Earth Trend Modeller in Idrisi Selva might suit your needs. :)

I'm not an expert in time series analysis but perhaps this CRAN task view gives you a starting point which package to use:

http://cran.r-project.org/web/views/TimeSeries.html

for raster data time series analysis in R you should have a look at this Website about the recently developed greenborwn package:

http://greenbrown.r-forge.r-project.org/

GeogWeighted Regression, Available in both R and ArcGIS. To include time simply calculate the % change in landcover between periods and use this as your dependent variable.

If your landcover data is qualitative classes you can use spatial logistic regression by giving each possible transfer between classes a unique code, the regression will then provide the co-variates to the probability of a given transition.

Dear Indranil, ArcGIS provides Geographically Weighted regression http://resources.arcgis.com/en/help/main/10.1/index.html#//005p00000021000000

where you can do regression analysis of rasters, but I am not sure it will fulfill your requirement or not.

I used Idrisi Selva for my land use modeling, however, for driving factor analysis it has another module called "Cramer's V Analysis". What I did and I suggest you is:

1. convert your rasters to ASCII files,

2. convert the matrices into each column by avaliable software

3. Feed into SPSS and conduct regression Analysis

Hope that helps, if you want more clarification, you can ask me!!

You may use Arc/GIS software (Temporal Analyst ). For more information see the following links:

http://www.crwr.utexas.edu/gis/gishydro05/Time/DHI_Temporal_Analyst/TemporalAnalystForArcGIS.htm

http://blogs.library.duke.edu/data/2011/07/18/time-series-visualizations-in-arcgis-an-introduction/

http://www.esri.com/software/arcgis/extensions/trackinganalyst

I haven't applied but you can try QGIS + SEXTANTE + R. Here is details: http://www.r-bloggers.com/getting-raster-histogram-in-qgis-using-sextante-and-r/

I would like to suggest not using regression model but to use Change detection or change vector analysis in order to find land use changes during the time and you can do that by using different software such SAGA GIS (open source) or ERDAS Imagine or Idrisi...... otherwise you can use Geographically Weighted regression in ArcGIS as some colleagues mentioned.

I would recommend IDRISI modules for change detection including Land Change Modeler and Earth Trends Modeler, bit with rather sophisticated trends analysis, visualisation, stats etc (http://clarklabs.org/). Reasonable learning curve (that is not very hard) and well documented software with good tutorials

IDRISI provides ready to use tools for time series analysis of raster data. The tools are collected in an application called Earth Trends Modeler (ETM), that facilitates the visualization pre-processing and analysis of image time series. Regarding regression between time series, you can do this in ETM within the tool called Linear modeling. You can find more information on ETM here: http://clarklabs.org/products/Earth-Trends-Modeler.cfm

javascript:

The examples and responses here that are pointing you towards time series analyses or geostatistical approaches are probably not what you're looking for.

I think what many of the answers here are missing is that what Indrani, the original poster, is asking about is *not* a time series in the traditional, statistical sense. Since it is a three date comparison most of the approaches, software and methods that are put forth here as solutions aren't really applicable. Nor is the problem particularly well suited for traditional geostatistical techniques since the response variable is a classification, or categorical representation, of land use or land cover.

You can certainly accomplish what I think he's asking about, i.e. modeling land cover or land use transitions between classified states, using binary or multinomial logistic regression, with or without the added complexity of a spatial dependence structure, using R or other freely available tools.

If R is your tool of choice, I would definitely echo Xuanlong Ma's suggestion to check out Hijman's raster package for R, as it makes manipulating data and covariates as rasters much more efficient. However, if you're not familiar with R and its packages

Continuing the previous answer, in the package raster of R, you can find the command stack, which can help you to pack all rasters

GRASS GIS 7 (development version) has a new temporal framework: https://archive.org/details/TheTemporalGrassGisFrameworkIntroductionAndApplication

Xuanlong Ma's suggestion is a cool one. I have enjoyed using the raster package in R

For general analysis ERDAS is good . Also as some one mentioned that TIMESAT is a good software package. It helps for analysing time-series of satellite sensor data. It has scope to include dynamic properties of vegetation, such as phenology etc. The temporal domain holds important information about short- and long-term vegetation changes. It was earlier used for noise suppression of NOAA AVHRR data.

We are offering an excellent forum for this type of discussions. Why not attending ForestSat 2014? Deadline for abstracts on April 30th, visit http://forestsat2014.com

Dear

you can check in

r-gis.net

You can try by using Geographically Weighted Regression software described in Fotheringham et al. (2002). You can find the latest release at the website  http://geoinformatics.wp.st-andrews.ac.uk/gwr/gwr-software/, or you can use the Geographically Weighted Regression tool in the Spatial Statistics Toolbox of ArcGIS. There is also an R tool at http://cran.r-project.org/web/packages/spgwr/index.html.

You can use Idrisi kilimanjaro software.

the elaboration by Xuanlong Ma is wonderful and something I am also working with

Of course, using the R package 'raster'. The stack or brick method in this package can combine some satellite images into a time series. afterward, many time series analysis could be performed for each pixel, such as break-point checking, data filter. 

R has many packages which you can use to do any time series analysis of raster data. The packages "Raster", "rasterVis" and "zoo", can help you analyse your time series data.

IDRISI software has a great set of tools for that

Dear I suggest to you use Python software

My regards

Dear if you go through the help of R language you will find model ready to use

regards

If you have an attribute data as a table (e.g. csv) derived from your raster image, you can then process the data using {ggplot} library in R for regression analysis, for example as here, applied for 25 bath profiles. The final image will be as 25 profiles plotted as multi-facetted image on one layout: Code R script for the regression analysis of the 25 bathymetric p...