Multi criteria analysis technique may help you in GIS.

https://www.researchgate.net/publication/269392574_Corafi_Bilgi_Sistemleri_Kullanarak_Bakonu_Danda_%28Kahramanmara%29_Orman_Yangn_Risk_Alanlarnn_Belirlenmesi_Determination_of_Forest_Fire_Risk_Areas_Using_Geographical_Information_Systems_in_Bakonu_Mountain_%28Kahramanmara%29

This is a paper about determination of forest fire risk areas Mt. Baskonus (in Mediterranean region). Language is Turkish, but you can see what variables used in the model.

Article Coğrafi Bilgi Sistemleri Kullanarak Başkonuş Dağında (Kahram...

Concerning your second question: as no/insufficient vegetation available for combustion (fuel load) inhibits the occurrence of a fire, I would at least include some quantitative measure of vegetation. Furthermore, humans are the major source of ignition in many ecosystems (and I could imagine in the Mediterraneans as well) - it thus makes sense to somehow account for those (e.g. proximity to roads, accessability,...) when determining fire risk or probability of ignition.  

Dear Manuel

Thank you for this answer to my question.

Just an idea about the vegetation.

As dried vegetation presents higher fire risk than healthy vegetation, it could be interesting to combine images from spring and summer, and try to use drop of a vegetation index between spring and summer as one of the risk indicators in undisturbed areas.

Or the combination of thermal and VNIR suggested by Amir Aghassi could possibly work even better, if the VNIR band was from spring, when the vegetation was green, and the thermal band from summer, when the dried vegetation areas get hot and prone to ignition.

Thank you Thomas for your contribution

Good day

Dear  Driss  

you can use MODIS LST data for forest fire detection. and also download KML file of MODIS prodect  . it will open in google earth and directly show on surface and you can show coordinate  of fire points nothing do to any analysis.....  

Dear Kuldeep

thank your for this answear.

Sincerly your's

hello,

A lot of things have been done in this field. I'll give some elements, but it would be interesting to have more details about your idea of "fire risk". Indeed, it exists many ways to interpret this concept and it would be helpful to know your idea in order to give you a better guidance.

The way I understand your need is that you want to characterize fire hazard. The MODIS product is probably not the best, except if you want near-real time assessment at coarse scale.

Usually, characterizing fire hazard with remote-sensing is done by monitoring vegetation's water content with diverse vegetation indices. But you will also need to characterize vegetation stand attributes (structure for instance) to derive potential fire behavior. You could also adopt a diachronic approach to compute other fie attributes of an area, like historical fires, fire interval, fire cycle, fire severity...basically to define the fire regime, wich is probably the best indicator of the average fire hazard over an area.

Dear François-Nicolas.

First, thank you for interesting answer

You hear about the fire regime, which is probably the best indicator of average risk of fire. However, is it possible to define from remotely sensed data, or only through a characterization on ground, according to the ecological component of the forest and / or structure? In the second case, how can we proceed or there's there any approved approach?

Thank you still again.

Defining a fire regime is far from simple and it requires a thorough knowledge of your area of interest. You basically need to know your fire return interval, the average type of fire and its intensity/severity and the fire season. You can derive some of those information from remote-sensing data. For instance calculating the severity using the Normalized Burn Ratio, among other indices. For anything related to historical fires, going on the field to collect tree samples with fire scars is important. Going on burn sites after a fire is important as well. But in any case you'll have to refer to literature on similar areas. I won't advise any method, as it depends a lot on your available data, computer resources, team size, budget...and also you actual goal. This study might help you though: http://www.publish.csiro.au/paper/WF10008

You may try to use a combination of land surface temperature (LST) and dryness index with Landsat imagery. Simply put LST image in one channel and the dryness image (MIR/NIR, eg, TM5/TM4) in the other channel. A pixel with high LST value as well as high dryness value could be at the risk of forest fire.

Dear Hanqiu Xu

Thank you very much .

There I find very interesting method. I'll test it and see what will give.

Best Regards

Land use and land management have a deep impact on the risk of fire. You should include a differentiation between land uses and major land use changes...

Dear Matteo

I do agree when you said the land use and land management have a profound impact on the risk of fire. But how can we include a differentiation between land uses and major changes of use? there's a scientific méhode (cartographic approaches eg using geomatics for example).

Thank you

Dear Alessandro

Thank you for all this informations and documents.

Best regards

Allessandro is right, niche modelling using the maximum of entropy approach has proven its usefulness. There are several software which implement it, like MAXENT or ModEco.

This paper might be of some interest as well: "Spatial variability in wildfire probability across the western United States", Parisien & al

In addition to the recommended answers above, it seems to me that the fuel load on the ground is essential for assessment of wildfire risk, but difficult to assess. Many 'forest' fires do not start within the forest, but along in road verges and/or abandoned grassland with a high fuel load (e.g. our 2010 article in FORECO).

dear you can use any satellite has thermal bands specially Landsat series satellite

good luck