Browsing by Subject "remote sensing"
Now showing items 21-25 of 25
-
(2022)Airborne LiDAR (Light Detection And Ranging) produces high-resolution and cost-efficient 3D data. Currently, forest inventories combine the use of both LiDAR and passive imaging by cameras, and the possibility of using LiDAR only is very tempting as it would lead to cost reduction. Focus of this study is on the full-waveform observations that extent the information content compared to conventional point clouds and are somewhat rarer to have access to. This study explores basic dependencies between structural canopy features and LiDAR signals over time and aims at augmenting our understanding of LiDAR-vegetation interactions and factors limiting our current ability to use pulsed LiDAR data for species detection, and how possibilities to overcome those limitations. Motivation is to understand how different waveform features can be interpreted and how the features behave over time with changing vegetation phenology. The study material consists of three consecutive LiDAR campaigns and aerial imaging surveys done in the area during a 38-month period and field reference trees that have been measured during this period. I use multi-temporal data that comprise three repeated acquisitions, which all applied same sensor, trajectories, as well as sensor and acquisition settings. As I had repeated LiDAR observations of the same trees where the acquisition settings are comparable, I could study the so-called ‘tree effect’ and overall co-variation between waveform features in the repeated acquisitions. Phenological changes are available as the data comprises winter (leaf-off), early summer (low LAI in conifers) and late summer data (full leaf, high LAI). The influence of scan zenith angle (SZA) on waveform features and attributes is also considered, as the same tree can be seen from multiple strips. The results showed that by using careful experimentation it is possible to detect intra- and interspecies phenological changes from multitemporal full-waveform data, while SZA did not have markable effect on the WF features. I was also able to perform well with the tree species classification task in varying phenological conditions. The phenological changes were very apparent on deciduous trees, but rather small on evergreen conifers. In a 45-year-old stand, the overall accuracies in tree species classification were 92, 87 and 88 % for winter, early summer, and late summer, respectively. These figures were 84, 81, and 83 % for in an old growth forest. The ‘tree effect’ was shown to be significant, i.e., many of the WF features of trees were correlated over time. The intra-species feature variance that is due to the tree effect represents natural variation between trees of the same species.
-
(2020)A warming trend of annual average surface temperatures since pre-industrial times has been observed globally. High-arctic area of Svalbard, Norway is undergoing amplified change of annual average temperatures when compared to the global average. Decline of glaciers in western Svalbard has been ongoing for several decades, and in the recent past, rapid biological successions have taken place. These changes have likely had effect on regional scale carbon dynamics at Svalbard’s moss tundra areas. Possibly indicating onset of paludification process of these areas. However, palaeoecological studies from the area are scarce, and the response of high-latitude moss tundra areas to past or ongoing climate change, are still not fully understood. This thesis aimed to bring forward information of changes in recent organic matter and carbon accumulation rates at Svalbard, Norway. Soil profiles were collected from four moss tundra sites, located on coastal areas and fjords descending towards Isfjorden, on the western side of Spitsbergen island. Radiocarbon (14C) and lead (210Pb) dating methods with novel age-depth modelling and soil property analyses, were used to reconstruct recent organic matter and carbon accumulation histories from 1900 AD to 2018 AD. Accumulation histories were supported by meteorological measurements from the area. In addition, annual maximum value Normalized Difference Vegetation Indices for 1985 AD till 2018 AD period were produced, to study vegetation succession in the recent past. Lastly, possibility to predict spatiotemporal variation of soil carbon accumulation with satellite derived vegetation indices was assessed. Development from predominantly mineral soils to organic soils was distinguishable within multiple soil profiles, pointing to potential paludification. Recent apparent carbon accumulation rates showed an increasing trend. Supporting meteorological data and literature suggest that regional abiotic and biotic factors in synergy with weather and climate are contributing to this observed trend. Vegetation indices pointed to major changes in vegetation composition and productivity. However, investigation of relationship between recent carbon accumulation rates and vegetation indices did not produce reliable results. Spatiotemporal heterogeneity of carbon soil-atmosphere fluxes presently imposes large challenges for such modelling. To alleviate this problem, efforts for more efficient synergetic use of field sampling and remote sensing -based material should be undertaken, to improve modelling results.
-
(2009)The purpose of this study was to investigate the usability of aerial images and Landsat TM in estimating Scots pine defoliation. Estimation methods tested were unsupervised classification, maximum likelihood method, mixed model and linear regression model. Image features for needle loss detection were selected with stepwise linear regression and mixed model technique. As a part of the study the relationship between needle loss and leaf area index (LAI) was examined. The relationship between image features, needle loss and leaf area index was also examined. Numerical aerial images and Landsat TM satellite images were used. Textural features were calculated from aerial images and spectral vegetation indices from the satellite image. The study site was located in Ilomantsi, Finland. 71 field sample plots were measured and located with GPS. Field plots were circular plots. Trees with diameter brest height (dbh) over 13,9 cm were measured from 13 meter radius and trees with dbh 5,0 - 13,8 cm were measured from 7 meter radius. Needle loss of all pines was estimated. Needle loss for the plot was calculated as an average weighted by tree height. Four different class combinations were tested in classification. Plots were divided in 2, 3, 4 and 9 classes depending on their needle loss. Different image feature combinations and classification methods were tested. Classification was done by cross validation. Classification results were compared with original classes. The reliability of the classification was tested using accuracy matrix and kappa value. A mixed model was also used for aerial image features. The best image feature combination with all classification methods was the aerial image feature combination selected with stepwise selection method. Both spectral and textural features were included in the stepwise selection result. Classification accuracy varied between 38,0 % (9 classes) and 88,7 % (2 classes). The best explanatory variable for needle loss was the aerial image NIR channel maximum radiation (r2=0,69). However, unsupervised and supervised classification might have produced too positive results because of correlation in the data. Mixed model technique was used to select the variables for the linear model. Mixed model was used to reduce the effects of the correlation. The model classification accuracy varied between 35,2 % (9classes) and 87,3 % (2classes). According to mixed model selection result no textural features were significant predictors for needle loss. Classification results with Landsat image features were slightly poorer than with the best aerial image feature set (accuracy between 25,4 % and 88,7 %). The relationship between needle loss and LAI was poor (r2=0,27). Needle loss and LAI also correlated with different image features. LAI correlated slightly better with textural features than needle loss. Spectral vegetation indices calculated from Landsat TM correlated moderately with both needle loss and LAI. Indices VI3 (r2=0,56), MIR/NIR (r2=0,51) and RSR (r2=0,44) had the strongest connection to needle loss. Spectral vegetation indices could be a potential way for large area needle loss detection.
-
(2021)Trees face an increasing variety of health threats. The overall effects of climate change on trees and forests are difficult to predict. As a result of the warming climate, the growing season is lengthening, improving the growth of the trees, but at the same time drought and insect damages may become more common and the risk of storm damage increases. There are many benefits to monitoring tree mortality. It can be used to assess the health status of forests, productivity, carbon sequestration and the ecological impacts of dead trees on forest ecosystems. Causes leading to tree death can include biological, climatic or human related factors. Monitoring can increase understanding of the causes of death and potentially help to protect forests better. Tree-related mortality is a spatially and temporally irregular process that is difficult to monitor using traditional inventory methods. Remote sensing makes it possible to map and monitor tree mortality more effectively. The purpose of this thesis was to find out how remote sensing data can be utilized in monitoring tree mortality. The aim was to find out how tree mortality has varied regionally and quantitatively in the Central Park of Helsinki and how accurately dead trees can be identified from aerial imagery. The study period was 2005–2019, during which orthophotos of seven different years were examined. Reference data of 14 212 trees were collected from the aerial time series covering a 15-year period by visual image interpretation. The data included healthy, weakened and dead trees. Heatmap time series were created from the locations of weakened and dead trees to examine quantitative and regional variability in mortality. The average temperatures over the years as well as the rainfall were compared with the dead tree numbers and the correlations between the observations were examined. The collected reference data was also utilized in health status classifications, which were implemented using semi-automatic machine learning methods. The object of the classifications was to identify healthy, weakened and dead trees as well as possible from each other. The canopies of individual trees were delimited by canopy segments obtained from laser scanning data. From the pixels contained in the delimited canopies, image features describing individual trees were calculated. Considerable changes in tree mortality were observed. The number of dead trees at the beginning of the study period increased significantly from year 2005 to year 2009. An exceptionally dry summer in 2006 was identified as a possible reason. In the following years, the situation remained moderate, but in quantitative and regional terms, mortality was at its highest in 2017. Overall, there was an upward trend in mortality during the study period, and average annual temperatures were found correlating strongly with the number of dead trees (r=0.73). The classification accuracies of tree health status varied annually between 89–96%. The seven-year average accuracy was 93.6% with a kappa value of 0.88. The most important features in the classification were the features calculated from the blue channel, such as the maximum value of the channel (B_max), the difference between the maximum and minimum of the channel (B_range) and the skewness of the distribution (B_skew). The results of the thesis showed that tree mortality can be monitored using remote sensing data. Clear changes in the number of dead trees were observed using the time series and possible causes were identified. By identifying the causes behind rising mortality, the effects of climate change can also be better understood. Tree health status classification accuracies were at a good level and dead trees can be mapped from aerial imagery by semi-automatic methods. Based on the thesis, it can be rightly stated that changes in tree mortality can be observed with aerial imagery time series. In addition, the semi-automatic identification of dead trees from aerial imagery can be said to be accurate enough for large-scale use.
-
(2021)Trees face an increasing variety of health threats. The overall effects of climate change on trees and forests are difficult to predict. As a result of the warming climate, the growing season is lengthening, improving the growth of the trees, but at the same time drought and insect damages may become more common and the risk of storm damage increases. There are many benefits to monitoring tree mortality. It can be used to assess the health status of forests, productivity, carbon sequestration and the ecological impacts of dead trees on forest ecosystems. Causes leading to tree death can include biological, climatic or human related factors. Monitoring can increase understanding of the causes of death and potentially help to protect forests better. Tree-related mortality is a spatially and temporally irregular process that is difficult to monitor using traditional inventory methods. Remote sensing makes it possible to map and monitor tree mortality more effectively. The purpose of this thesis was to find out how remote sensing data can be utilized in monitoring tree mortality. The aim was to find out how tree mortality has varied regionally and quantitatively in the Central Park of Helsinki and how accurately dead trees can be identified from aerial imagery. The study period was 2005–2019, during which orthophotos of seven different years were examined. Reference data of 14 212 trees were collected from the aerial time series covering a 15-year period by visual image interpretation. The data included healthy, weakened and dead trees. Heatmap time series were created from the locations of weakened and dead trees to examine quantitative and regional variability in mortality. The average temperatures over the years as well as the rainfall were compared with the dead tree numbers and the correlations between the observations were examined. The collected reference data was also utilized in health status classifications, which were implemented using semi-automatic machine learning methods. The object of the classifications was to identify healthy, weakened and dead trees as well as possible from each other. The canopies of individual trees were delimited by canopy segments obtained from laser scanning data. From the pixels contained in the delimited canopies, image features describing individual trees were calculated. Considerable changes in tree mortality were observed. The number of dead trees at the beginning of the study period increased significantly from year 2005 to year 2009. An exceptionally dry summer in 2006 was identified as a possible reason. In the following years, the situation remained moderate, but in quantitative and regional terms, mortality was at its highest in 2017. Overall, there was an upward trend in mortality during the study period, and average annual temperatures were found correlating strongly with the number of dead trees (r=0.73). The classification accuracies of tree health status varied annually between 89–96%. The seven-year average accuracy was 93.6% with a kappa value of 0.88. The most important features in the classification were the features calculated from the blue channel, such as the maximum value of the channel (B_max), the difference between the maximum and minimum of the channel (B_range) and the skewness of the distribution (B_skew). The results of the thesis showed that tree mortality can be monitored using remote sensing data. Clear changes in the number of dead trees were observed using the time series and possible causes were identified. By identifying the causes behind rising mortality, the effects of climate change can also be better understood. Tree health status classification accuracies were at a good level and dead trees can be mapped from aerial imagery by semi-automatic methods. Based on the thesis, it can be rightly stated that changes in tree mortality can be observed with aerial imagery time series. In addition, the semi-automatic identification of dead trees from aerial imagery can be said to be accurate enough for large-scale use.
Now showing items 21-25 of 25