Desertification In Tunisia

[stephenhann]

Hi,

I have a project to map 'Current' Desertification in Tunisia. I need to produce a 1:800,000 map.

My Initial Idea is to measure desertification through an NDVI to classify desertification using LandSat 8 images as a data source.

Questions:

Is LandSat 8's spatial resolution suitable to map a 1:800,000 scale map?

Should I use one year of images i.e monthly 2014 images or a longer time frame such as 10 years of monthly images or be even more specific i.e daily? to get a mean NDVI score, or should I not use a mean score?

Would all the images I have to use need to be corrected for atmospheric variation even though I have used the same satellite?

 

Thanks in advance!

 

[sigologo]

Hi Stephenhann

 you should try with MODIS Sensor is many product like EVI, NDVI, LST  & NPP

 

Regards!!

[intertronic]

hi, with landsat (30m resolution) you can go up to 100'000 easily, you can get new imagery with landsat but you have to do a mosaic in order to have the complete territory of tunisia. 

http://landsat.usgs.gov/Landsat_Search_and_Download.php

http://earthexplorer.usgs.gov/

 

 

like sigologo said, if you need another scale (like you said 1:800'000) is maybe better to do with modis (terra/aqua), you have a resolution of 250m and you can get new imagery on daily basis.

https://earthdata.nasa.gov/labs/worldview/

 

Desertification with NDVI and landsat

http://www.isprs.org/proceedings/2005/isrse/html/papers/669.pdf

Edited December 7, 2015 by intertronic

[intertronic]

MODIS over Tunisia

 

https://earthdata.nasa.gov/labs/worldview/?p=geographic&l=MODIS_Aqua_CorrectedReflectance_TrueColor(hidden),MODIS_Terra_CorrectedReflectance_TrueColor,Reference_Labels,Reference_Features,Coastlines&t=2015-12-05&v=-3.796875,26.929687499999986,15.732421875,38.091796874999986

Edited December 5, 2015 by intertronic

[stephenhann]

Hi,

The pdf link would not work, but thanks for your advice it was very helpful.

one other question would be if I compare a mean NDVI score for 2010 compared to 2015 but need to define desertification in 4 categories would I use the boundaries for 2010 as a classification system for 2015?

[intertronic]

Hi,

The pdf link would not work, but thanks for your advice it was very helpful.

one other question would be if I compare a mean NDVI score for 2010 compared to 2015 but need to define desertification in 4 categories would I use the boundaries for 2010 as a classification system for 2015?

sorry, a little mistake in the link, edited in original post..

 

 

http://www.isprs.org/proceedings/2005/isrse/html/papers/669.pdf

Edited December 7, 2015 by intertronic

[mamadouba]

Before you start choosing sensor data and methodologies, you need a good understanding of your landscape.  Desertification needs to relate to the rate of change in the landscape.  If rate of change is extremely high, then comparing data over a few years may work.  If rate of change is low, and the environmental conditions of the landscape are extremely variable, then that must be accounted for in your analysis, otherwise you will result in erroneous rates of change.  My suggestion, take the entire time series record of 250-meter MODIS NDVI over your study area; you can use the 16-day composites for your work.  This will give you over 350 observations from February 2000 to present.  You can temporally smooth the dataset using a Savitzky-Golay filter or harmonic analysis (Fourier Transform), or a combination of both.  After the time series has been preprocessed, you can choose to run a mean or median deseasoning function if you wish, but it's not necessary.  The final step of the analysis is to run a pixel-wise trend statistic using the Mann-Kendall or Seasonal Mann-Kendall.  There are two ways to do this. (1) Extract only the seasons for each year that correspond with peak greenness and run the Mann-Kendall on the subset.  The median trend (Thiel-Sen trend) will give you the rate of change (negative slope is decreasing and positive slope is increasing).  The Mann-Kendall statistic provides a degree of significance expressed as Z-scores (-1.96 < >1.96 can be considered significant change). (2) The second approach is running a Seasonal Mann-Kendall on the entire dataset and extract the same statistics.  There are a number of software packages to facilitate this analsysi, including IDL, Matlab, R, etc.  However, Idrisi (now called Terrset) has all of these functionalities built into the Earth trends Modeler (ETM).  Just be aware that desertification is a time series issue that requires exploitation of the temporal domain. It is not a simple image difference between a few dates.   

[sigologo]

Dear Mamadouba!!!

Very complete his explanation!!! Almost completely agree...

just to help discussion for me NDVI is not good Index Vegetation,

EVI MODIS is better performance and repair problems of saturation and overestimate vegetation in Arid environment in my case living Chilean Desert....EVI is better....

rest of the analysis is good very good but could you comment a little more than Savitzky-Golay filter or harmonic analysis (Fourier Transform)....

 ...

Thanks!!

 

 

[oz1]

Mamadouba is dead right about the need to understand the landscape & the rate of change.  Also, you need to be aware of what NDVI actually measures.  The following points apply:

 

1. NDVI or a similar index (they all have pluses & minuses) will tell you something about how green vegetation changes.  It is not a good indicator of dry vegetation.Thus, after rain, you will get a strong NDVI signal but after green vegetation dries off, it becomes yellow, brown or grey & your NDVI signal drops off, even though you may have exactly the same amount of ground cover by plants.  Essentially, all that happens is that you are swapping green biomass for dry biomass.  This is not desertification. Also, a strong NDVI signal may indicate encroachment by woody shrubs at the expense of herbage.  In some parts of the world, this is a form of land degradation.

 

2. NDVI varies in response to rainfall.  If your rainfall varies seasonally, you will need to filter out the wet/dry season variation & to extract a longer term trend over time.  This will require a time series of data.  If your rainfall is variable over longer periods (i.e. subject to ENSO generated wet & dry periods), that will give you substantial NDVI variation over time.  Again, to detect true desertification (i.e. loss of the ability of a landscape to produce plant biomass from rainfall), as opposed to short term drought, you will need to filter out the effects of rainfall variability.  Hence, the need for a long time series of whatever index you plan to use.  The length of the time series needed will depend on the amount of rainfall variability in your environment.

 

3. Desertification is a term that means different things to different people.  I have seen it widely misused & frequently confused with changes in vegetation cover associated with long term changes in rainfall (i.e. confused with the effects of drought).  I think what you are after is a loss of landscape resilience, i.e. a capacity to bounce back from imposed stresses, be they natural or man-made.  When you develop your methods of analysing a time series of satellite data, you need to show that loss of landscape resilience can be proven.  Almost certainly, this will require a long time series of data which forces you to use Landsat or AVHRR.  There are exceptions but these are restricted to situations where you have rapid & extreme land degradation.

 

4.  There is loads of this stuff in the literature.  Australian researchers probably produced the best developed methods for separating land degradation from rainfall variability & drought using both time series & spatial patterns in remotely sensed data.

[mamadouba]

Thanks for the contributions sigologo and oz1. This has turned into a very informative thread. Choosing an optimal spectral index is important since no index is perfect. I tend to apply NDVI for most landscapes except tropical because saturation typically is not an issue for arid or poor yield regions. NDVI is also a simple calculation because it does not require extra multiplicative factors or coefficients such as the high dynamic range indices. Unless ground truthing field data have been collected, many of these factors are arbitrarily chosen through experience and/or heuristics (e.g. SAVI soil factor of 0.5).

In response to sigologo about Savitzky-Golay (SG) and FFT, Thse are built-in functions in IDL, but requires extra coding to loop through image slices and pixel-wise observations (i,j) because they are meant to process 1-D data. However, there are other non-commercial options using Python, SPIRITS, 52North, TimeSat, HANTS, etc. SG is simply a pice-wise polynomial regression that requires a user-defined window to calculate the regression and degree of polynomial. SG alone does a great job at smoothing data, but does not deal with outliers. FFT, which I use Harmonic Analysis of Time Series (HANTS), is an iterative FFT that calculates the underlying sine waves (seasonal signal) and uses this signal to rebuild the series to a user-defined range of good data. I use the MODIS QAQC flags to flag bad pixels and serve as a mask for HANTS. Excellent at modeling missing data or outliers unless cloud cover is too persistent (tropics).

HANTS is freely available from here, http://gdsc.nlr.nl/gdsc/en/tools/hants, and the developer, Allard de Wit provides good documentation on research and application.

I use HANTS and SG together. I first remove the noise in my Tim series using HANTS and then I apply minimal smoothing using SG.

[sigologo]

Thanks a Lot Mamadouba!!!!!

 

I appreciate your time and your words,very interesting your approach for Time series analysis,

 HANTS analysis good tools,  If you don't mind, may you please share to me. IDL code Savitzky-Golay (SG)

Best Regards!!!!

[mamadouba]

sigologo, even better, Allard de Wit has posted all of his IDL code to Github and it includes a Savitzky-Golay routine that models data within a user-defined range.  

 

https://github.com/ajwdewit/idl_adewit

 

You will find HANTS under the hants directory and Savitzky-Golay under the sagof directory.  These scripts were developed to process time series image stacks so no need for looping (I wish I had these a year ago).

 

Here's what he says about SavGol, "This is an implementation of the Savitsky-Golay filter for processing time-series of satellite data. It uses ENVI for tiling over the stack of satellite images. This implementation is very close to the original implementation by Chen et al (2004) but it has some drawbacks that it does not do iterative filtering like HANTS does (could be added easily though)."

[sigologo]

sigologo, even better, Allard de Wit has posted all of his IDL code to Github and it includes a Savitzky-Golay routine that models data within a user-defined range.  

 

https://github.com/ajwdewit/idl_adewit

 

You will find HANTS under the hants directory and Savitzky-Golay under the sagof directory.  These scripts were developed to process time series image stacks so no need for looping (I wish I had these a year ago).

 

Here's what he says about SavGol, "This is an implementation of the Savitsky-Golay filter for processing time-series of satellite data. It uses ENVI for tiling over the stack of satellite images. This implementation is very close to the original implementation by Chen et al (2004) but it has some drawbacks that it does not do iterative filtering like HANTS does (could be added easily though)."

Thanks a lot Mamadouba

Very good link......

 

Best Regards