Remote Sensing Applications - Science topic

Of course the answer depends a lot on which kind of analyses you need to do.

Anyway, as already mentioned by others, RTKLIB is free and open source software with a graphical user interface running on Windows. It's written in C in case you need to modify the source code. http://www.rtklib.com/

GpsTools was developed by the same author of RTKLIB, but it's written in MATLAB (mixed with compiled C programs). http://gpspp.sakura.ne.jp/gpstools/help/gpstools.htm

Another free and open source solution is goGPS (but only for single-frequency analyses): http://www.gogps-project.org/

The MATLAB version of goGPS is more complete, while its Java version implements only the main positioning engine. goGPS MATLAB comes with a graphical user interface, while goGPS Java has to be included in your project as a library.

goGPS is a work-in-progress, so it may require the user to tweak the code here and there.

As one of goGPS developers, I would be happy to get back code contributions / bug fixes / suggestions. Collaboration proposals on goGPS development are welcome.

Using a water index such as MNDWI (for images with SWIR band) or NDWI (for images without SWIR band) to highlight water features, and then separating wetland form other features with proper threshold values.

That really depends on the technique you use.

Full wave form lidar is very expensive, but provides a very good estimate of above ground woody biomass (irrespective of whether the tree is alive, dead or in variable condition).

SAR also provides good estimates, but needs a thorough understanding of backscatter characteristics of both the trees and the surrounding landscape, but is likely to be problematic in urban areas due to the array of different scattering elements.

Traditional airborne or spaceborne high resolution optical will allow you to identify/count the number of trees and you'll need to have some sort of allometric relationship between crown size and wood volume to estimate carbon stock. The more stratified and locally validated your allometric relationship...the more reliable the result.

Read Help doc in Matlab about: "HDF Import Tool", maybe useful.

MODIS is not the best instrument to derive a DEM

I agree with Akhtar in using ArcGIS. When loading the polygons and the SPOT images, check their projection, coordinate system, and datum. Then use the projection tools to reproject the polygons in the same system of the raster images (not the opposite). About the second question on comparing aerial photos among different years, you should first classify them, and then perform a change detection analysis among the different years. Make sure that the aerial photos are in the same projection, datum, etc., and then choose an appropriate classification method which will depend on case by case.

You can extract depth information from satellite images (or orthophotos) of clear shallow waters. You can use just optical bands, as Leo wrote above. It is possible to use even Google images. Just, you need to have several points with known depth for calibration.

This possibility has been examined by many authors:

Lyzenga, D.R., 1978. Passive remote sensing techniques for mapping water depth and bottom features. Applied Optics 17 (3), 379-383.

Clark, R.K., Fay, T., Walker, C., 1987. Bathymetry calculations with Landsat 4 TM imagery under a generalized ratio assumption. Applied Optics 26 (19), 4036-4038.

Benny, A.H., Dawson, G., 1983. Satellite imagery as aid to bathymetric charting in the Red Sea. The Cartographic Journal 20 (1), 5-16.

Stove, G., 1985. Use of high resolution satellite imagery in optical and infrared wavebands as an aid to hydrographic and coastal engineering. Proceedings Conference on Electronics in Soil and Gas, London, January 1985 (Twickenham: Cahners Exhibitors), pp. 509-530.

Stumpf, R.P., Holderied K., Sinclair M., 2003. Determination of water depth with highresolution satellite imagery over variable bottom types. Limnology and Oceanography 48 (1, part 2), 547-556.

A review of passive remote sensing of shallow-water bathymetry Yarbrough, L.D., Easson, G., 2003. Comparison of techniques for deriving bathymetry from

remotely sensed data. AMRS – Conference 2003: Hyperspectral Issues for Coastal Zone Environments.

Baban, S.M.J., 1993. Evaluation of different algorithms for bathymetric charting of lakes using Landsat imagery. International Journal of Remote Sensing 14 (12), 2263-2273.

Bramante, J.F.; Raju, D.K., Min T.S., 2010. Derivation of bathymetry from multispectral imagery in the highly turbid waters of Singapore’s south islands: A comparative study. DigitalGlobe 8 – Band Research Challenge 2010.

There isn't a 250 m thermal product in MODIS; however, there are techniques where you can fuse both MODIS and Landsat products to get better spatial and temporal resolution.

See this article.

dear Victor,

as you may know, the problem of oil spill detection on water is still partly unsolved, especially for what concerns deep oil leaks and thickness detection,

you may be interested to look at this review to have a general listing of all the methods and their literature references, even though the paper is not really recent:

Some recent advances on the topic have been proposed based on hyperspectral methods (see e.g. http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6080935&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6080935) and interference approaches (see e.g. http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-20-22-24496)

But the literature on the argument is very large so I am far from complete in my answer!

Very good question !!!! : )

GRASS GIS r.in.xyz is VERY fast. It can create a DEM from millions of LiDAR points in a few seconds. And of course it is free and open-source

I agree with the previous answers. The success depend on many issues including the nature of vegetation that you want to map i.e. the extent, the mixing between different type of vegetation and also on what kind of properties that you will map from these vegetation. If the vegetation

Based on its spectral resolution, hyperion is a good sensor. If the area under investigation is large enough, and the cloud cover is low, it should be able to provide many valuable information about wetland vegetation, especially when its coupled with field data or spectral library.

when digitizing in google earth, the data should be saved as *.kml file.

This file should be imported to arcgis through import utility.

Multispectral instruments such as Landsat and ASTER have been proven in Belize, where Cherrington et al. aimed to measure at change in mangrove extent over multiple decades (see link to technical report from 2010). Spatial resolutions here ranged from 30m (Landsat's multispectral bands ranging from blue to near-infrared) to 15m (ASTER's green, red, and near-infrared channels). In brief, spectral mixture analysis allowed for the identification of mangroves. It should be noted that a detailed baseline mangrove map that included extensive field reports provided a reference for validation.

Globally, Giri et al. (2011) mapped mangroves using Landsat as well (30 m spatial resolution).

To identify dominant species (or to distinguish one species from the next), greater spectral resolution would be needed. Unfortunately, while the MODIS and MERIS instruments have better spectral resolution than instruments like Landsat and ASTER, their spatial resolution is lacking (~250m - 1km) for the fine level of detail that mangrove mapping requires. Hyperion is a hyperspectral instrument on board the Earth Observing-1 satellite and provides 30m spatial resolution, but the signal to noise ratio can limit its utility.

I agree with Emmanuel that the relationship to CDOM is probably a more promising approach to take. Our Aquatic Remote Sensing Group at CSIRO recently published work on the relationship of CDOM to salinity for river plumes impacting the Great Barrier Reef. Check out:

Schroeder, T., Devlin, M., Brando, V., Dekker, A., Brodie, J., Clementson, L., and McKinna, L. (2012). Inter-annual variability of wet season freshwater plume extent into the Great Barrier Reef lagoon based on satellite coastal ocean colour observations. Marine Pollution Bulletin, 65: 210-223.

Abstract:

Riverine freshwater plumes are the major transport mechanism for nutrients, sediments and pollutants into the Great Barrier Reef (GBR) lagoon and connect the land with the receiving coastal and marine waters. Knowledge of the variability of the freshwater extent into the GBR lagoon is relevant for marine park management to develop strategies for improving ecosystem health and risk assessments. In this study, freshwater extent has been estimated for the entire GBR lagoon area from daily satellite observa- tions of the Moderate Resolution Imaging Spectroradiometer (MODIS) between 2002 and 2010. To enable a reliable mapping of freshwater plumes we applied a physics-based coastal ocean colour algorithm, that simultaneously retrieves chlorophyll-a, non-algal particulate matter and coloured dissolved organic mat- ter (CDOM), from which we used CDOM as a surrogate for salinity (S) for mapping the freshwater extent.

Dear Victor,

Infrared and hyperspectral data will be good for detecting freshwater. However, geophyical mapping may help some times for fresh water hunting.

Hope this will be helpful for you

here's a link to the U.S. Geological Survey Spectral Libraries...http://speclab.cr.usgs.gov/spectral-lib.html

Indeed, some good correspondence between satellite retrievals and in-situ observations can be obtained. See:

Albergel et al. 2009: An evaluation of ASCAT surface soil moisture products with in-situ observations in Southwestern France. Hydrol. Earth Syst. Sci., 13, 115–124.

(http://www.bartalis.com/publications/albergel_etal_2009.pdf)

Prigent, C., F. Aires, W. B. Rossow, and A. Robock (2005), Sensitivity of satellite microwave and infrared observations to soil moisture at a global scale: Relationship of satellite observations to in situ soil moisture measurements, J. Geophys. Res., 110, D07110, doi:10.1029/2004JD005087.

I second Richard Gloaguen's comment. Additionally, using freely available Landsat ETM+ and Landat 7 imagery is a good place to start. There's a small body of literature available from authors around the world on the band ratios that worked in their study areas for epi-and mesothermal gold deposits.

In unvegetated areas, the bands and band ratios that highlight alteration products found on the surface, mainly clays, were reasonably good indicators and predictors of subsurface and nearsurface veins.

Check out the following paper "Primary analysis for enhancing the Iron Oxide and alteration minerals, using ETM+ data: a Case study of Kuh-e-Zar gold deposit, NE Iran" S Saadat, M Ghoorchi - IRANIAN JOURNAL OF EARTH SCIENCES (IJES), 2009 (http://www.sid.ir/En/VEWSSID/J_pdf/1019920090106.pdf) and others like it.

The turbidity, organic and inorganic matter content of the water, and the nature of the atmospheric layer between the sensors and the water body (often influenced by time of the year) and the nature of the surface at the boundary between water and land could introduce challenges. These variables would affect the albedo of the water body and make it more difficult to distinguish. There have been a number of studies utilising bio-optical and radiation transport models to enhance mapping accuracy. 

There are quite a lot of methods that will apply on different settings with more or less success... We are currently working on several approaches so if you are interested contact me. But basically, we use preprocessed DEMs and PAN-data to extract linear features that we further merge using polynomial fits and or splines. It works quite well.

But, to be honest, one of the most important step is still to pre process the data in order to be able to extract the right stuff... There is a huge gap between image discontinuities and geological structures...

We do better than that now but look at: Mallast, U., Gloaguen, R., Geyer, S., Rödiger, T., & Siebert, C. (2011). Derivation of groundwater flow-paths based on semi-automatic extraction of lineaments from remote sensing data. Hydrology and Earth System Sciences, 15(8), 2665–2678. doi:10.5194/hess-15-2665-2011

It might give you some ideas...

OMI's Level 2 products are stored as two-dimensional arrays in the instrument's frame of reference. The first dimension of these arrays has 60 entries in across-track direction, which are obtained at the same time (push broom scanner). The second dimension has about 1600 entries in along-track direction. In along-track direction, entries are sorted by time from the first to the last measurement in an orbit. As there are about 14 orbits per day, you will find 14 files for each day.

The Level 2G product are also stored in a two-dimensional array, but with first and second dimension being longitude and latitude, respectively. Each entry of these arrays has a list of all measurements obtained at this longitude-latitude position. The Level 2G product contains all data of a single day in one file.

To summarize, the Level 2 product is using measurement time as array dimensions and the Level 2G product is using longitude-latitude ground position as array dimensions. Thus Level 2 data are ordered by time, while Level 2G is ordered by ground position.

I hope my explanations help to answer your questions.

Yes, you can. Most of the GIS/RS software should do it. If you have access to Arc GIS you can use Arc Map and categorize the data (classification). Thanks.

Thank you for the answer.

Thx for the adress, but it seems this server isn't online anymore...

Hi Martin,

I can mention two large networks of ground measurements.

1) The VALERI Initiative

The objectives of the VALERI project are to provide high spatial resolution maps of biophysical variables (LAI, fAPAR, fCover) estimated from ground measurements to validate products derived from satellite observations.

For this purpose, the VALERI project offers:

• a methodological framework designed for the derivation of the high spatial resolution maps;

• a pool of instrumentation and tools for ground measurements and processing;

• a network of sites distributed over the Earth’s surface;

• a database of processed and available high spatial resolution maps.

Have a look at: http://w3.avignon.inra.fr/valeri/

2) FLUXNET

In the CO2 Eddy covariance FLUXNET network (global) several sites are equipped with optical instruments to measure in situ NDVI. For more details and a description of the approach see the publication here below:

Cheers,

Frank

As the method of first choice I would suggest a linear interpolation.

In your example: in the common area 28S-> 32S value of each pixel set as a weighted average of the two images with respect to latitude, ie:

X1 - the value of an image 1

X2 - the value of image 2

X - the value of the merged image

fi - latitude

If fi <= 28

X (fi) = X1 (fi)

If fi> = 32

X (fi) = X2 (fi)

If fi> 28 and fi <32

X (fi) = X1 * (32-fi) / (32-28) + X2 * (fi-28) / (32-28)

If you are using NetCDF, you can do the calculations for example in ArcMap (version> 9) or in MATLAB

Is it possible to get information on estimation of the biomass of mediterranean type bushlands (maquis, chapparals etc) using remote sensing and/or GIS techniques?

I have asked a question, but stiil waiting for a reply:

Have you read up at all on BDRF? I think there are a number of formulae for correcting for the angle of your incoming solar radiation.

Thanks Abdelaziz, may I have your email address if possible?

I think this link contains everything you need for Matlab: http://homepage.uibk.ac.at/~c7071028/metpack.zip

Anyway, there is a special package dedicated for R programming language, which is called Radiosonde (http://www.image.ucar.edu/Software/RadioSonde/) with nice skewT examples. R programming language is in some aspects very similar to Matlab syntax, so maybe using part of this code may be really helpful also in your case.

You can contact Jeff Masek at NASA to obtain the STARFM software.

Based in the information from USGS (http://landsat.usgs.gov/ESUN.php) you should not use ESUN but the information provided in the metadata.

In the GRASS manual is a method described how to do it.

Hope this helps?

Dear Victor,

The relationship depends much on the change in land use and the change in activities and management of that land. The run-off quality is highly influenced by the retention time and contact with polluted substances. In general land cover change from nature to roads or cities results in lower quality of the run-off water. Roads will add oil and others, buildings can increase the concentration of e.g. lead and copper in the run-off water (from e.g. drain pipes). Key is, what kind of change is it, what was the prior situation and what is the difference and what is the surface area of that change.

In my upcoming project I will more or less take this relation into account, regarding soil structure improving measures and their effect on the watersystem. Effects are all ready known on a rough scale, but the affiliated costs and benefits aren't. I will focus on the latter.

Hopefully this helps you. If you have more questions don't hesitate to ask.

Greetings,

Jeroen

The issue is rather complex and depends a lot on the forest characteristics. While in general dense forests can be easily detected the problem becomes critical in open forests such as savannahs, degraded forests etc. Here, in the optical domain the background reflected energy often masks the signal from trees resulting in classification errors. This raises as well the definition of the class "forest", as for instance in tropical countries agro-forestry (e.g. rubber, cocoa, orchards) are more considered as agriculture, while mainly comprise trees. What can help are time series (esp. when seasonal changes are present) which will allow to differentiate for instance agriculture practice from permanent forestry.

Nevertheless, open forests remain a challenge in the optical domain.

VIIRS is a new and very wide dataset, you can find it here

Also EUMETSAT products are available here

I suggest looking at this very interesting paper:

"A three-dimensional gap filling method for large geophysical datasets: Application to global satellite soil moisture observations", freely available here:

Sure. I will post it here as soon as it comes out.

Before doing anything you should probably take into consideration some of the calibration issues currently being dealt with for Band 11 (http://landsat.usgs.gov/calibration_notices.php)

There is a correction factor that can be applied to Band 10 but it is suggested you currently steer clear of Band 11 and the split window correction process until further investigation is done on the calibration.

Data assimilation is a natural approach for merging observations (in-situ and satellite) and model data to obtain estimates of either observed or unobserved variables. For instance, we are interested in obtaining estimates of the North Atlantic meridional overturning circulation (NAMOC) by applying that methodology. Preliminary results suggest that this could be a successful endeavor.

Hi Solomon, it's an interesting question, one I'll respond to with a different question :) What sort of land cover change are you trying to detect? land use intensification, tree clearing, urban expansion, shifts in grazing pressure, vegetation responses to climate dynamics? All of the above?

Depending on what sort of change you're trying to detect, and your existing skills there are a range of options:

Easiest: Identify the best available GIS data from the three decades, research the methods and classification schemas used to generate the GIS layers and then create look up tables to cross walk between classifications. Once you've standardised the three classifications then a standard GIS change analysis will give you a first order estimate of the changes that have taken place. Advantages, relatively quick and straightforward and requires intermediate GIS skills and no remote sensing expertise. The major limitation is the availablity and comparability and thematic/temporal/spatial limitations of the coverages of available GIS layers.

Moderate: Source surface reflectance corrected Landsat data from the three decades in question, 1 cloud free scene per decade, preferably acquired at a time of year when the dominant land cover classes are easy to discriminate (ie dry season rather than wet season), perform the same image classification technique to all data and then compare results through time. The national land cover dataset for the US is a good example of this approach http://www.mrlc.gov/nlcd2001.php

Advantages, the stable nature of Landsat allows you to hindcast the classification provided you have decent field data for current land cover classes (worst case you can do this from Google Earth with it's date specific imagery and some image interpretation). Disadvantages, you need a reasonable level of remote sensing and image processing techniques to undertake such an approach and need access to earth observation image processing software i.e. ENVI, Erdas, ERMapper, GRASS to run the classification. Another disadvantage is that you have no insight into when the change occurred between image dates, and you may fail to detect short live land cover change events/signals. This is especially problematic in tropical environments where the shift from forest to oil palm (both have strong green canopy signal once established) can occur in a short space of time.

Challenging: Use a full multi-temporal approach using all available Landsat data such as that described in http://www.sciencedirect.com/science/article/pii/S0034425713001235 or http://www.sciencedirect.com/science/article/pii/S0034425712000387 or http://www.tandfonline.com/doi/abs/10.1080/01431161.2013.796099

These approaches are at the cutting edge of land cover mapping but require a strong understanding of both the time series analysis techniques and land cover change types that you're trying to detect, and access to high performance computing so that you can bring change detection algorithms to bear on terrabytes of data.

Good luck!

I wouldn't suggest to apply any direct spatial interpolation method before checking whether there is any spatial pattern or not. I would recommend to perform trend surface analysis and Mantel test before you proceed. If there is any spatial pattern, it's also important to identify what actually describes such spatial variability. There is a recent good paper on it if you are interested:

Peres-Neto, P. R.; Legendre, P., Estimating and controlling for spatial structure in the study of ecological communities. Global Ecol. Biogeogr. 2010, 19, (2), 174-184.

Hope that helps.

Thanks for this information. I'm currently working in Vietnam in the frame of different project dealing with ocean color, my main research topic. I'm based at Spatial Technological Institut in Hanoi and promote ocean color in Vietnam in collaboration with different researchers and institution (ESA, CNES, NASDA, IRD, USTH).

Hope to hear from you,

best

hubert loisel

It is very useful, especially ASTER satellite. There is a model called "SEBAL, Surface Energy Balance Algorithm for Land" deals with estimating the evapotranspiration. Also, it can be correlated with the vegetation cover using vegetation indices (VI), such as NDVI and SAVI.

Please find below some useful links:

1) http://phylares.vub.ac.be/Thesissen/2007%20Ahmed%20Qudaih.pdf

2) http://www.dca.ufcg.edu.br/DCA_download/ISR/UFPE/Final%20Sebal%20Manual.pdf

Please don't hesitate if you want to contact me for more information about this issue.

Here is a recent publication that should help answer your question:

Di Michele et al, 2013: Quality Assessment of Cloud-Top Height Estimates From Satellite IR Radiances Using the CALIPSO Lidar. IEEE Transactions on Geoscience and Remote Sensing, Volume:51 , Issue: 4, Pages 2454 - 2464.

maybe this pub will help some. part of the document discusses coastal flood inundation mapping with SAR and touches on influences of HH and VV polarizations.

Ramsey, E., III, D. Werle, Y. Suzuoki, A. Rangoonwala, and Z. Lu, 2012. Limitations and Potential of Optical and Radar Satellite Imagery to Monitor Environmental Response to Coastal Emergencies in Louisiana, USA. Journal of Coastal Research, 28(2), 457–476.

colour can be assessed from satellite images, especially due to suspended sediments and algal blooms

p.s. based on these there is possible to make some rough correlations, but if You want to set-up some kind of water-quality classes than You will have to combine the spectral analyses with field-data (physico-chemical and biological parameters from the same periods when the respective images were taken)

Dear Ali,

It is a very good question, specially for those who are working in the field...

The following paper might help: http://www.environment.uwaterloo.ca/research/rsgtl/publications/2004/2004_Tulloch_Li_EIA.pdf

Moreover, you can read my paper entitled " APPLICATION OF DUBAISAT-1 IMAGERY" which shows some of DubaiSat-1 Environmental Applications.

Best Regards,

Saeed

As your objective is just to seperate the vegetation from others, I suggest you calculate the NDVI, then conduct segmentation and classification by setting a threshold.

You can get Landsat metadata for this site

I will strongly disagree with the predecessors. Satellite images will never directly allow you to measure forest degradations. Sadly, RS is a bit cursed because it deals with data that are easily represented as images, fancy pictures with nice colours. One can easily forget that it is a bit more complicated than that.

As as said in a previous comment, one has to be extremely cautious with change detection based on RS.

To summarise: find a proxy for your type of "degradation", find the proper data/technique and then, maybe, you will find that RS is adequate.

On the other hand, there is no real debate in the RS community regarding the potential of RS for any kind of land degradation monitoring. RS has, without any doubt, a huge potential. What people do though, is trying to find suitable methodologies for all the different ecological/climatic/... settings at different time and spatial scales, with limited data availability.

I can't tell you how to measure forest degradation because you have first to define "degradation".

There is no specific answer to your questions. In our last research on that topic we found out that the MLR (multiple linear regression) of textures and PolInSAR decompositions and coherence gives the best results. But I guess we are only beginning to understand the matter ;-)

Anyway, I can tell you that backscatter coefficients only will not do at all, in any polarisations.

We are currently publishing a few papers in that directions. Stay tuned.

But here some current thoughts:

Liesenberg, V., & Gloaguen, R. (2013). Evaluating SAR polarization modes at L-band for forest classification purposes in Eastern Amazon, Brazil. International Journal of Applied Earth Observation and Geoinformation, 21, 122–135. doi:10.1016/j.jag.2012.08.016

Wijaya, A., Reddy Marpu, P., & Gloaguen, R. (2010). Discrimination of peatlands in tropical swamp forests using dual-polarimetric SAR and Landsat ETM data. International Journal of Image and Data Fusion, 1(3), 257–270. doi:10.1080/19479832.2010.495323

It is on process now, no published yet. We have a research on Moscow case study, but only in Russian. Here http://www.geosys.ru/images/site/journal/3-2011/vihod_3_2011.pdf - you can find an abstract of Moscow UHI study in English/

Here: Experience of satellite images application for ecological research of Moscow area. I. Labutina, E. Baldina, M. Grischenko, T. Khaibrakhmanov; -http://zikj.ru/index.php/en/archive/issue12 - the study of Moscow with our Moscow UHI as a fragment.

its better go for some other data than Landsat since it is very coarse to map mangroves. Mangroves cover along coast line there will be a chance of miscalculation of area with coarse resolution. u will be face mixed pixel problem

I would start by looking at what others have done toward creating integrated MW precipitation products, as differences in performance/calibration from different sensors must be dealt with in this context. For instance, see what was done at Remote Sensing Systems (http://www.ssmi.com):

Hilburn, KA, FJ Wentz, 2008, Intercalibrated passive microwave rain products from the unified microwave ocean retrieval algorithm (UMORA), Journal of Applied Meteorology and Climatology, 47, 778-794.

i think thermal remote sensing will give some clue.. or you can try high resolution data which will provide the detail picture and extract the features by taking separated from others.

pls have a look

http://www.trans-servisgroup.ru/pdf/trans-servis_-_advertising_material_(thermal_remote_sensing).pdf.

or

http://www.netl.doe.gov/kmd/cds/disk37/B%20-%20Native%20American%20Program/NT15453_Final%20Presentation,%20Feb%202004.pdf.

There is ongoing development on that issue in the DINEOF software suite, (http://modb.oce.ulg.ac.be/mediawiki/index.php/DINEOF ) , A " EOF-based method to fill in missing data from geophysical fields, such as clouds in sea surface temperature". Which also handle mutlivariate satellite information .

As i'm only indirectly involved in the project i can only redirect you to the website, where you can find papers and documentation about the tool, and eventually download the tool itself.

Good work !

Thank u Selavm and Tashi:)

I think it would be difficult to distinguish paddy from other vegetation using landsat image

Please go through this paper

The application of ASTER remote sensing data to porphyry copper and epithermal

gold deposits

Amin Beiranvand Pour, Mazlan Hashim

doi:10.1016/j.oregeorev.2011.09.009

Consider using L band radar if your areas are covered with sand. L band sees through sand to materials below due to it's long wavelength. Keep in my that with all other data products, you are seeing only the surface sand/dirt/ veg/ rock whatever. Aster is great, but I am a fan of believing the more data, the better the project!

I use a variety of methods when exploring data. I like using Sprectral Angle Mapper (SAM) but I also like using Minimum Noise transform (MNF). Both are available in ENVI.

Hi there,

Depending on your term of canopy cover percentage, the analysis may varies. If you're talking about cover percentage, then, you should consider the nature (i.e. diameter, composition) of the tree that you want to identify. It will determine the specification of image that can be effectively used.

using very high spatial resolution data won't be effective because each pixel is sub-canopy extent, and thus, at per-pixel analysis, there will be many pixels that have 100% cover because the pixel consist only tree leaves.

the term percentage itself means that it is the percentage of area covered by tree canopy per unit area. Thus, you should define your unit of analysis before progressing. If not, quantifying the percentage will be difficult and the result will be difficult to compare.

If you have already finish with the aforementioned issues, there are many method that can be used to improve satellite-based canopy cover percentage identification such as vegetation index and unmixing. Furthermore, to obtain a real quantiative estimate of canopy cover, you should at least perform field survey, or if you use medium spatial resolution image, you can use higher resolution data to get canopy cover information. Afterward, you can perform empirical modeling using vegetation index, PC bands, image fractions or other approach.

And if you want to separate tree from background reflectance i.e. grass, you can try applying PCA with vegetation mask on. This will maxed out the variation of vegetation pixels and object separation will be easier.

Hope this help.

Thank you.

First of all, check somehow all papers written by Foody and Stehman. These will answear all your questions.

Generally, there is no "perfect" way. It depends on your application, classifier, study area, budget, limitations, and many more ...

There is a European project about it with good information:

Good luck

Its quite obvious that different users will have different needs regarding habitat characterization. Two issues seem noteworthy:

One should be aware of the difficulties in classifying a habitat from remote sensing data, even with the use of good ancillary data - field work is most recommended. From an ecologic point of view its critical to have a clear notion of the bioclimatic and biogeographic context of each habitat (hence the need for good ancillary data), together with a set of diagnostic species that will assist you to make a correct diagnostic on the field regarding the presence/absence of a certain habitat. Having said that, its also evident that this is a valid approach for most habitats (meaning NATURA 2000 habitats), but not all of them: Cavities (habitat 8310) have no diagnostic species and are mostly dependent on geological/geomorphological context and habitats like Estuaries (habitat 1130) are composed of a complex mosaic of plant communities, many of them included in other NATURA 2000 habitats.

Finally, from the manager point of view some patch metrics may be of high interest, especially if trends over time can be studied using these metrics. This may require retrospective studies but because of the reasons I mentioned earlier, many limitations may occur in this process. Also a good diagnostic of the (potential) threats and disturbance sources (some of which may be useful to the habitats persistence) may be of high interest.

In the Portuguese case some of this information was put together in what was called: 'NATURA 2000 characterization and management files', that can be found here (http://www.icn.pt/psrn2000/caract_habitat.htm ) but because these are in Portuguese, a synthesis of this may be found here (https://www.researchgate.net/publication/231521517_The_application_of_the_Habitats_Directive_in_Portugal). Hope you find it useful.

this algorithm may help you retrieve the temperature from Landsat image

Ts = γ [e-1 (ψ1 Lsensor + ψ2) + ψ3] + δ

with

γ = {c2 Lsensor / Tsensor2 + [λ4 / c1 Lsensor + λ-1]}-1

and

δ = - γ Lsensor + Tsensor

where :

Ts – surface temperature calculated using generalized single-channel method;

Lsensor – at-sensor radiance in W m-2 sr -1 μm-1;

Tsensor – at-sensor brightness temperature in K;

γ – effective wavelength (11.457 μm for TM6 band);

e – emissivity;

c1, c2 – constants (c1 = 1.19104 108 W μm4m-2 sr -1; c2 = 14387.7 μm K);

ψ1, ψ1, ψ1 – atmospheric functions obtained as a function of the total atmospheric water vapour content (W)

I would like to do a premise first. Usually the urban heat island (UHI) phenomenon is intended as the difference of Air temperature (AT) in the urban areas in respect of surrounding peripheral or rural ones. From airborne/satellite satellite what can be "directly" inferred is the Land Surface Temperature (LST), which is sometimes used as input for modeling to retrieve AT. A term used to remind this difference is the Surface Urban Heat Island (SUHI), which is related to the difference of LST.

So, in trying to answer to your question I'll refer to satellite data to calculate LST as a proxy to then model AT.

To make a choice (or an integration) between the sensors some parameters to focus are: 1) spectral resolution, 2) spatial resolution and 3) temporal resolution.

1) Concerning the spectral resolution, you should consider that at least one band with wavelength in the Thermal Infra Red range (~8-12micrometers) (TIR) is required. With 1 TIR band (e.g. Landsat family) you can apply single-channel algorithms: needs a reliable calibration, very sensible to the area of application and type of material, very dependent on the removal of the atmospheric effects. With 2 TIR (e.g. MODIS and AVHRR or the old (A)ATSR or the future VIIRS-OLCI) bands you can apply split windows algorithm, mitigating atmospheric effect, but the type of material is still critical (emissivity). With >3 TIR bands (e.g. ASTER or usuallly airborne sensors like the ATLAS or AHS40) you can apply advanced algorithms (like the Temperature Emissivity Separation - TES), but still the atmospheric correction is critical.

2) Concerning the spatial resolution you can go from the <=5 meters of airborne sensors, to the 60/120m of Landsat, the 90m of ASTER and then there is a jump to the 1km resolution of MODIS/AVHRR, up to the 3-4km of MSG.

3) Finally concerning the temporal resolution, the golden rule is usually: higher resolution means lower repetition frequency (e.g. >2 daily acquisitions for MODIS/AVHRR/MSG, about every 16 days ASTER/Landsat).

Then depending on the application you have to do a trade off between those 3 elements.

For a big city (e.g. of >20km2), 1km data could be enough for historical analysis and/or monitoring. This could be integrated with few acquisitions at high resolution for a "zonation" of the city and to have a more precise characterization in terms of emissivity.

For medium and small cities 1km become no more enough and it is more difficult to find data suitable for monitoring, while you can more easily found ASTER/Landsat or airborne data for a limited historical study or again zonation.

Hope to be of help.

PS

I suggest you to give a look to the Urban Heat Island and Thermography ESA project (www.urbanheatisland.info) and to Sobrino, J. A., Jiménez-Muñoz, J. C., Sòria, G., Romaguera, M., Guanter, L., Moreno, J., Plaza, A. and Martínez, P., 2008. Land Surface Emissivity Retrieval From Different VNIR and TIR Sensors. IEEE Transactions on Geoscience and Remote Sensing, 46, 316 – 327

Thanks for the reply first.

It's about terrestrial GPP and by Chlorphyll index I mean the index derived from reflectance at 750 nm and 710 nm, which is derived as CI_re = (R750nm/R710nm -1, where R750 and R710 are reflectance at 750 nm and 710 nm respectively.

Regards,

Error matrix is used to find out the accuracy of classified raster layer. for more information you can go through this papers.

you can find out lot from google search

The scikit-learn library for python is best and easy to use–many machine learning algorithms a implemented in this library.

Dear Ali,

Sorry for not being able to get back to you promptly. I've downloaded the data. It seems that it was in Level 1 format. You need to use software from the website first. Then convert it to normal desktop image processing format.

You can download it from here --> http://geography.osu.edu/grads/xzhu/

we can also use aster DEM to create drainage layer as it is also of more resolution than SRTM provided in the cgiar website which has resolution of 90 m. At times the major drawback with the aster DEM fails in the way as to which algorithm has been used for the "fill" command in the ArcHydro tool in ArcGIS. If proper algorithm is provided for the Aster DEM data also then the drainage extraction from this data can provide better results than any other DEM data of less resolution. In SRTM the algorithm used for "Fill" command has given a better result for drainage delineation.

You can have a look at this paper [1]. You certainly need to preprocess your images, e.g. using [2].

[1] C. Sun and P. Vallotton, Fast Linear Feature Detection Using Multiple Directional Non-Maximum Suppression. Journal of Microscopy, 234(2):147-157, May 2009.

[2] R. Su, C. Sun, C. Zhang, and T. D. Pham, Linear Feature Enhancement based on Morphological Operation and Gabor Function, In Image and Vision Computing New Zealand, Dunedin, New Zealand, 26-28 November 2012, pp.91-96.

Hi Victor, a couple of options come to mind:

Fluorescence can be used to track hydrocarbons. I believe Turner Designs makes submersible sensors for this purpose. I am not sure of Turner's depth and platform capabilities.

In situ particle sizing instruments (Sequoia Scientific LISST-100X and -Deep) have also been used for oil plume tracking and for examining the effects of dispersants on the oil in situ and in test facilities (a couple sources below). The LISST-Deep is capable of deployment to 3000m. These instruments can be profiled, and can be adapted to various platforms, though I do not know of any researchers deploying specifically on an AUV for oil studies.

Li Z, Lee K, Kepkay P, King T, Yeung W, Boufadel MC, Venosa AD (2008): Wave Tank Studies on Chemical Dispersant Effectiveness: Dispersed Oil Droplet Size Distribution. In: Davidson WF, Lee K, Cogswell A (Eds.): Oil Spill Response: A Global Perspective. NATO Science for Peace and Security Series C: Environmental Security, Springer, pp. 143 – 157.

doi:10.1007/978-1-4020-8565-9

Li Z, Lee K, King T, Boufadel MC, Venosa AD (2010): Effects of temperature and wave conditions on chemical dispersion efficacy of heavy fuel oil in an experimental flow-through wave tank. Marine Pollution Bulletin 60(9): 1550-1559.

doi:10.1016/j.marpolbul.2010.04.012

Trudel K, Belore RC, Mullin JV, Guarino A (2010): Oil viscosity limitation on dispersibility of crude oil under simulated at-sea conditions in a large wave tank. Marine Pollution Bulletin 60(9): 1606-1614.

doi:10.1016/j.marpolbul.2010.01.010

Hi Victor, I don't think that there is any point using TIR. The simplest reason is spatial resolution. Because of low SNR the footprint of TIR sensor is usually 2 times the size of VIR and 4 times of PAN...

As oil slicks are dominantly detected by their geometric properties (SAR roughness, form and size...) you would not add any specific information to the algorithm with TIR.

The companies promotional material after the trial is attached

Actually, the same authors are very active in this field:

Hain, Christopher R., John R. Mecikalski, Martha C. Anderson, 2009: Retrieval of an Available Water-Based Soil Moisture Proxy from Thermal Infrared Remote Sensing. Part I: Methodology and Validation. J. Hydrometeor, 10, 665–683. doi: http://dx.doi.org/10.1175/2008JHM1024.1

Yilmaz, M.T., W.T. Crow, M.C. Anderson and C. Hain, "An objective methodology for merging satellite and model-based soil moisture products," Water Resources Research, 48, W11502, doi:10.1029/2011WR011682, 2012.

Hain, C.R., W.T. Crow, M.C. Anderson and J.R. Mecikalski, "An EnKF dual assimilation of thermal-infrared and microwave satellite observations of soil moisture into the Noah land surface model," Water Resources Research, 48, W11517, doi:10.1029/2011WR011268, 2012.

... and continue to publish on that

It is outside my field

R.G,Rastogi

Some of the trends in your AVHRR data look like they are related to sensor or orbital drift.

Here is one paper where the authors discuss this problem and their attempts to correct for it. This may help you find a solution.

Tucker,CJ, Pinzon, JE, Brown, ME, Slayback, DA, Pak, EW, Mahoney, R, Vermote, EF, and Saleous, NE. 2005. An extended AVHRR 8-km NDVI dataset compatible with MODIS and SPOT vegetation NDVI data. International Journal of Remote Sensing 26:4485-4498.

Emmanuel,

I guess the 5/beam attenuation coefficient distance includes total pulse travel distance down and up, right?

Surface roughness has influence on air dynamics by influencing its velocity due to friction. In high level of roughness it may also divert direction of wind flow through gullies or narrow troughs. As is known that wind speed, with humidity level and sunshine bear heavily in the rate of evaporation greater the wind velocity, the larger amount of the local moisture in the air, due to vapour pressure, will move away and increasing the process of evaporation of soil moisture. Surface roughness mostly retard the wind velocity therefore decreases the intensity of evaporation of soil moisture. However, in some situations roughness may act conversely. For example, if the wind is directed into a narrow valley, its velocity shall increase so may increase evaporation of soil moisture on valley floor.

Victor, it has been a long time since I played with giant lasers, so I'm not sure this helps, but back in the day (ca. 1999) topographic LIDAR vertical resolution was about 15 cm. See e.g. Meridith, Andrew W., David Eslinger, and Dirk Aurin. An evaluation of hurricane-induced erosion along the North Carolina coast using airborne LIDAR surveys. Coastal Services Center, 1999.

Full-text available on my RG page.

Check the link below:

Dear Joshi,

from my stand of fiew, the following definition is possible:

Land Cover

Land cover is a natural scientific category. That means that we describe the fact what is located on a specified area (e.g. wood, soil, rock).

Land Use

In contrast to this, land use is more anthropocentrically defined. That means that we give an impression what we do on a specified area. We give an information of the intensity (e.g. intensive, extensive, not used) of our usage.

With best wishes, Erik!

If there is, I think the place to look would be with NOAA. The GTSPP may be what you are looking for.

Try this website

If you are still interested in calculating hypsometric curve look at attached image and try to do the same.

Steps:-

Delineate watershed in ARCGIS or what ever tool like to do so.

Export the table information of DEM into ASCII and import the same into MS Excel and try to arrange as arranged in the attached image file

Value represents the elevation information and arranged in descending order, count represents number of pixel of particular elevations. Area is represented in terms of pixel (since pixel size is fixed and hence represent fixed area)

Elevation above baseline = A2-MIN(A:A)

Area Above given line of elevation = number of pixel above that particular elevation

Normalised area = (D2-MIN(D:D))/(MAX(D:D)-MIN(D:D))

Normalized Elevation = (C2-MIN(C:C))/(MAX(C:C)-MIN(C:C))

Check on this link: http://landsathandbook.gsfc.nasa.gov/handbook.html

Vaijayanti,

If you are still in the process of looking for documentation, here after is the reference to a paper recently released:

"Reconstructing satellite images to quantify spatially explicit land surface change caused by fires and succession: A demonstration in the Yukon River Basin of interior Alaska", Huang et al., 2013, Rem Sensing Environ, at the address

Even if this paper deals with LULC changed caused by fires, I consider that reading it will help you in a proper selection of the further steps in your study.

I am not wholly familiar with the method, but just by following the math in the paper, which indicates that Eqn. 15 is derived by differentiation of Eqn. 13, it seems that (theta)n+1 does not refer to value from adjacent pixel, rather it is a small deviation from the actual (theta)n value. Essentially: delta(theta) = [(theta)n+1] - [(theta)n], where delta(theta) is the small deviation that is used to define the differentiation approximation. Similarly for delta(phi).

I don't know how exactly this equation is used in simulation or phase-to-height conversion, perhaps that would clear things up more.

Also maybe have a look at a practical we have for msc students which explores some of the issues ...http://www2.geog.ucl.ac.uk/~plewis/geogg124/phenology.html

LiDAR has been used to measure depth successfully in coastal and other clear waters. Hence its good for bathymetry in such waters. However if there are any chemical or suspended solids that are leading to differences in light dispersal within the water column then LiDAR is no use at all. For Australia that means most inland waters cant apply this technology becuase we highly dispersive clays.

http://sirius.spotimage.com/PageSearch.aspx?language=UK&AspxAutoDetectCookieSupport=1 (for archived data)

Related to purchase you have to contact with the local vendor.

Dear all, thank you for your answers. I found the method to perform the fuzzy accuracy assessment in this paper: Gopal, S., and Woodcock, C. E., 1994, Accuracy assessment of Thematic Maps using fuzzy sets I: theory and methods. Photogrammetric Engineering and Remote Sensing, 60, 181±188.

What kind of course are you thinking of? basic level or advanced? to do what precisely.