Browsing by Subject "Lidar"

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    2015 Twin Cities Metropolitan Area Urban Tree Canopy Assessment
    (2017-01-03) Knight, Joe F; Rampi, Lian P; Host, Trevor K; jknight@umn.edu; Knight, Joseph, F
    A high-resolution (1-meter) tree canopy assessment was completed for the Twin Cities Metropolitan Area. Mapping of existing and potential tree canopy is critical for urban tree management at the landscape level. This classification was created from combined 2015 aerial imagery, LIDAR data, and ancillary thematic layers. These data sets were integrated using an Object-Based Image Analysis (OBIA) approach through multi-resolution image segmentation and an iterative set of classification commands in the form of customized rulesets. eCognition® Developer was used to develop the rulesets and produce raster classification products for TCMA. The results were evaluated using randomly placed and independent verified assessment points. The classification product was analyzed at regional scales to compare distributions of tree canopy spatially and at different resolutions. The combination of spectral data and LiDAR through an OBIA method helped to improve the overall accuracy results providing more aesthetically pleasing maps of tree canopy with highly accurate results.
  • Thumbnail Image
    Item
    Evaluating state-of-the-art remotely sensed data and methods for mapping wetlands in Minnesota
    (2013-12) Rampi, Lian Pamela
    Appropriate management of our natural resources requires constant improvement and update of natural resource inventories. Remote sensing data and techniques offer an effective way to map and estimate changes in our current natural resources. The research presented in this dissertation will demonstrate state-of-the art remote sensing based methods for mapping natural and man-made features, including wetlands, general land cover, and building footprints. High resolution remotely sensed data used in this research included: lidar (light detection and ranging) data (low and high lidar posting density) and multispectral (NIR, blue, green and red bands) leaf-off aerial imagery.This research examined high resolution lidar data through the evaluation of various lidar posting densities and their influence on the accuracy of building footprints and DEMs. The lidar DEM analysis was extended by creating a Compound Topographic Index (CTI) from the DEM to evaluate the potential of the CTI's information for identifying wetland's location. Finally, the results from the second chapter were integrated into the third chapter by combining CTI, high resolution imagery, Digital Surface Model (DSM) and lidar intensity for mapping four land cover classes, including: wetlands, urban, agricultural and forest. A state-of-the-art remote sensing technique known as Object-Based Image Analysis (OBIA) was used to integrate lidar derived products and high resolution imagery. Results and findings of this research are important in two ways: First, advancing the understanding of lidar and lidar derivatives for mapping natural and manmade landscape features. Second, providing needed information to the scientific and civilian community, particularly in the state of Minnesota, to help with the process of updating wetland inventories such as the NWI and increasing the accuracy of mapping wetlands efforts with state-of-the-art techniques.
  • Thumbnail Image
    Item
    Minnesota Land Cover Classification and Impervious Surface Area by Landsat and Lidar: 2013-14 Update
    (2016-08-03) Rampi, Lian P; Knight, Joe F; Bauer, Marvin; jknight@umn.edu; Knight, Joe F; Remote Sensing and Geospatial Analysis Laboratory, University of Minnesota
    This is a 15-meter raster dataset of a land cover and impervious surface classification for 2013-14, level two classification. The classification was created using a combination of multitemporal Landsat 8 data and LiDAR data with Object-based image analysis. By using objects instead of pixels we were able to utilize multispectral data along with spatial and contextual information of objects such as shape, size, texture and LiDAR-derived metrics to distinguish different land cover types. While OBIA has become the standard procedure for classification of high resolution imagery we found that it works equally well with Landsat imagery. For the objects classified as urban or developed, a regression model relating the Landsat greenness variable to percent impervious was developed to estimate and map the percent impervious surface area at the pixel level.
  • Thumbnail Image
    Item
    On the geometric and statistical signature of landscape forming processes.
    (2009-12) Passalacqua, Paola
    The physical processes that shape landscapes are complex and involve the interaction of water, soil, vegetation and biota over a range of scales. Yet, as complex as these interactions may seem to be, the landscapes we see around us often exhibit striking hierarchical order and geometric/statistical properties that are subject to scale renormalization. The overall goal of this research is to contribute to the theoretical foundation of extracting geomorphic features of interest from high resolution topography and use the extracted features for process understanding and for advancing landscape evolution modeling. Specifically, an advanced methodology for geomorphic feature extraction is developed and implemented on several high resolution data sets of different characteristics, from a steep and landslide-dissected basin, to a mountainous region, to a flat and partly artificially drained area. This new methodology incorporates nonlinear diffusion for the pre-processing of the data, both to focus the analysis on the scales of interest and to enhance features that are critical to the network extraction. Following this pre-processing, channels are defined as curves of minimal effort, or geodesics, where the effort is measured based on fundamental geomorphological characteristics such as flow accumulation and iso-height contours curvature. The developed channel network extraction methodology is compared in terms of performance to a previously proposed channel extraction methodology based on wavelets. The results show that the geometric nonlinear framework is more computationally efficient and achieves better localization and robust extraction of features, especially in areas where gentle slopes prevail. The automatic extraction of channel morphology, such as cross-section, banks location, water surface elevation, is also addressed, as well as the possibility of distinguishing the signature of natural features such as channels from the one of artificial features, such as drainage ditches. One motivation for extracting detailed geomorphic features from landscapes is the premise that this will lead to improved process understanding (e.g., by relating the observed form to physical processes that gave rise to that form) and improved modeling (e.g., incorporate important localized features in hydrologic or sediment transport models or develop sophisticated metrics for testing the performance of landscape evolution models). With this premise in mind, work herein presents preliminary results along a particular new direction related to geomorphic transport laws and landscape evolution modeling. Specifically, motivated by: (a) our experience that geomorphic attributes, such as slope and curvature, are scale-dependent and thus the resulting sediment flux computed from the typical transport laws would also be scale-dependent, and (b) that landscapes present heterogeneity over a large range of scales, we put forward the idea of a non-local sediment flux formulation to be explored in landscape evolution modeling. Along these lines, a simple landscape evolution model is proposed and its ability to reproduce some common statistical properties of real landscapes is examined.
  • Thumbnail Image
    Item
    Provably Learning From Data: New Algorithms And Models For Matrix And Tensor Decompositions
    (2019-09) Rambhalta, Sirisha
    Learning and leveraging patterns in data has fueled the recent advances in data driven services. As these solutions become more ubiquitous, and get incorporated into critical applications in healthcare and transportation, there is an increasing need to understand the limits of these learning algorithms and to develop algorithms with guarantees. Moreover, with data being generated at unprecedented rates, these algorithms need to be fast, learn on-the-fly (online), handle large volumes of data (scalable), and be computationally efficient, while possessing guarantees on their behavior. Furthermore, to make the learning-based products widely applicable there is also a need to make their reasoning and decision making process transparent (interpretable). These challenges inspire and motivate this dissertation. Specifically, we focus on analyzing various matrix/tensor demixing and factorization tasks, where we leverage the inherent interpretability endowed by the structure of problem (such as sparsity and low-rankness) to characterize the (theoretical) conditions for successful recovery, and analyze their performance in real-world settings. To this end, we make contributions on three fronts. First, we develop algorithm-aware theoretical guarantees for sparse matrix and tensor factorization tasks. Second, we establish algorithm-agnostic theoretical results for matrix demixing models and demonstrate their applications on real-world datasets. Lastly, we develop application-specific techniques for navigation and source separation. Bringing together Algorithms, Theory, and Applications, the techniques and theoretical results developed as part of this dissertation facilitate and motivate future explorations into the inner workings of learning algorithms for their safe use in critical applications.