27/10/2011

Remove haze with GRASS GIS through QGIS


Memadamkan kabus menggunapakai GRASS GIS melalui QGIS

Semalam, saya menerima permintaan emel untuk menasihat bagaimana untuk import imej remote sensing (Landsat) kedalam GRASS GIS bagi membantu dalam memadamkan kabus dengan source code. Saya menjawab saya tidak mahir dengan GRASS GIS tetapi menasihat untuk menggunapakai  GeoReferencer dalam QGIS yang mudah dan pasti dapat import imej raster kedalam GIS. Bagaimana pun, saya heran mengapa walaupun kabus dapat dipadamkan, adakah ini bermakna bumi di bawah dapat dilihat? Perkara ini diluar bidang pengkhususan saya tetapi saya berpendapat tidak salah kalau dapat saya membantu. Seterusnya dibuat Google search dan tiba-tiba saya sampai disini. Nampaknya, laman ini adalah sebahagian daripada AWF-Wiki, landasan untuk berkongsi maklumat, ilmu dan kepakaran dalam konteks inventori perhutanan dan remote sensing yang dimulakan oleh Georg-August-Universität Göttingen, Germany. Dua perkara menarik perhatian: Pertama, saya dalam mendalami pasal automated cloud detection iaitu bidang raster dan kedua, terdapat syarahan memperkenalkan QGIS. Boleh tahan! Mendalami automated cloud detection, saya menyedari dapat saya melaksanakannya dengan pakej QGIS-GRASS tanpa penggunaan source code dan bergantung pada modul GRASS seperti r.composite, r.univari, r.mapcalc, r.buffer dan r.grow yang diterangkan  oleh AWF-Wiki. Ah...kalau macam ni, boleh saya lanjutkan pelajaran ke tahap PhD. ;-D  Memandangkan saya berpendapat laman AWF-Wiki amat berguna, saya telah meninggal pautan pada blog ini bagi mereka yang berminat.

Remove haze with GRASS GIS through QGIS

Yesterday, I received an email asking for help how to insert the remote sensing image (Landsat) in GRASS GIS to eventually help remove haze with a source code of haze removal. I replied I was not familiar with GRASS GIS but advised to use GeoReferencer in QGIS which is easy and will definitely solve the task of importing raster images in a GIS. However, I could not understand why even if haze could be removed , does that make it possible to see the land form underneath? Although this is beyond my area of expertise, I thought there was nothing wrong if I could help. I immediately made a Google search and eventually landed here. This appears to the part of a site belonging to AWF-Wiki, a platform for sharing information, knowledge and expertise in the context of forest inventory and remote sensing which was initiated  at the Georg-August-Universität Göttingen, Germany. Two things attracted me: Firstly, I got to know more about automated cloud detection which is a raster domain and secondly, there were lectures on the introduction of QGIS. Not bad! Reading more about automated cloud detection, I realized I could easily do it with the QGIS-GRASS package and instead of using source code, I could still rely on GRASS modules like r.composite, r.univari, r.mapcalc, r.buffer and r.grow as explained clearly by AWF-Wiki. Ah...like this, I can pursue a PhD. ;-D  Since I found the AWF-Wiki site such a great wealth of resource, I have left it's link at this blog for those interested.

15 comments:

  1. Che' Abbas,

    Wow, thank you for the link on the amazing QGIS tutorial from Germany (in English, no less) ! If you don't mind, I will include that link in my list of QGIS resources. (I think I'm going to study the GRASS portions myself, since I've been tinkering with SAGA GIS for the spatial analysis portion and not bothering to check out GRASS!) Thank you for your great posts!

    Howard Yamaguchi

    ReplyDelete
  2. It pays to help others doesn't it? because there seems to be a snowball effect in life for things done in sincerity. I am just happy I could help what is not my domain.

    ReplyDelete
  3. Tuan ini memang GENIUS GIS. Tahniah Tuan. Teruskan. Tuan , apa pandangan tentang setiap Jabatan / Agensi perlu ada sekurang kurangnya Bahagian GIS masing masing yang diketuai oleh orang macam Tuan. :)

    ReplyDelete
  4. The line between Genius and crazy people is very fine :-D Hanya Jabatan yang ada kaitan dengan data spatial elok tubuhkan unit GIS dan orang mana-mana pun boleh jadi boss cuma mesti ada minat.

    ReplyDelete
    Replies
    1. yup true. without interest it's just liability to gov other words to Malaysian

      Delete
    2. So how does one generate another person's interest? Here, different people have different needs, therefore, there is a need to study the subject and try to answer "What benefit do I get out of this?" from that subject's angle. To some, it is cost-benefit, to others convenience. This makes it challenging but at the same time interesting and rewarding if you can win a person's attention and undivided interest. To me, it's just a game, sometimes I win, sometimes I lose but I won't know until I try so I try ...

      Delete
  5. Saya rasa setiap Jabatan Perlu ada GIS kerana GIS boleh digunakan untuk membantu kerja perancangan.
    Nak tanya juga, boleh kah imej GOOGLE digunakan untuk mengemaskini data? Google x marah ke?

    ReplyDelete
  6. Google tak benar datanya dimuat turn dalam bentuk fitur dan atribut tapi data yang hampir sama boleh dimuat turun dengan OpenStreetMap.

    ReplyDelete
  7. Maksud saya menggunakan imej google save dia jadi jpeg kemudian georeferencing balik. Boleh ke tu?

    ReplyDelete
  8. Lagi satu nak tanya JPBD beli imej dengan remote sensing ke nak buat Land Use?

    ReplyDelete
  9. >Maksud saya menggunakan imej google save dia jadi jpeg kemudian georeferencing balik.

    Oh...tu tak jadi masalah

    >JPBD beli imej dengan remote sensing ke nak buat Land Use?

    Biasanya digunakan untuk buat kerja verifikasi gunatanah terutama bagi kawasan yang jauh kedalam dari jalan raya.

    ReplyDelete
  10. Thanks for the link Tuan Abbas! I've never seen that approach to landsat atmospheric correction.. Normally I use GRASS built-in modules such as i.atcorr (for atmospheric correction), and i.landsat.acca (for cloud removal) to do something in-line to what you're trying to achieve.. Those modules require no complex understanding in mathematical formulas (I suck at Math, hehe..)

    As for your inquiry about the visibility of the land under the clouds after the GRASS processing:
    1. i.atcorr corrects for the effects of atmospheric gases, aerosols, and thin cirrus clouds (not dense clouds). To my understanding it utilized those landsat multi-spectral band images to penetrate the atmosphere, leaving the land features intact..
    2. For cloud removal (i.landsat.acca), the land under the cloud won't be visible after processing, because there is no data under there in the 1st place. The cloud's removal will leave a hole. So what we need to do next is to overlay this processed image over other Landsat images for the same scene, but in different dates - hopefully the area image lost under the clouds will be covered by those other images..

    ReplyDelete
    Replies
    1. Haziq

      Orang minta tolong, kita cuba tolong, sejauhmana boleh. Bukannya tahu pun subject matter tu. Soalnya tak ke tak nak? Anyway, it was a calculated guess that I got that link. I am surprised but am happy my effort is not wasted.

      Delete
  11. hello.Thank you for your enlightening post.The link to the use of GRASS GIS for automated cloud detection seems to have been moved.are there any alternative tutorials you could recommend?

    ReplyDelete