Dynamic stratification for estimating pointwise forest characteristics
This paper deals with the testing of dynamic stratification for estimating stand level forest characteristics (basal areas, mean diameter, mean height and mean age) for a 117 ha study areas in Finland. The results do not show possibilities to achieve more accurate estimates using only Landsat TM pri...
Main Author: | |
---|---|
Format: | Article |
Language: | English |
Published: |
Finnish Society of Forest Science
1996-12-01
|
Series: | Silva Fennica |
Subjects: |
id |
doaj-5653079df10c4e4f92fe1eeeeca7c789 |
---|---|
record_format |
Article |
spelling |
doaj-5653079df10c4e4f92fe1eeeeca7c7892020-11-25T03:26:41ZengFinnish Society of Forest ScienceSilva Fennica2242-40752242-40751996-12-0130110.14214/sf.a9220Dynamic stratification for estimating pointwise forest characteristicsGintautas MozgerisThis paper deals with the testing of dynamic stratification for estimating stand level forest characteristics (basal areas, mean diameter, mean height and mean age) for a 117 ha study areas in Finland. The results do not show possibilities to achieve more accurate estimates using only Landsat TM principal components as auxiliary data opposed to static stratification. It was found that in dynamic stratification non-measured observations should be assigned the mean characteristics of the measured observations that belong to the same cube (class) instead of stratification variable classes until a certain limit. If only one principal component is used the number of classes has, however, little influence. Low field values are overestimated and high values underestimated. The only successful results were obtained using two variables of different origin – the qualitative development stage class and the quantitative 1st principal component. The lowest root mean square error in estimating basal area was 6.40 m2/ha, mean diameter 3.34 cm, mean height 2.65 m and mean age 14.06 years. This increase of stratification accuracy is mainly resulted by the use of development stage class as an auxiliary variable.smi forest management planning systemdynamic stratificationstand characteristicsstatic stratificationforest mensuration |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Gintautas Mozgeris |
spellingShingle |
Gintautas Mozgeris Dynamic stratification for estimating pointwise forest characteristics Silva Fennica smi forest management planning system dynamic stratification stand characteristics static stratification forest mensuration |
author_facet |
Gintautas Mozgeris |
author_sort |
Gintautas Mozgeris |
title |
Dynamic stratification for estimating pointwise forest characteristics |
title_short |
Dynamic stratification for estimating pointwise forest characteristics |
title_full |
Dynamic stratification for estimating pointwise forest characteristics |
title_fullStr |
Dynamic stratification for estimating pointwise forest characteristics |
title_full_unstemmed |
Dynamic stratification for estimating pointwise forest characteristics |
title_sort |
dynamic stratification for estimating pointwise forest characteristics |
publisher |
Finnish Society of Forest Science |
series |
Silva Fennica |
issn |
2242-4075 2242-4075 |
publishDate |
1996-12-01 |
description |
This paper deals with the testing of dynamic stratification for estimating stand level forest characteristics (basal areas, mean diameter, mean height and mean age) for a 117 ha study areas in Finland. The results do not show possibilities to achieve more accurate estimates using only Landsat TM principal components as auxiliary data opposed to static stratification. It was found that in dynamic stratification non-measured observations should be assigned the mean characteristics of the measured observations that belong to the same cube (class) instead of stratification variable classes until a certain limit. If only one principal component is used the number of classes has, however, little influence. Low field values are overestimated and high values underestimated.
The only successful results were obtained using two variables of different origin – the qualitative development stage class and the quantitative 1st principal component. The lowest root mean square error in estimating basal area was 6.40 m2/ha, mean diameter 3.34 cm, mean height 2.65 m and mean age 14.06 years. This increase of stratification accuracy is mainly resulted by the use of development stage class as an auxiliary variable. |
topic |
smi forest management planning system dynamic stratification stand characteristics static stratification forest mensuration |
work_keys_str_mv |
AT gintautasmozgeris dynamicstratificationforestimatingpointwiseforestcharacteristics |
_version_ |
1724591310112817152 |