5. ValidationFollowing Fraser, Hall, and Landry (2001) and Smith,Woost translation - 5. ValidationFollowing Fraser, Hall, and Landry (2001) and Smith,Woost Indonesian how to say

5. ValidationFollowing Fraser, Hall

5. Validation
Following Fraser, Hall, and Landry (2001) and Smith,
Wooster, Powell, and Usher (2002), the ability of the rulebased
method described in Section 4 to accurately map
burned area was tested by comparison to burned area maps
derived from multi-spectral supervised maximum likelihood
classification of Landsat-7 ETM+ imagery (Table 2). In
total, 14 separate burns varying in size by two orders of
magnitude were located and mapped using the ETM+ data,
with the resulting burned area maps compared to our VGTderived
results. These burns occurred in deciduous needleleaf
forest, deciduous broadleaf forest, mixed forest, and
wooded tundra, which are the land cover classes subject to
severest burns in 2001 (as shown in Table 4). We can be
sure that these were new rather than old burns because we
were able to identify their date of formation using the VGT
time series. Table 3 and Fig. 4 indicate that all 14 newly
burned areas identifiable in the ETM+ data were successfully
detected by VGT, but in all cases, Fire 1 aside, VGT
underestimates the size of the burnt area by between 3% and
62%. This is because the rule-based method that we applied
to the VGT data tends not to select all pixels on the
perimeter of burned areas because many of these pixels will
be dominated by unburned vegetation. The spectral reflectance
characteristics of these particular ‘‘mixed’’ perimeter
pixels prevent them being classified as burnt by our criteria. This is a markedly different result to that of Fraser et al.,
who found SPOT VGT to overestimate the size of burnt
areas in Canadian forests by an average of 71% when
compared to ETM+. It is possible that the method used by
Fraser et al. selected a higher percentage of pixels containing
sub-pixel bunt patches than did our method, though,
conversely, this means that many of the 1-km pixels selected
as burnt in the Fraser et al. study were, in fact, only partly
burnt. This is something that is easily confirmed via
comparison to the much higher spatial resolution ETM+
data. Actually, even at the 30-m scale, individual ETM+
pixels maybe comprised of a fragmented mosaic of burnt
and unburnt patches, but at present, we have little very high
resolution information (e.g., aerial photography) to assess
this effect in Russian forests. In summary, the threshold
values used in our rule-based procedure are set stringently to
minimise errors of commission (i.e., to reduce ‘false’ fire
scar detection) but with the disadvantage that fire-affected
pixels containing considerable amounts of unburnt vegetation
may quite frequently remain unidentified. This effect
can, to some extent, be observed in Fig. 3, in which the
VGT-derived product (Fig. 3d) appears to have missed the
leftmost parts of the burn scar when compared to the ETM+-
derived product (Fig. 3f). This, however, is counteracted by
the fact that in many other regions of the burn, the ETM+
data confirms that the coarse 1-km2 VGT pixels classified as
‘burnt’ are in fact a mix of burnt and unburnt areas. In this
case, the final burnt area estimate derived from the ETM+
and VGT data sets differed by less than 2 km2 (10%).
There is a particularly large underestimate (>50%) for the
VGT-derived areas of Fires 7 and 14 in Table 3. The reason
for this is that these two fire scars are crossed by a river,
whose width in the AARS land–water mask is 90% greater
than in the VGT imagery. Thus, some of the burnt pixels Whilst the ETM+ data used in the validation effort do
encompass the land cover classes most severely affected by
fire in 2001, it is clearly the case that increased confidence
could be placed in the validation procedure if the number of
ETM+ scenes used was increased and was able to cover all
the affected land cover classes. This, however, is difficult
because of the limited number of ETM+ scenes available
(usually three to six scenes each year for any particular
location) and the fact that many scenes are corrupted by
cloud and ice/snow cover. We have checked another 12
ETM+ quicklook (240 m) resolution scenes containing 40
additional fire scars (in the same land cover classes as the
four validation ETM+ images). An example of this comparison
is shown in Fig. 5, and comparison between these
ETM+ quicklooks and the VGT-derived fire scar map
indicates that VGT detected 98% of the burned areas present
in the quicklook imagery, with the missed scars having areas
1 km2 or smaller. In general, as with the previous validation
data set, the VGT mapped fire scars again appear smaller
than they do in ETM+. We plan to adopt Landsat TM and
ETM+ imagery as the validation data for the remaining
years of our study and to produce a much more comprehensive
validation report, covering aspects of fire timing,
position, and burned area land cover, along with confirmation
of the hypothesis that the degree of area under- or
overestimation is a function of fire scar size.
0/5000
From: -
To: -
Results (Indonesian) 1: [Copy]
Copied!
5. ValidationFollowing Fraser, Hall, and Landry (2001) and Smith,Wooster, Powell, and Usher (2002), the ability of the rulebasedmethod described in Section 4 to accurately mapburned area was tested by comparison to burned area mapsderived from multi-spectral supervised maximum likelihoodclassification of Landsat-7 ETM+ imagery (Table 2). Intotal, 14 separate burns varying in size by two orders ofmagnitude were located and mapped using the ETM+ data,with the resulting burned area maps compared to our VGTderivedresults. These burns occurred in deciduous needleleafforest, deciduous broadleaf forest, mixed forest, andwooded tundra, which are the land cover classes subject toseverest burns in 2001 (as shown in Table 4). We can besure that these were new rather than old burns because wewere able to identify their date of formation using the VGTtime series. Table 3 and Fig. 4 indicate that all 14 newlyburned areas identifiable in the ETM+ data were successfullydetected by VGT, but in all cases, Fire 1 aside, VGTunderestimates the size of the burnt area by between 3% and62%. This is because the rule-based method that we appliedto the VGT data tends not to select all pixels on theperimeter of burned areas because many of these pixels willbe dominated by unburned vegetation. The spectral reflectancecharacteristics of these particular ‘‘mixed’’ perimeterpixels prevent them being classified as burnt by our criteria. This is a markedly different result to that of Fraser et al.,who found SPOT VGT to overestimate the size of burntareas in Canadian forests by an average of 71% whencompared to ETM+. It is possible that the method used byFraser et al. selected a higher percentage of pixels containingsub-pixel bunt patches than did our method, though,conversely, this means that many of the 1-km pixels selectedas burnt in the Fraser et al. study were, in fact, only partlyburnt. This is something that is easily confirmed viacomparison to the much higher spatial resolution ETM+data. Actually, even at the 30-m scale, individual ETM+pixels maybe comprised of a fragmented mosaic of burntand unburnt patches, but at present, we have little very highresolution information (e.g., aerial photography) to assessthis effect in Russian forests. In summary, the thresholdvalues used in our rule-based procedure are set stringently tominimise errors of commission (i.e., to reduce ‘false’ firescar detection) but with the disadvantage that fire-affectedpixels containing considerable amounts of unburnt vegetationmay quite frequently remain unidentified. This effectcan, to some extent, be observed in Fig. 3, in which theVGT-derived product (Fig. 3d) appears to have missed theleftmost parts of the burn scar when compared to the ETM+-derived product (Fig. 3f). This, however, is counteracted bythe fact that in many other regions of the burn, the ETM+data confirms that the coarse 1-km2 VGT pixels classified as‘burnt’ are in fact a mix of burnt and unburnt areas. In thiscase, the final burnt area estimate derived from the ETM+and VGT data sets differed by less than 2 km2 (10%).There is a particularly large underestimate (>50%) for theVGT-derived areas of Fires 7 and 14 in Table 3. The reasonfor this is that these two fire scars are crossed by a river,whose width in the AARS land–water mask is 90% greaterthan in the VGT imagery. Thus, some of the burnt pixels Whilst the ETM+ data used in the validation effort doencompass the land cover classes most severely affected byfire in 2001, it is clearly the case that increased confidencecould be placed in the validation procedure if the number ofETM+ scenes used was increased and was able to cover allthe affected land cover classes. This, however, is difficultbecause of the limited number of ETM+ scenes available(usually three to six scenes each year for any particularlocation) and the fact that many scenes are corrupted bycloud and ice/snow cover. We have checked another 12ETM+ quicklook (240 m) resolution scenes containing 40additional fire scars (in the same land cover classes as thefour validation ETM+ images). An example of this comparisonis shown in Fig. 5, and comparison between theseETM+ quicklooks and the VGT-derived fire scar mapindicates that VGT detected 98% of the burned areas presentin the quicklook imagery, with the missed scars having areas1 km2 or smaller. In general, as with the previous validationdata set, the VGT mapped fire scars again appear smallerthan they do in ETM+. We plan to adopt Landsat TM andETM+ imagery as the validation data for the remainingyears of our study and to produce a much more comprehensivevalidation report, covering aspects of fire timing,position, and burned area land cover, along with confirmationof the hypothesis that the degree of area under- oroverestimation is a function of fire scar size.
Being translated, please wait..
Results (Indonesian) 2:[Copy]
Copied!
5. Validasi
Setelah Fraser, Hall, dan Landry (2001) dan Smith,
Wooster, Powell, dan Usher (2002), kemampuan BerbasisPeran
metode yang dijelaskan dalam Bagian 4 secara akurat memetakan
daerah terbakar diuji dengan perbandingan peta area yang terbakar
berasal dari multi-spektral diawasi kemungkinan maksimum
klasifikasi Landsat-7 ETM + citra (Tabel 2). Dalam
total, 14 luka bakar terpisah bervariasi dalam ukuran dengan dua perintah
besarnya berada dan dipetakan menggunakan ETM + data,
dengan menghasilkan peta daerah terbakar dibandingkan dengan VGTderived kami
hasil. Luka bakar ini terjadi di daun needleleaf
hutan, hutan berdaun lebar daun, hutan campuran, dan
berhutan tundra, yang merupakan kelas tutupan lahan tunduk
luka bakar parah pada tahun 2001 (seperti yang ditunjukkan pada Tabel 4). Kami dapat
memastikan bahwa ini adalah baru daripada luka bakar lama karena kami
mampu mengidentifikasi tanggal mereka dari formasi menggunakan VGT
time series. Tabel 3 dan Gambar. 4 menunjukkan bahwa ke-14 yang baru
daerah terbakar diidentifikasi di ETM + data berhasil
terdeteksi oleh VGT, tetapi dalam semua kasus, Api 1 samping, VGT
meremehkan ukuran area yang terbakar oleh antara 3% dan
62%. Hal ini karena metode berbasis aturan bahwa kita diterapkan
untuk data VGT cenderung tidak memilih semua piksel pada
perimeter dari daerah yang terbakar karena banyak piksel ini akan
didominasi oleh vegetasi terbakar. Reflektansi spektral
karakteristik khusus ini '' dicampur '' perimeter
piksel mencegah mereka yang diklasifikasikan sebagai dibakar dengan kriteria kami. Ini adalah hasil yang sangat berbeda dengan yang Fraser et al.,
Yang menemukan SPOT VGT untuk melebih-lebihkan ukuran dibakar
area di hutan Kanada dengan rata-rata 71% ketika
dibandingkan dengan ETM +. Ada kemungkinan bahwa metode yang digunakan oleh
Fraser et al. dipilih persentase yang lebih tinggi dari piksel yang mengandung
patch sub-pixel bunt daripada metode kami, meskipun,
sebaliknya, ini berarti bahwa banyak piksel 1-km dipilih
sebagai dibakar di Fraser et al. Penelitian yang, pada kenyataannya, hanya sebagian
terbakar. Ini adalah sesuatu yang mudah dikonfirmasi melalui
dibandingkan dengan ETM resolusi spasial lebih tinggi +
data. Sebenarnya, bahkan pada skala 30-m, ETM individu +
piksel mungkin terdiri dari mosaik terfragmentasi dari bakaran
patch dan tidak terbakar, tapi saat ini, kami memiliki sedikit sangat tinggi
informasi resolusi (misalnya, foto udara) untuk menilai
efek ini di hutan Rusia. Singkatnya, ambang
nilai yang digunakan dalam prosedur berbasis aturan kami ditetapkan ketat untuk
meminimalkan kesalahan komisi (yaitu, untuk mengurangi kebakaran 'palsu'
deteksi bekas luka) tetapi dengan merugikan bahwa api yang terkena dampak
piksel mengandung jumlah yang cukup dari vegetasi terbakar
mungkin cukup sering tetap tidak teridentifikasi. Efek ini
dapat, sampai batas tertentu, diamati pada Gambar. 3, di mana
produk VGT yang diturunkan (. Gambar 3d) tampaknya telah melewatkan
bagian paling kiri dari bekas luka bakar jika dibandingkan dengan ETM + -
berasal produk (Gambar 3f.). Ini, bagaimanapun, adalah menetral oleh
fakta bahwa di banyak daerah lain luka bakar, yang ETM +
Data menegaskan bahwa kasar piksel VGT 1-km2 diklasifikasikan sebagai
'terbakar' sebenarnya campuran daerah yang terbakar dan tidak terbakar. Dalam hal ini
kasus, estimasi daerah yang terbakar akhir berasal dari ETM +
dan data VGT set berbeda dengan kurang dari 2 km2 (10%).
Ada meremehkan sangat besar (> 50%) untuk
daerah VGT yang diturunkan dari Kebakaran 7 dan 14 pada Tabel 3. alasan
untuk ini adalah bahwa dua bekas luka api ini dilintasi oleh sungai,
yang lebarnya di AARS tanah-air mask adalah 90% lebih besar
dari pada citra VGT. Dengan demikian, beberapa piksel bakaran Sementara ETM + data yang digunakan dalam upaya validasi dilakukan
meliputi kelas tutupan lahan yang paling parah terkena dampak
kebakaran pada tahun 2001, itu jelas kasus yang meningkatkan keyakinan
bisa ditempatkan dalam prosedur validasi jika jumlah
ETM + adegan digunakan meningkat dan mampu mencakup semua
kelas tutupan lahan yang terkena dampak. Ini, bagaimanapun, adalah sulit
karena terbatasnya jumlah ETM + adegan yang tersedia
(biasanya 3-6 adegan setiap tahun untuk setiap tertentu
lokasi) dan fakta bahwa banyak adegan yang rusak oleh
awan dan lapisan es / salju. Kami telah memeriksa lain 12
ETM + QuickLook (240 m) adegan resolusi yang berisi 40
bekas luka api tambahan (di kelas tutupan lahan sama dengan
ETM empat validasi + gambar). Contoh perbandingan ini
ditunjukkan pada Gambar. 5, dan perbandingan antara ini
ETM + quicklooks dan VGT diturunkan peta api bekas luka
menunjukkan bahwa VGT terdeteksi 98% dari daerah yang terbakar hadir
dalam citra tampilan cepat, dengan bekas luka rindu memiliki area
1 km2 atau lebih kecil. Secara umum, seperti sebelumnya validasi
data set, yang VGT dipetakan bekas kebakaran lagi tampak lebih kecil
daripada yang mereka lakukan di ETM +. Kami berencana untuk mengadopsi Landsat TM dan
ETM + citra sebagai validasi data untuk sisa
tahun penelitian kami dan untuk menghasilkan yang jauh lebih komprehensif
laporan validasi, meliputi aspek waktu kebakaran,
posisi, dan dibakar tutupan lahan daerah, bersama dengan konfirmasi
dari hipotesis bahwa tingkat daerah memahami atau
terlalu tinggi adalah fungsi dari ukuran api bekas luka.
Being translated, please wait..
 
Other languages
The translation tool support: Afrikaans, Albanian, Amharic, Arabic, Armenian, Azerbaijani, Basque, Belarusian, Bengali, Bosnian, Bulgarian, Catalan, Cebuano, Chichewa, Chinese, Chinese Traditional, Corsican, Croatian, Czech, Danish, Detect language, Dutch, English, Esperanto, Estonian, Filipino, Finnish, French, Frisian, Galician, Georgian, German, Greek, Gujarati, Haitian Creole, Hausa, Hawaiian, Hebrew, Hindi, Hmong, Hungarian, Icelandic, Igbo, Indonesian, Irish, Italian, Japanese, Javanese, Kannada, Kazakh, Khmer, Kinyarwanda, Klingon, Korean, Kurdish (Kurmanji), Kyrgyz, Lao, Latin, Latvian, Lithuanian, Luxembourgish, Macedonian, Malagasy, Malay, Malayalam, Maltese, Maori, Marathi, Mongolian, Myanmar (Burmese), Nepali, Norwegian, Odia (Oriya), Pashto, Persian, Polish, Portuguese, Punjabi, Romanian, Russian, Samoan, Scots Gaelic, Serbian, Sesotho, Shona, Sindhi, Sinhala, Slovak, Slovenian, Somali, Spanish, Sundanese, Swahili, Swedish, Tajik, Tamil, Tatar, Telugu, Thai, Turkish, Turkmen, Ukrainian, Urdu, Uyghur, Uzbek, Vietnamese, Welsh, Xhosa, Yiddish, Yoruba, Zulu, Language translation.

Copyright ©2025 I Love Translation. All reserved.

E-mail: