Calendar of Vietnam
Based on Crop Calendar, images in June selected as Low NDVI images. Which means near start of the Dry season
Based on Crop Calendar, images in August selected as High NDVI images. Which means near harvesting of Dry season

22. Paddy Land Extraction – Flow Chart
Stack 1 of NDVI Images in Low NDVI period
Stack 2 of NDVI Images in Max. NDVI period
Calculate Pixel-Wise Mean
Visualize as a False Color Composition Image
ISODATA Classification
Area Calculation
This is a flow chart for main process for Paddy land map
Stack1 as example image NDVI less than 0.4
Stack 2 as NDVI image higher than 0.4
It has error need to take mean to reduce error

23. Multispec Software Paddy Land Extraction
MultiSpec developed by Purdue University, West Lafayette, Indiana, is a multispectral image data analysis software.
Freely download
OS Requirements
Macintosh
Windows

24. Overview Paddy Land Extraction
Software – MultiSpec
Import Data
Stack Images
Calculate Pixel-wise Average of NDVI data sets
Visualizing Time Series Data as False Color Composite
ISODATA Classification and Merging output classes

25. MODIS in Study Area High NDVI Low NDVI Stack 2 High NDVI period
Yellow box represent for image in Start of the Season
Blue box represent for MODIS image in March as Maximum NDVI period.
Low NDVI
High NDVI
Stack 1 Low NDVI period
Stack 2 High NDVI period

26. Open Images as Stack – (Start of the Season)
1. Click File Menu and Open Image
3. Browse location (Data\2_Paddy_Land_Extraction\Input) and Select one Image File (Philippines_NDVI_ tif)
4. Click Open
2. Select Multispectral Type

27. Open Images as Stack – (Start of the Season)
5. Click OK
6. Click OK

28. Open Images as Stack – (Start of the Season)

29. Open Images as Stack – (Start of the Season)
7. Click File Menu and Open Image
9. Browse location (Data\2_Paddy_Land_Extraction\Input) and Select "Philippines_NDVI_ tif" "Philippines_NDVI_ tif" images to make a composite of 3 images.
10. Click Open
8. Select “Link selected file(s) to Active Image Window”

30. Overview Paddy Land Extraction
Software – MultiSpec
Import Data
Stack Images
Calculate Pixel-wise Average of NDVI data sets
Visualizing Time Series Data as False Color Composite
ISODATA Classification and Merging output classes

31. Calculating Pixel-Wise Average
11. Click Process Menu and Reformat
12. Click Change Image Format

32. Calculating Pixel-Wise Average
18. Click OK
15. Select Average
14. Tick “New Channel from Function”
13. Turn On Transform Data
16. Enter 1,2,3
17. Click OK

33. Calculating Pixel-Wise Average
19. Save as Mean_NDVI_2010_01.tif in Data\2_Paddy_Land_Extraction\Output Folder
20. Click Save

34. Follow Same Procedure - 3 Images (Max. NDVI period)
Follow Same Procedure for Images in 2007-August month (3 images).
Summery of steps as following
Open as Stack of 3 images
Calculate Pixel Wise Average using “Click Change Image Format” tool and Save as

35. Overview Paddy Land Extraction
Software – MultiSpec
Import Data
Stack Images
Calculate Pixel-wise Average of NDVI data sets
Visualizing Time Series Data as False Color Composite
ISODATA Classification and Merging output classes
False colour composite scheme allows vegetation to be detected readily in the image.
In this type of false colour composite images, vegetation appears in different shades of red
depending on the types and conditions of the vegetation, since it has a high reflectance in the NIR band

36. Flow Chart False Color Composition
Stack 1 in Start of the Season
Stack 2 in Max. NDVI period
Mean_NDVI_2010_01.tif
Mean_NDVI_2010_03.tif
Stack Image Mean_NDVI_Stack.tif
Now, there are 2 image that are mean of NDVI in Start of season as stack1
and mean of NDVI in Max period as stack2
Next process, we will Stack 2 images.

37. Open Above Result as False Color Composition
(Mean_NDVI_2010_01.tif, Mean_NDVI_2010_03.tif)
22. Browse location (Data\2_Paddy_Land_Extraction\Output) and Select both Images as once, (Mean_NDVI_2010_01.tif, Mean_NDVI_2010_03.tif)
21. Close all existing Images and Click File Menu and Open Image
23. Click Open

38. Open Above Result as False Color Composition
24. Select “3-Channel Color”
25. Enter 2 as Red, 1 as Green, 1 as Blue
26. Click Ok

39. Open Above Result as False Color Composition - Output
Here area in RED shows area that has high NDVI in 2nd image and low NDVI in 1st image which is correspond to paddy area
Next Target is to Separate that Area using simple unsupervised classification method

40. Save Image this Stack Image
27. Click Process Menu and Reformat
28. Click Change Image Format
29. Select Header as GeoTIFF
30. Click OK

41. Save Image this Stack Image
31. Save as Mean_NDVI_Stack.tif in Data\2_Paddy_Land_Extraction\Output Folder
32. Click Save

42. Overview Paddy Land Extraction
Software – MultiSpec
Import Data
Stack Images
Calculate Pixel-wise Average of NDVI data sets
Visualizing Time Series Data as False Color Composite
ISODATA Classification and Merging output classes
Next step is to Separate that Area using simple unsupervised classification method as ISODATA

43. Unsupervised classification ISODATA
Iterative Self Organizing Data Analysis Technique
More robust
User specifies
Iteration
Clusters
Merging and Splitting parameters
ISODATA stand for Iterative Self Organizing Data Analysis Technique

44. Unsupervised classification ISODATA
Assign each pixel to nearest cluster (mean spectral distance)
Re-calculate cluster means and standard deviations
If distance between two clusters < some threshold, merge them
If standard deviation in any one dimension > some threshold, split into two clusters
Delete clusters with small number of pixels
Re-assign pixels, re-calculate cluster statistics etc.
until changes of clusters < some fixed threshold

45. ISODATA Classification
34. Browse location (Data\2_Paddy_Land_Extraction\Output) and Select Stacked Images in Last Step (Mean_NDVI_Stack.tif)
33. Close all existing Images and Click File Menu and Open Image
35. Click Open

46. ISODATA Classification
36. Click Processor Menu and Cluster
37. Select ISODATA
38. Enter Number clusters (here 5)
39. Click OK
40. Select “Cluster Stats:” as Do Not Save
42. Click OK
41. Select “Cluster mask file:” as GeoTIFF

47. ISODATA Classification – Save Image
43. Save as Mean_NDVI_Stack_clMask.tif in Data\2_Paddy_Land_Extraction\Output Folder
44. Click Save

48. ISODATA Classification – Open Classified Image
With Visual Inspection, we can distinguish, Cluster 2 as Paddy Land.
Next step is to merge classed in order to classify to 2 classes (paddy and non-paddy)
Usually, Starting from high number of classes and merging together gives better results. Optimum Number of Classes should be estimated by trial and error here.

49. ISODATA Classification – Merging Classes (Create Paddy/Non-Paddy Output)
46. Drag and Drop Clusters in to Groups Cluster 1 and Cluster 2 as shown in Figure
45. Select “”Groups/Classes”

50. ISODATA Classification – Merging Classes (Create Paddy/Non-Paddy Output)
47. Select “”Groups”

51. ISODATA Classification – Merging Classes -Save Image
48. Click Process Menu and Reformat
49. Click Change Image Format
50. Select Header as GeoTIFF
51. Click OK

52. ISODATA Classification – Merging Classes -Save Image
52. Save as Mean_NDVI_Stack_clMask_Merge.tif in Data\2_Paddy_Land_Extraction\Output Folder
53. Click Save

53. Non-Paddy
Paddy

54. Limitation
1. Cloud – Cloud blocks the view of optical images making almost impossible to analyze data. Especially, this is prominent in Start of the Season which associated with Wet-Season (Especially in Rain fed paddy lands)
2. Thin Cloud – when Thin Cloud presents over highly vegetated area, NDVI values are reduced, leading to miss classification. (Even, associated Thin Cloud is not reflected well in standard MODIS vegetation quality data)
3. Flood Water – when flood water present in the ground NDVI value reduced producing wrong NDVI profile
4. Most NDVI time series variations are not a smooth curves due to Atmospheric effect, Flooding effects, Mix-pixel effect (small size paddy fields), etc. which leads to erroneous classifications. (Especially at the start of the season, NDVI variation is very vague and bumpy. Because of this, in above flow chart pixel-wise Mean calculates before thresh-holding)

55. Thank you

56. METADATA of MODIS NDVI Data
Appendix:
METADATA of MODIS NDVI Data

57. Donwload http://e4ftl01.cr.usgs.gov/MOLT/MOD13Q1.005/2010.01.01/
0. METADATA file for MODIS images is also available for download by corresponding name with .xml extension

58. Add Data
2. Browse location (Data\1_Pre_Processing_MODIS\Input) and Select Image, (MOD13Q1.A h29v hdf)
1. Add Raster Data
3. Select “[GDAL] Hierarchical Data . . “
4. Click Open

59. Select Layer
6. Click OK
5.Select “MODIS_Grid_16Day_250m_500m_VI:250m_16_days_NDVI”
This Widow shows all layers that available in selected HDF image including NDVI, EVI, VI Quality data, etc. (Total 11 layers). Here, we select only NDVI for the analysis

60. Open METADATA 7. Right Click on the Layer and Select Properties
8. Select “Metadata” tab and “Properties” section

61. Open METADATA – Part of Output METADATA File
Driver
GDAL provider
HDF4Image
HDF4 Dataset
Dataset Description
HDF4_EOS:EOS_GRID:"D:\Crop_Monitoring\RESTEC_Training\Phase_II\Material\Data\1_Pre_Processing_MODIS\Input\MOD13Q1.A h29v hdf":MODIS_Grid_16DAY_250m_500m_VI:250m 16 days NDVI
_FillValue=-3000
add_offset=0
add_offset_err=0
ALGORITHMPACKAGEACCEPTANCEDATE=102004
ALGORITHMPACKAGEMATURITYCODE=Normal
ALGORITHMPACKAGENAME=MOD_PR13A1
ALGORITHMPACKAGEVERSION=5
ASSOCIATEDINSTRUMENTSHORTNAME=MODIS
ASSOCIATEDPLATFORMSHORTNAME=Terra
ASSOCIATEDSENSORSHORTNAME=MODIS
AUTOMATICQUALITYFLAG=Passed