Mika Karjalainen

Comparison of Laser and Stereo Optical, SAR and InSAR Point Clouds from Air- and Space-Borne Sources in the Retrieval of Forest Inventory Attributes

It is anticipated that many of the future forest mapping applications will be based on three-dimensional (3D) point clouds. A comparison study was conducted to verify the explanatory power and information contents of several 3D remote sensing data sources on the retrieval of above ground biomass (AGB), stem volume (VOL), basal area (G), basal-area weighted mean diameter (Dg) and Lorey’s mean height (Hg) at the plot level, utilizing the following data: synthetic aperture radar (SAR) Interferometry, SAR radargrammetry, satellite-imagery having stereo viewing capability, airborne laser scanning (ALS) with various densities (0.8–6 pulses/m2) and aerial stereo imagery. Laser scanning is generally known as the primary source providing a 3D point cloud. However, photogrammetric, radargrammetric and interferometric techniques can be used to produce 3D point clouds from space- and air-borne stereo images. Such an image-based point cloud could be utilized in a similar manner as ALS providing that accurate digital terrain model is available. In this study, the performance of these data sources for providing point cloud data was evaluated with 91 sample plots that were established in Evo, southern Finland within a boreal forest zone and surveyed in 2014 for this comparison. The prediction models were built using random forests technique with features derived from each data sources as independent variables and field measurements of forest attributes as response variable. The relative root mean square errors (RMSEs) varied in the ranges of 4.6% (0.97 m)–13.4% (2.83 m) for Hg, 11.7% (3.0 cm)–20.6% (5.3 cm) for Dg, 14.8% (4.0 m2/ha)–25.8% (6.9 m2/ha) for G, 15.9% (43.0 m3/ha)–31.2% (84.2 m3/ha) for VOL and 14.3% (19.2 Mg/ha)–27.5% (37.0 Mg/ha) for AGB, respectively, depending on the data used. Results indicate that ALS data achieved the most accurate estimates for all forest inventory attributes. For image-based 3D data, high-altitude aerial images and WorldView-2 satellite optical image gave similar results for Hg and Dg, which were only slightly worse than those of ALS data. As expected, spaceborne SAR data produced the worst estimates. WorldView-2 satellite data performed well, achieving accuracy comparable to the one with ALS data for G, VOL and AGB estimation. SAR interferometry data seems to contain more information for forest inventory than SAR radargrammetry and reach a better accuracy (relative RMSE decreased from 13.4% to 9.5% for Hg, 20.6% to 19.2% for Dg, 25.8% to 20.9% for G, 31.2% to 22.0% for VOL and 27.5% to 20.7% for AGB, respectively). However, the availability of interferometry data is limited. The results confirmed the high potential of all 3D remote sensing data sources for forest inventory purposes. However, the assumption of using other than ALS data is that there exist a high quality digital terrain model, in our case it was derived from ALS.

Comparing Accuracy of Airborne Laser Scanning and TerraSAR-X Radar Images in the Estimation of Plot-Level Forest Variables

Abstract: In this study we compared the accuracy of low-pulse airborne laser scanning (ALS) data, multi-temporal high-resolution noninterferometric TerraSAR-X radar data and a combined feature set derived from these data in the estimation of forest variables at plot level. The TerraSAR-X data set consisted of seven dual-polarized (HH/HV or VH/VV) Stripmap mode images from all seasons of the year. We were especially interested in distinguishing between the tree species. The dependent variables estimated included mean volume, basal area, mean height, mean diameter and tree species-specific mean volumes. Selection of best possible feature set was based on a genetic algorithm (GA). The nonparametric k -nearest neighbour ( k -NN) algorithm was applied to the estimation. The research material consisted of 124 circular plots measured at tree level and located in the vicinity of Espoo, Finland. There are large variations in the elevation and forest structure in the study area, making it demanding for image interpretation. The best feature set contained

TerraSAR-X Stereo Radargrammetry and Airborne Scanning LiDAR Height Metrics in Imputation of Forest Aboveground Biomass and Stem Volume

Our objective is to evaluate the boreal forest aboveground biomass (AGB) and stem volume (VOL) imputation accuracy when scanning LiDAR or TerraSAR-X stereo radargrammetry-derived point-height metrics are used as predictors in the nearest neighbor imputation approach. Treewise measured field plots are used as reference data in the AGB and VOL imputations and accuracy evaluations. The digital terrain model (DTM) that is produced by the National Land Survey of Finland is used to obtain aboveground elevation values for the TerraSAR-X stereo radargrammetry. The DTM that is used (i.e., grid size 2 m) is derived from LiDAR surveys with an average point density of ~ 0.5 points/m2. The respective DTM and point data are used in LiDAR imputations of AGB and VOL. The relative root mean square errors (RMSEs) for AGB and VOL are 29.9% (41.3 t/ha) and 30.2% (78.1 m3/ha) when using TerraSAR-X stereo radargrammetry metrics. The respective LiDAR estimation accuracy values are 21.9% (32.3 t/ha) and 24.8% (64.2 m3/ha). LiDAR imputations are clearly more accurate than imputations that are made by using TerraSAR-X stereo radargrammetry metrics. However, the difference between imputation accuracies of LiDAR- and TerraSAR X-based features are smaller than in any previous study in which LiDAR and different types of synthetic aperture radar materials are compared in the variable predictions regarding forests. We conclude that TerraSAR X stereo radargrammetry is a promising remote-sensing technique for large forest-area AGB and VOL mapping and monitoring when an accurate LiDAR-based DTM is available.

Artificial target detection with a hyperspectral LiDAR over 26-h measurement

Abstract. Laser scanning systems that simultaneously measure multiple wavelength reflectances integrate the strengths of active spectral imaging and accurate range measuring. The Finnish Geodetic Institute hyperspectral lidar system is one of these. The system was tested in an outdoor experiment for detecting man-made targets from natural ones based on their spectral response. The targets were three camouflage nets with different structures and coloring. Their spectral responses were compared against those of a Silver birch (Betula pendula), Scots pine shoots (Pinus sylvestris L.), and a goat willow (Salix caprea). Responses from an aggregate clay block and a plastic chair were used as man-made comparison targets. The novelty component of the experiment was the 26-h-long measurement that covered both day and night times. The targets were classified with 80.9% overall accuracy in a dataset collected during dark. Reflectances of four wavelengths located around the 700 nm, the so-called red edge, were used as classification features. The addition of spatial aggregation within a 5-cm neighborhood improved the accuracy to 92.3%. Similar results were obtained using a set of four vegetation indices (78.9% and 91.0%, respectively). The temporal variation of vegetation classes was detected to differ from those in man-made classes.

Cereal Yield Modeling in Finland Using Optical and Radar Remote Sensing

During 1996–2006, the Ministry of Agriculture and Forestry in Finland (MAFF), MTT Agrifood Research and the Finnish Geodetic Institute performed a joint remote sensing satellite research project. It evaluated the applicability of optical satellite (Landsat, SPOT) data for cereal yield estimations in the annual crop inventory program. Four Optical Vegetation Indices models (I: Infrared polynomial, II: NDVI, III: GEMI, IV: PARND/FAPAR) were validated to estimate cereal baseline yield levels (yb) using solely optical harmonized satellite data (Optical Minimum Dataset). The optimized Model II (NDVI) yb level was 4,240 kg/ha (R2 0.73, RMSE 297 kg/ha) for wheat and 4390 kg/ha (R2 0.61, RMSE 449 kg/ha) for barley and Model I yb was 3,480 kg/ha for oats (R2 0.76, RMSE 258 kg/ha). Optical VGI yield estimates were validated with CropWatN crop model yield estimates using SPOT and NOAA data (mean R2 0.71, RMSE 436 kg/ha) and with composite SAR/ASAR and NDVI models (mean R2 0.61, RMSE 402 kg/ha) using both reflectance and backscattering data. CropWatN and Composite SAR/ASAR & NDVI model mean yields were 4,754/4,170 kg/ha for wheat, 4,192/3,848 kg/ha for barley and 4,992/2,935 kg/ha for oats.

Monitoring Changes in Rice Cultivated Area from SAR and Optical Satellite Images in Ben Tre and Tra Vinh Provinces in Mekong Delta, Vietnam

The objective of this study was to obtain up-to-date information on land use and to identify long term changes in land use, especially rice, aquaculture and other crops in Ben Tre and Tra Vinh provinces in Vietnam’s Mekong Delta. Long-term changes in land-use of the study area have not been studied using long time series of SAR and optical Earth observation (EO) data before. EO data from 1979–2012 was used: ENVISAT ASAR Wide Swath Mode, SPOT and Landsat imagery. An unsupervised ISODATA classification was performed on multitemporal SAR images. The results were validated using ground truth data. Using the Synthetic Aperture Radar (SAR) imagery maps for 2005, 2009 and 2011 were obtained. Different rice crops, aquaculture and fruit trees could be distinguished with an overall accuracy of 80%. Using available optical imagery the time series was extended from 2005 to 1979. Long-term decrease in the rice acreage and increase in the aquaculture acreage could be detected.

Prediction of Forest Stand Attributes Using TerraSAR-X Stereo Imagery

Consistent, detailed and up-to-date forest resource information is required for allocation of forestry activities and national and international reporting obligations. We evaluated the forest stand attribute prediction accuracy when radargrammetry was used to derive height information from TerraSAR-X stereo imagery. Radargrammetric elevations were normalized to heights above ground using an airborne laser scanning (ALS)-derived digital terrain model (DTM). Derived height metrics were used as predictors in the most similar neighbor (MSN) estimation approach. In total, 207 field measured plots were used in MSN estimation, and the obtained results were validated using 94 stands with an average area of 4.1 ha. The relative root mean square errors for Loreys height, basal area, stem volume, and above-ground biomass were 6.7% (1.1 m), 12.0% (2.9 m 2 /ha), 16.3% (31.1 m 3 /ha), and 16.1% (15.6 t/ha). Although the prediction accuracies were promising, it should be noted that the predictions included bias. The respective biases were −4.6% (−0.7 m), −6.4% (−1.6 m 2 /ha), −9.3% (−17.8 m 3 /ha), and −9.5% (−9.1 t/ha). With detailed DTM, TerraSAR-X stereo radargrammetry-derived forest information

Integrating Vegetation Indices Models and Phenological Classification with Composite SAR and Optical Data for Cereal Yield Estimation in Finland (Part I)

During 1996–2006 the Ministry of Agriculture and Forestry in Finland, MTT Agrifood Research Finland and the Finnish Geodetic Institute carried out a joint remote sensing satellite research project. It evaluated the applicability of composite multispectral SAR and optical satellite data for cereal yield estimations in the annual crop inventory program. Three Vegetation Indices models (VGI, Infrared polynomial, NDVI and Composite multispetral SAR and NDVI) were validated to estimate cereal yield levels using solely optical and SAR satellite data (Composite Minimum Dataset). The average R2 for cereal yield (yb) was 0.627. The averaged composite SAR modeled grain yield level was 3,750 kg/ha (RMSE = 10.3%, 387 kg/ha) for high latitude spring cereals (4,018 kg/ha for spring wheat, 4,037 kg/ha for barley and 3,151 kg/ha for oats).

A Comparison of Precise Leveling and Persistent Scatterer SAR Interferometry for Building Subsidence Rate Measurement

It is well known that the most accurate method to detect changes of height is the geodetic precise leveling method. Due to the high demand work and time needed for precise leveling alternative methods are studied to obtain high quality height information. Differential SAR interferometry techniques such as the Persistent Scatterer Interferometry (PSI) method are studied to detect millimeter level deformations in urban areas. Additionally, SAR analysis will provide spatially extensive information on subsidence. On the other hand, PSI subsidence rates have not yet been comprehensively compared to the precise leveling measurements of the subsidence of individual buildings. Typically subsidence rates are interpolated to a continuous spatial surface, but in this study, spatially discontinuous subsidence was measured for a set of individual buildings. Therefore, we conducted three precise leveling campaigns and measured in total 82 geodetic-grade bolts, which were tightly attached to the building foundations. Moreover, we used additional leveling data (obtained from the local authorities), which contained long time series of leveling data for individual buildings. In the present study, ERS and ENVISAT satellite SAR data were processed using a PSI algorithm and the results were compared to leveling data of individual buildings.

Urban change detection in the Helsinki metropolitan region using Radarsat-1 fine beam SAR images

This paper describes the potential of Radarsat-1 fine beam intensity SAR images in urban mapping and change detection. The main objective was to determine the possibility of building detection e.g. as a function of height, orientation, roof material and roof type in the Helsinki metropolitan region. Altogether, 22235 buildings were used to estimate the effect of height, and then 14270 buildings were used to estimate the effect of orientation. Additionally, a smaller and more detained set of buildings was selected from various environments e.g. from forested and dense urban areas. It was shown that height and orientation were the main factors affecting on the detectability of the buildings. The results indicate that approximately 39% of the buildings can be detected using the Radarsat-1 SAR; densely build up areas are clearly visible in Radarsat fine beam images, but in small-house areas buildings that are surrounded by forest are visible only in very rare cases.