| 1. |
- Ehlers, Sarah, et al.
(författare)
-
Assessing Error Correlations in Remote Sensing-Based Estimates of Forest Attributes for Improved Composite Estimation
- 2018
-
Ingår i: Remote Sensing. - : MDPI AG. - 2072-4292. ; 10
-
Tidskriftsartikel (refereegranskat)abstract
- Today, non-expensive remote sensing (RS) data from different sensors and platforms can be obtained at short intervals and be used for assessing several kinds of forest characteristics at the level of plots, stands and landscapes. Methods such as composite estimation and data assimilation can be used for combining the different sources of information to obtain up-to-date and precise estimates of the characteristics of interest. In composite estimation a standard procedure is to assign weights to the different individual estimates inversely proportional to their variance. However, in case the estimates are correlated, the correlations must be considered in assigning weights or otherwise a composite estimator may be inefficient and its variance be underestimated. In this study we assessed the correlation of plot level estimates of forest characteristics from different RS datasets, between assessments using the same type of sensor as well as across different sensors. The RS data evaluated were SPOT-5 multispectral data, 3D airborne laser scanning data, and TanDEM-X interferometric radar data. Studies were made for plot level mean diameter, mean height, and growing stock volume. All data were acquired from a test site dominated by coniferous forest in southern Sweden. We found that the correlation between plot level estimates based on the same type of RS data were positive and strong, whereas the correlations between estimates using different sources of RS data were not as strong, and weaker for mean height than for mean diameter and volume. The implications of such correlations in composite estimation are demonstrated and it is discussed how correlations may affect results from data assimilation procedures.
|
|
| 2. |
- Granholm, Ann-Helen, et al.
(författare)
-
Estimating vertical canopy cover using dense image-based point cloud data in four vegetation types in southern Sweden
- 2017
-
Ingår i: International Journal of Remote Sensing. - : Informa UK Limited. - 0143-1161 .- 1366-5901. ; 38, s. 1820-1838
-
Tidskriftsartikel (refereegranskat)abstract
- This study had the aim of investigating the utility of image-based point cloud data for estimation of vertical canopy cover (VCC). An accurate measure of VCC based on photogrammetric matching of aerial images would aid in vegetation mapping, especially in areas where aerial imagery is acquired regularly. The test area is located in southern Sweden and was divided into four vegetation types with sparse to dense tree cover: unmanaged coniferous forest; pasture areas with deciduous tree cover; wetland; and managed coniferous forest. Aerial imagery with a ground sample distance of 0.24 m was photogrammetrically matched to produce dense image-based point cloud data. Two different image matching software solutions were used and compared: MATCH-T DSM by Trimble and SURE by nFrames. The image-based point clouds were normalized using a digital terrain model derived from airborne laser scanner (ALS) data. The canopy cover metric vegetation ratio was derived from the image-based point clouds, as well as from raster-based canopy height models (CHMs) derived from the point clouds. Regression analysis was applied with vegetation ratio derived from near nadir ALS data as the dependent variable and metrics derived from image-based point cloud data as the independent variables. Among the different vegetation types, vegetation ratio derived from the image-based point cloud data generated by using MATCH-T resulted in relative root mean square errors (rRMSE) of VCC ranging from 6.1% to 29.3%. Vegetation ratio based on point clouds from SURE resulted in rRMSEs ranging from 7.3% to 37.9%. Use of the vegetation ratio based on CHMs generated from the image-based point clouds resulted in similar, yet slightly higher values of rRMSE.
|
|
| 3. |
- Lindgren, Nils, et al.
(författare)
-
Data assimilation in stand level forest inventory – first results
- 2015
-
Ingår i: Natural resources and bioeconomy studies. - 2342-7639. ; 29, s. 37-37
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Data assimilation in stand-level forest inventory – first results Nils Lindgren 1 , Mattias Nyström1 , Jörgen Wallerman 1 , Sarah Ehlers 1 , Anton Grafström1 , Anders Muszta 1 , Kenneth Nyström1 , Erik Willen 2 , Johan Fransson 1 , Jonas Bohlin 1 , Håkan Olsson 1 , Göran Ståhl 1 1Swedish University of Agricultural Sciences, Umeå, Sweden 2Skogforsk, Uppsala, Sweden As we are entering an era of increased supply of remote sensing data, we believe that data assimilation has a large potential for keeping forest stand registers up to date (Ehlers et al. 2013). Data assimilation combines forecasts of previous estimates with new observations of the current state in an optimal way based on the uncertainties in the forecast and the observations. These forecasting and updating steps can be repeated with new available observations to get improved estimations. In the present study, we use canopy height models obtained from matching of digital aerial photos over the test site Remningstorp in Sweden, acquired 2003, 2005, 2007, 2009, 2010 and 2012 and normalized with a DEM from airborne laser scanning. Stem volume was estimated for each data acquisition and stand, using regression functions based on field reference data from sample plots. Forecasting was done with growth functions constructed from National Forest Inventory plots. The remote sensing estimates for each time point were assimilated with the forecasts of the previous estimates, using extended Kalman filtering. Validation was done on 40 m radius sample plots dominated by Norway spruce. Early results for three stands show that the variances were lower when using assimilation of new estimates and there was less fluctuation compared to repeated remote sensing estimates. The results for the assimilated data at year 2011 were also consistently closer to the validation data measured in 2011 compared to the remote sensing estimates from year 2011.
|
|
| 4. |
- Lindgren, Nils, et al.
(författare)
-
Data Assimilation of Growing Stock Volume Using a Sequence of Remote Sensing Data from Different Sensors
- 2022
-
Ingår i: Canadian Journal of Remote Sensing. - : Informa UK Limited. - 0703-8992 .- 1712-7971. ; 48, s. 127-143
-
Tidskriftsartikel (refereegranskat)abstract
- Airborne Laser Scanning (ALS) has implied a disruptive transformation of how data are gathered for forest management planning in Nordic countries. We show in this study that the accuracy of ALS predictions of growing stock volume can be maintained and even improved over time if they are forecasted and assimilated with more frequent but less accurate remote sensing data sources like satellite images, digital photogrammetry, and InSAR. We obtained these results by introducing important methodological adaptations to data assimilation compared to previous forestry studies in Sweden. On a test site in the southwest of Sweden (58 degrees 27 ' N, 13 degrees 39 ' E), we evaluated the performance of the extended Kalman filter and a proposed modified filter that accounts for error correlations. We also applied classical calibration to the remote sensing predictions. We evaluated the developed methods using a dataset with nine different acquisitions of remotely sensed data from a mix of sensors over four years, starting and ending with ALS-based predictions of growing stock volume. The results showed that the modified filter and the calibrated predictions performed better than the standard extended Kalman filter and that at the endpoint the prediction based on data assimilation implied an improved accuracy (25.0% RMSE), compared to a new ALS-based prediction (27.5% RMSE).
|
|
| 5. |
- Lindgren, Nils, et al.
(författare)
-
Improved Prediction of Forest Variables Using Data Assimilation of Interferometric Synthetic Aperture Radar Data
- 2017
-
Ingår i: Canadian Journal of Remote Sensing. - : Informa UK Limited. - 0703-8992 .- 1712-7971. ; 43, s. 374-383
-
Tidskriftsartikel (refereegranskat)abstract
- The statistical framework of data assimilation provides methods for utilizing new data for obtaining up-to-date forest data: existing forest data are forecasted and combined with each new remote sensing data set. This new paradigm for updating forest database, well known from other fields of study, will provide a framework for utilizing all available remote sensing data in proportion to their quality to improve prediction. It also solves the problem that not all remote sensing data sets provide information for the entire area of interest, since areas with no remote sensing data can be forecasted until new remote sensing data become available. In this study, extended Kalman filtering was used for assimilating data from 19 TanDEM-X InSAR images on 137 sample plots, each of 10-meter radius at a test site in southern Sweden over a period of 4 years. At almost all time points data assimilation resulted in predictions closer to the reference value than predictions based on data from that single time point. For the study variables Lorey's mean height, basal area, and stem volume, the median reduction in root mean square error was 0.4 m, 0.9 m(2)/ha, and 15.3 m(3)/ha (2, 3, and 6 percentage points), respectively.
|
|
| 6. |
- Nordkvist, Karin, et al.
(författare)
-
Vegetation classification in the Swedish sub-arctic using a combination of optical satellite images and airborne laser scanner data
- 2011
-
Konferensbidrag (refereegranskat)abstract
- The aim of this pilot study was to investigate to which degree the accuracy of automated vegetation classification in the Swedish sub-arctic could be improved by combining optical satellite data with airborne laser scanner (ALS) data, compared to using satellite data only. This information is of interest in an ongoing discussion about the possible inclusion of the mountains in northern Sweden in the national laser scanning that started in 2009. A SPOT 4 scene and ALS data from an Optech ALTM Gemini scanner, both from 2010, were used in maximum likelihood classification. Data for training and validation was obtained from 279 plots with 20 m radius that were visited in field 2010. These plots were located near Abisko in northern Sweden (lat. 68° 23' N, long. 18° 53' E), on the north and south side of Lake Torne Träsk. A classification scheme with 7 classes based on the Swedish mountain vegetation map was used. Classification using only SPOT data gave an over-all accuracy of 75.6%, and the combination of SPOT data and ALS data increased the accuracy to 81.4%
|
|
| 7. |
- Nyström, Mattias, et al.
(författare)
-
Assimilating remote sensing data with forest growth models
- 2015
-
Konferensbidrag (refereegranskat)abstract
- As we are entering an era of increased supply of remote sensing data, we believe that dataassimilation that combines growth forecasts of previous estimates with new observations of thecurrent state has a large potential for keeping forest stand registers up to date (Ehlers et al. 2013).The data assimilation will update a forest model e in an optimal way based on the uncertainties inthe forecast and the observations, each time new data becomes available. These forecasting andupdating steps can be repeated with new available observations to get improved estimations. In thisstudy we present the first practical results from data assimilation of mean tree height, basal area andgrowing stock. The remote sensing data used were canopy height models obtained from matching ofdigital aerial photos over the test site Remningstorp in Sweden. The photos were acquired 2003,2005, 2007, 2009, 2010 and 2012 and normalized with a DEM from airborne laser scanning.The procedure for the data assimilation was as follows: mean tree height, basal area and growingstock were predicted on 18 m × 18 m raster cells using the area based method. Ten meter radiussample plots were used as field calibration data. For each photo year, the field data were adjustedfor growth to have the same state year as each acquisition year of the photos. Growth models wereconstructed from National Forest Inventory plot data. Data assimilation could then be performed onraster cell level by initially start with the estimates from 2003 year´s photos. This prediction was thenforecasted to year 2005 by calculating the growth for the raster cell. This forecasted value is thenblended with the new remote sensing estimation collected 2005. The process was then repeated forthe following years where new measurements were available. In this study, extended Kalmanfiltering was used to blend the forecasted values with the new remote sensing measurements.Validation was done for 40 m radius field plots. Further, the results were also compared with twoalternative approaches: the first was to forecast the first remote sensing estimate to the endpointand the second was to use remote sensing data acquired at the endpoint only.The preliminary results for the eight forest stands show that the variances were lower when usingassimilation of new estimates and there were less fluctuation compared to only using remote sensingdata from the endpoint. However, the mean deviation from the measured value 2011 was lowerwhen only data from the endpoint were used. The assimilated values 2011 were consistently closerto the validation data compared to only forecasting the starting estimate from 2003 to 2011.
|
|
| 8. |
- Nyström, Mattias, et al.
(författare)
-
Change detection of mountain birch using multi-temporal ALS point clouds
- 2013
-
Ingår i: Remote Sensing Letters. - 2150-704X .- 2150-7058. ; 4, s. 190-199
-
Tidskriftsartikel (refereegranskat)abstract
- The use of multi-temporal laser scanner data is potentially an efficient method for monitoring of vegetation changes, for example, at the alpine treeline. Methods for relative calibration of multi-temporal airborne laser scanning (ALS) data sets and detection of experimental changes of tree cover in the forest–tundra ecotone was tested in northern Sweden (68° 20′ N, 19° 01′ E). Trees were either partly or totally removed on 6 m radius sample plots to simulate two classes of biomass change. Histogram matching was successfully used to calibrate the laser metrics from the two data sets and sample plots were then classified into three change classes. The proportion of vegetation returns from the canopy was the most important explanatory variable, which provided an overall accuracy of 88%. The classification accuracy was clearly dependent on the density of the forest.
|
|
| 9. |
- Nyström, Mattias, et al.
(författare)
-
Change detection of mountain vegetation using multi-temporal ALS point clouds
- 2011
-
Konferensbidrag (refereegranskat)abstract
- Multi-temporal laser scanner data to be used in change detection studies will most likely be acquired with different sensors, flying altitudes, and system parameters. Therefore, calibration is probably needed in order to make laser returns from vegetation comparable between two laser data acquisitions. In this study, two ALS point clouds were acquired with different sensors and flying altitudes. The first data set had 11.5 points m-2 and was obtained in 2008 with a TopEye MKII scanner and the second with a density of 1.1 points m-2 was obtained in 2010 with an Optech ALTM Gemini scanner. The test site was located in Abisko in northern Sweden with forest dominated by mountain birch. Six meter radius sample plots were placed in the forest-tundra ecotone and assigned one of the following treatments: (1) reference with no removal of trees, (2) removal of 50% of the total number of stems above 1.5 m, and (3) removal of 100% of the total number of stems above 1.5 m. Histogram matching was used to calibrate the two data sets and sample plots were then classified into the three treatments. The overall classification accuracy was 82% using only the proportion of vegetation returns from the canopy as explanatory variable. Features created from gridded laser data had overall higher classification accuracy than laser features created directly from the point cloud. Histogram matching made the two data sets comparable by reducing the difference between them. These early results show how changes can be detected even with different sensors, flying altitudes, and system parameters
|
|
| 10. |
- Nyström, Mattias, et al.
(författare)
-
Data assimilation : a prototype system to assimilate forest stand information
- 2016
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- The purpose of this report is to describe a data assimilation prototype program(Appendix A) developed to estimate forest stand data. The program was developed and tested on data col-lected on the forest estate Remningstorp in southern Sweden. Data assimilation can be used to sequentially combine remote sensing based estimates of forest variables with predictions from growth models. The assimilation routine implemented was the extended Kalman Filter. The program supports two different ways to assimilate the forest data: (1) pixel-wise and (2)stand-wise. In the pixel-wise way, raster cells are used as assimilation unit and can beaggregated to a stand for evaluation. In the stand-wise way, the whole stand is assimilatedas one unit. The two methods has pros and cons. The pixel-wise way is simple to use as nostand-delineation is needed, but might be subject to boundary effects and noise due to geo-metric errors. Using the developed code, it has been shown in three case studies that thecombination of time series of remote sensing data and forest growth functions provides bet-ter estimates of forest variables than only using forecasting, or only using the latest remotesensing data. This opens up for a new way to keep forest stand registers up to date.
|
|
