Photogrammetry, together with radar interferometry, is the most popular of the remote sensing techniques used to monitor stability of high mountain slopes. By using two images of an area taken from different view angles, photogrammetry produces digital terrain models (DTM) and orthoprojected images. Repeat digital terrain models (DTM) are differenced to compute elevation changes. Repeat orthoimages are matched to compute the horizontal displacement and deformation of the masses. The success of the photogrammetric approach in the computation of horizontal displacement (and also the generation of DTM through parallax matching, although not covered in this work) greatly relies on the success of image matching techniques. The area-based image matching technique with the normalized cross-correlation (NCC) as its similarity measure is widely used in mass movement analysis. This method has some limitations that reduce its precision and reliability compared to its theoretical potential. The precision with which the matching position is located is limited to the pixel size unless some sub-pixel precision procedures are applied. The NCC is only reliable in cases where there is no significant deformation except shift in position. Identification of a matching entity that contains optimum signal-to-noise ratio (SNR) and minimum geometric distortion at each location has always been challenging. Deformation parameters such as strains can only be computed from the inter-template displacement gradient in a post-matching process. To find appropriate solutions for the mentioned limitations, the following investigations were made on three different types of mass movements; namely, glacier flow, rockglacier creep and land sliding. The effects of ground pixel size on the accuracy of the computed mass movement parameters such as displacement were investigated. Different sub-pixel precision algorithms were implemented and evaluated to identify the most precise and reliable algorithm. In one approach images are interpolated to higher spatial resolution prior to matching. In another approach the NCC correlation surface is interpolated to higher resolution so that the location of the correlation peak is more precise. In yet another approach the position of the NCC peak is computed by fitting 2D Gaussian and parabolic curves to the correlation peak turn by turn. The results show that the mean error in metric unit increases linearly with the ground pixel size being about half a pixel at each resolution. The proportion of undetected moving masse increases with ground pixel size depending on the displacement magnitudes. Proportion of mismatching templates increases with increasing ground pixel size depending on the noise content, i.e. temporal difference, of the image pairs. Of the sub-pixel precision algorithms, interpolating the image to higher resolution using bi-cubic convolution prior to matching performs best. For example, by increasing the spatial resolution (i.e. reducing the ground pixel size) of the matched images by 2 to 16 times using intensity interpolation, 40% to 80% of the performances of the same resolution original image can be achieved. A new spatially adaptive algorithm that defines the template sizes by optimizing the SNR, minimizing the geometric distortion and optimizing the similarity measure was also devised, implemented and evaluated on aerial and satellite images of mass movements. The algorithm can also exclude ambiguous and occluded entities from the matching. The evaluation of the algorithm was conducted on simulated deformation images and in relation to the image-wide fixed template sizes ranging from 11 to 101 pixels. The evaluation of the algorithm on the real mass movements is conducted by a novel technique of reconstructing the reference image from the deformed image and computing the global correlation coefficient and the corresponding SNR between the reference and the reconstructed image. The results show that the algorithm could reduce the error of displacement estimation by up to over 90% (in the simulated case) and improve the SNR of the matching by up to over 4 times compared to the globally fixed template sizes. The algorithm pushes terrain displacement measurement from repeat images one step forward towards full automation. The least squares image matching (LSM) matches images precisely by modeling both the geometric and radiometric deformation. The potential of the LSM is not fully utilized for mass movement analysis. Here, the procedures with which horizontal surface displacement, rotation and strain rates of glacier flow, rockglacier creep and land sliding are computed from the spatial transformation parameters of LSM automatically during the matching are implemented and evaluated. The results show that the approach computes longitudinal strain rates, transverse strain rates and shear strain rates reliably with mean absolute deviation in the order of 10-4 as evaluated on stable grounds. The LSM also improves the accuracy of displacement estimation of the NCC by about 90% in ideal (simulated) case and the SNR of the matching by about 25% in real multi-temporal images of mass movements. Additionally, advanced spatial transformation models such as projective and second degree polynomial are used for the first time for mass movement analysis in addition to the affine. They are also adapted spatially based on the minimization of the sum of square deviation between the matching templates. The spatially adaptive approach produces the best matching, closely followed by the second-order polynomial. Affine and projective models show similar results closely following the two approaches. In the case of the spatially adaptive approach, over 60% of the entities matched for the rockglacier and the landslide are best fit by the second-order polynomial model. In general, the NCC alone may be sufficient for low resolution images of moving masses with limited or no deformation. To gain...

Nation-wide Sentinel-2 mosaics were used with National Forest Inventory (NFI) plot data for modelling and subsequent mapping of spruce-, pine-, and deciduous-dominated forest in Norway at a 16 m × 16 m resolution. The accuracies of the best model ranged between 74% for spruce and 87% for deciduous forest. An overall accuracy of 90% was found on stand level using independent data from more than 42 000 stands. Errors mostly resulting from a forest mask reduced the model accuracies by ∼10%. The produced map was subsequently used to generate model-assisted (MA) and poststratified (PS) estimates of species-specific forest area. At the national level, efficiencies of the estimates increased by 20% to 50% for MA and up to 90% for PS. Greater minimum numbers of observations constrained the use of PS. For MA estimates of municipalities, efficiencies improved by up to a factor of 8 but were sometimes also less than 1. PS estimates were always equally as or more precise than direct and MA estimates but were applicable in fewer municipalities. The tree species prediction map is part of the Norwegian forest resource map and is used, among others, to improve maps of other variables of interest such as timber volume and biomass.

This paper describes the development and utility of the Norwegian forest resources map (SR16). SR16 is developed using photogrammetric point cloud data with ground plots from the Norwegian National Forest Inventory (NFI). First, an existing forest mask was updated with object-based image analysis methods. Evaluation against the NFI forest definitions showed Cohen’s kappa of 0.80 and accuracy of 0.91 in the lowlands and a kappa of 0.73 and an accuracy of 0.96 in the mountains. Within the updated forest mask, a 16×16 m raster map was developed with Lorey’s height, volume, biomass, and tree species as attributes (SR16-raster). All attributes were predicted with generalized linear models that explained about 70% of the observed variation and had relative RMSEs of about 50%. SR16-raster was segmented into stand-like polygons that are relatively homogenous in respect to tree species, volume, site index, and Lorey’s height (SR16-vector). When SR16 was utilized in a combination with the NFI plots and a model-assisted estimator, the precision was on average 2-3 times higher than estimates based on field data only. In conclusion, SR16 is useful for improved estimates from the Norwegian NFI at various scales. The mapped products may be useful as additional information in Forest Management Inventories.

Displacement and deformation are fundamental measures of Earth surface mass movements such as glacier flow, rockglacier creep and rockslides. Ground-based methods of monitoring such mass movements can be costly, time consuming and limited in spatial and temporal coverage. Remote sensing techniques, here matching of repeat optical images, are increasingly used to obtain displacement and deformation fields. Strain rates are usually computed in a post-processing step based on the gradients of the measured velocity field. This study explores the potential of automatically and directly computing velocity, rotation and strain rates on Earth surface mass movements simultaneously from the matching positions and the parameters of the geometric transformation models using the least squares matching (LSM) approach. The procedures are exemplified using bi-temporal high resolution satellite and aerial images of glacier flow, rockglacier creep and land sliding. The results show that LSM matches the images and computes longitudinal strain rates, transverse strain rates and shear strain rates reliably with mean absolute deviations in the order of 10⁻⁴ (one level of significance below the measured values) as evaluated on stable grounds. The LSM also improves the accuracy of displacement estimation of the pixel-precision normalized cross-correlation by over 90% under ideal (simulated) circumstances and by about 25% for real multi-temporal images of mass movements.

The size and location of agricultural fields that are in active use and the type of use during the growing season are among the vital information that is needed for the careful planning and forecasting of agricultural production at national and regional scales. In areas where such data are not readily available, an independent seasonal monitoring method is needed. Remote sensing is a widely used tool to map land use types, although there are some limitations that can partly be circumvented by using, among others, multiple observations, careful feature selection and appropriate analysis methods. Here, we used Sentinel-2 satellite image time series (SITS) over the land area of Norway to map three agricultural land use classes: cereal crops, fodder crops (grass) and unused areas. The Multilayer Perceptron (MLP) and two variants of the Convolutional Neural Network (CNN), are implemented on SITS data of four different temporal resolutions. These enabled us to compare twelve model-dataset combinations to identify the model-dataset combination that results in the most accurate predictions. The CNN is implemented in the spectral and temporal dimensions instead of the conventional spatial dimension. Rather than using existing deep learning architectures, an autotuning procedure is implemented so that the model hyperparameters are empirically optimized during the training. The results obtained on held-out test data show that up to 94% overall accuracy and 90% Cohen’s Kappa can be obtained when the 2D CNN is applied on the SITS data with a temporal resolution of 7 days. This is closely followed by the 1D CNN on the same dataset. However, the latter performs better than the former in predicting data outside the training set. It is further observed that cereal is predicted with the highest accuracy, followed by grass. Predicting the unused areas has been found to be difficult as there is no distinct surface condition that is common for all unused areas.