<p>Unmanned aerial vehicles (UAV) have revolutionized data collection in large scale agronomic field trials (10+ ha). Vegetative index (VI) maps derived from UAV imagery are a potential tool to characterize temporal and spatial treatment effects in a more efficient and non-destructive way compared to traditional data collection methods that require manual sampling. The overall objective of this study was to characterize and quantify maize responses to experimental treatments in field-scale research using UAV imagery. The specific objectives were: 1) to assess the performance of several VI as predictors of grain yield and to evaluate their ability to distinguish between fertilizer treatments, and the effects of removing soil and shadow background, 2) to assess the performance of VI and canopy cover fraction (CCF) as predictors of maize biomass at vegetative and reproductive growth stages under field-scale conditions, and 3) to compare the performance of VI derived from consumer-grade and multispectral sensors for predicting grain yield and identifying treatment effects. For the first objective, the results suggest that most VI were good indicators of grain yield at late vegetative and early reproductive growth stages, and that removing soil background improved the characterization of maize responses to experimental treatments. For objective two, overall, CCF was the best to predict biomass at early vegetative growth stages, while VI at reproductive growth stages. Finally, for objective three, performance of consumer-grade and multispectral derived VI were similar for predicting grain yield and identifying treatment effects.</p>

Ecological indicators are defined as quantifiable metrics that can be used to monitor the state of ecosystems and their response to environmental perturbations. In the global oceans, commonly used indicators are typically based on the presence and distribution of phytoplankton (as indexed by the concentration of chlorophyll-a [Chl-a]), which form the base of oceanic food webs. Phytoplankton phenology (the timing of phytoplankton growth) and phytoplankton size structure are particularly important ecological indicators that can be derived via ocean colour remote sensing. Phytoplankton phenology has a direct control on food availability, which subsequently impacts the survival of higher trophic levels and the structure of marine ecosystems. Meanwhile, phytoplankton size structure can be used to define the major functional groups that ultimately influence marine food web structure, biogeochemical cycling and carbon export. The Red Sea is a relatively unexplored tropical marine ecosystem, particularly in relation to its large-scale biological dynamics. In light of recent evidence of rapid regional warming, the need to monitor the response of the Red Sea to potential future ecosystem modifications is becoming more imminent. Using a combination of contemporary oceanographic tools, with an emphasis on ocean colour remote sensing, this PhD thesis attempts to validate the retrieval of phytoplankton ecological indicators in the Red Sea - specifically phytoplankton abundance, phenology and size structure. The interannual variability of both indicators and their linkages with the regional physical environment are also explored.

The Greenland and Antarctic ice sheets are losing mass and contributing to global sea-level rise. Ice shelves, floating ice attached to the margins of the ice sheets, modulate sea-level rise by restraining ice-sheet flow out towards the oceans, but are sensitive to surface melting. The formation of surface meltwater lakes on ice shelves can trigger rapid ice-shelf collapse. However, surface meltwater also flows atop ice shelves through rivers. The impact of rivers on ice-shelf stability is unknown. Previous studies of ice-shelf hydrology hypothesize that rivers mitigate the damage-potential of lakes by removing surface water off of the ice shelf, but also suggest that rivers enhance ice-shelf fracturing by incising into areas of already thin ice. This dissertation is focused on exploring the role of rivers on ice-shelf stability using remote sensing datasets, conceptual models, and Augmented Reality (AR). Focusing on ice shelves in Greenland, I present the discovery of a new ice-shelf surface hydrology feature, an ice-shelf estuary, and demonstrate its potential to weaken ice shelves. I fully document this new process on the Petermann Ice Shelf, where flow reverses at the mouth of the Petermann Estuary. This study marks the first observation of ocean water atop an ice shelf. I also document the initiation and growth of fracturing along the estuary channel, and a history of rectilinear calving events, where icebergs calve along longitudinal rivers. Based on this analysis of the Petermann Estuary, I propose a new mechanism for damaging ice shelves: estuarine weakening. I present evidence that this process also occurs on the Ryder Ice Shelf in northwest Greenland. My analysis demonstrates that the role of rivers on ice-shelf stability depends on how the river mouth evolves. If ice-shelf waterfalls at the river mouth incise to sea level and form estuaries, flow reversal will modulate water export off the shelf and maintain the damage-potential of lakes, and estuarine weakening may lead to a new mode of ice-shelf calving. By analyzing the three-dimensional (3D) structure of the Petermann and Ryder Ice Shelves and Estuaries with remote sensing and radar data, I find that basal channels are an important driver of estuary development as they dictate the linearity of surface rivers. Determining the role that basal channels play in estuary formation requires accurate and appropriate data visualization tools. I develop AR applications to visualize radar data on ice shelves, towards enabling more intuitive and sophisticated interpretation of the ice-shelf structure in 3D. Through simple conceptual modeling, I suggest that although basal channels precondition ice-shelf estuary formation, estuary formation is strongly controlled by river incision. Finally, I present a model of ice-shelf estuary formation as a function of surface and basal melting. Using this conceptual model, I predict that ice-shelf estuaries could form in Antarctica in the near future. Surface melting in Antarctica is predicted to increase in under half a century. Estuary formation in Antarctica will be accelerated by lengthening of the melt season, and estuaries may form far from the calving front if rivers intersect upstream rifts. I show that ice-shelf estuaries could evolve from ice-shelf rivers in a warming Antarctica, introducing new ice-shelf weakening mechanisms. This increases the urgency to understand and include ice-shelf estuarine processes in ice-sheet models.

<p><a>Indiana is the leading state of cover crop adoption within the Upper Mississippi River Basin. However, since 2015 the cover crop adoption has slowed to a plateau. In order to regain the previous momentum, there must be an increased understanding of the spatiotemporal dynamics of cover crop adoption on the county and watershed scale. Currently, the cover crop adoption is monitored biannually through a driving transect survey method that investigates only 8.5% of the watershed and extrapolates to the entire county. However, the observations made by the driving transect survey can merely cover limited fields and is time-consuming. In addition, the driving transect survey did not provide comparative analysis among consecutive years. Therefore, we developed a rapid cover crop survey method by using remote sensing technology. The fundamental objectives of this research are: (1) evaluating the accuracy of the rapid cover crop survey method relative to the driving transect data and determining the best cut-off value (COV) of Normalized Difference Vegetation Index (NDVI); (2) performing a hindcasting analysis of cover crop adoption within the Big Pine Creek Watersheds within the period of 2014-2018 by employing a rapid cover crop survey remote sensing techniques; (3) accessing cover crop adoption management tendencies of farmers within the Big Pine Watersheds, and (4) determining the cover crop adoption tenure of farmers within the Big Pine Creek watersheds between 2014 and 2018. The cover crop management tendency represents the farmers’ preference on cash crop rotation method after harvesting cover crops, and the cover crop adoption tenure means that how often farmers adopt cover crops in a specific field in the research period.</a></p> <p>The results of this research demonstrated that relative to the conventional driving transect, remote sensing is a feasible method to successfully detect cover crop adoption on a county and watershed scale. Over a 4-year period (2015-2018), Producer’s Accuracy (PA) under the best COV, which represented how much vegetation-covered field recorded in transect data that can be captured in the processed NDVI map, was 89.02%. This PA value was relatively high compared with previous spatial crop classification research. The rapid remote sensing method also provided individual field locations of cover crop adoption over time within the entire watershed, compared to the driving transect that only gives extrapolated average of adoption. The hindcasting analysis of cover crop adoption revealed a 74% increase in cover crop acreage in the watershed from 2014 to 2018, which equated to a 0.71% increase in land receiving cover crops among all cultivated land annually. The evaluation of farmer cover crop adoption tendencies demonstrated that over a 4-year period, cover crop adoption going into corn was 19.7% greater on average relative to before soybean. Another key finding was that the level of cover crop adoption annually in the watershed was heavily influenced by the cash crop rotation. The cover crop tenure analysis demonstrated that agricultural fields of greater cover crop tenure represented the smallest portion of the cultivated land in the watershed, where 84.2% of the watershed was void of cover crop adoption and field that received cover crops for more than 4 consecutive years represented only 1% of cultivated land.</p> <p> To conclude, we are confident that the rapid cover crop survey method could replace the traditional driving transect survey. Our findings suggest that rapid assessment methods of cover crop adoption involving processed NDVI map could help advance the effectiveness, speed, and accuracy of cover crop adoption and assessment in the state of Indiana and the entire Mississippi River Basin region.</p>

Natural gas flaring from oil production is a pervasive yet understudied environmental issue. Recently available satellite imagery of gas flares has increased public awareness and concern over the severity and ubiquity of the problem. In the US, the relatively recent combination of hydraulic fracturing and horizontal drilling sparked the shale boom, leading to hundreds of thousands of wells being drilled within a decade, often in close proximity to residential populations. A major oil state that has emerged from the shale boom is North Dakota. In 2014, state regulators introduced a policy to limit the percentage of produced gas that oil-extracting companies are allowed to flare. Like many other places where flaring takes place, flared volumes are reported by oil companies themselves. What was the effect of North Dakota’s regulation on gas flaring according to self-reported and satellite data? What was the effect of the regulation on self-reporting behavior? In such a tight oil setting, how well does the prevailing satellite product used to monitor gas flares perform? This dissertation uses new data and methodologies from several disciplines to study these important questions around gas flaring. The results find that the predominant satellite product does not perform well in the on-shore oil production context. While regulation has reduced flaring in a major oil state, the reduction is smaller than thought because of underreporting by oil well operators. Further, the underreporting is associated with political economy and corporate culture factors.

<div>Sorghum, which is grown internationally as a cereal crop that is robust to heat, drought, and disease, has numerous applications for food, forage, and biofuels. When monitoring the growth stages of sorghum, or phenotyping specific traits for plant breeding, it is important to identify and monitor the panicles in the field due to their impact relative to grain production. Several studies have focused on detecting panicles based on data acquired by RGB and multispectral remote sensing technologies. However, few experiments have included hyperspectral data because of its high dimensionality and computational requirements, even though the data provide abundant spectral information. Relative to analysis approaches, machine learning, and specifically deep learning models have the potential of accommodating the complexity of these data. In order to detect panicles in the field with different physical characteristics, such as colors and shapes, very high spectral and spatial resolution hyperspectral data were collected with a wheeled-based platform, processed, and analyzed with multiple extensions of the VGG-16 Fully Convolutional Network (FCN) semantic segmentation model.</div><div><br></div><div>In order to have correct positioning, orthorectification experiments were also conducted in the study to obtain the proper positioning of the image data acquired by the pushbroom hyperspectral camera at near range. The scale of the DSM derived from LiDAR that was used for orthorectification of the hyperspectral data was determined to be a critical issue, and the application of the Savitzky-Golay filter to the original DSM data was shown to contribute to the improved quality of the orthorectified imagery.</div><div><br></div><div>Three tuned versions of the VGG-16 FCN Deep Learning architecture were modified to accommodate the hyperspectral data: PCA&FCN, 2D-FCN, and 3D-FCN. It was concluded that all the three models can detect the late season panicles included in this study, but the end-to-end models performed better in terms of precision, recall, and the F-score metrics . Future work should focus on improving annotation strategies and the model architecture to detect different panicle varieties and to separate overlapping panicles based on an adequate quantities of training data acquired during the flowering stage.</div>

The environmental and human impacts of wildfires have grown considerably in recent years due to an increase in their frequency and coverage. Effective wildfire management and suppression requires real-time data to locate fire fronts, model their propagation and assess the impact of biomass burning. Existing empirical wildfire models are based on fuel properties and meteorological data with inadequate spatial or temporal sampling. A geosynchronous space platform with the proposed set of high resolution infrared detectors provides a unique capability to monitor fires at improved spatial and temporal resolutions. The proposed system is feasible with state-of-the-art hardware and software for high sensitivity fire detection at saturation levels exceeding active flame temperatures. Ground resolutions of 100 meters per pixel can be achieved with repeat cycles less than one minute. Atmospheric transmission in the presence of clouds and smoke is considered. Modeling results suggest fire detection is possible through thin clouds and smoke. A semi-empirical cellular automata model based on theoretical elliptical spread shapes is introduced to predict wildfire propagation using detected fire front location and spread rate. Model accuracy compares favorably with real fire events and correlates within 2% of theoretical ellipse shapes. This propagation modeling approach could replace existing operational systems based on complex partial differential equations. The baseline geosynchronous fire detection system supplemented with a discrete-based propagation model has the potential to save lives and property in the otherwise uncertain and complex field of fire management.

With an increasing urban population in Sweden, expecting to reach 90% by 2050 (UN World Urbanization prospects, The 2011 Revision), this high level of urban population put pressure on functioning infrastructure, sufficient housing and need to monitor the environmental effects such as pollution and the effects of land use change. Stockholm County currently holds 22% of the population and accounts for nearly half of the urban growth in Sweden (Svensk Handelskammare).   Previous research on change detection using remote sensing cover the use of data sets from optical sensors, infrared spectrum, radar data and the use of additional derived data sets such as indices and texture measure (implemented on pixel or feature level). There is not yet any consensus regarding which change detection methods that is superior to others. Comparative studies often only test a few algorithms on one particular data set. Change detection of Stockholm urban area has not been well investigated in previous literature. This thesis is focused on a change detection analysis of Stockholm area between 1986 and 2006 using remote sensing data fusion. The data set used is SPOT-1 HRV XS data at 20m resolution from 1986, SPOT-1 HRV Panchromatic data at 10m resolution from 1987 and SPOT-5 HRG XS data of 10m resolution from 2006. The first challenge was to fuse the multispectral and panchromatic images from 1986 and 1987 to inject the details of the 10m panchromatic image into the 20m multispectral so that the resulting images will have similar spatial details as the 2006 images. This was done by wavelet transform. Haar, Daubechies, Coiflet and Biorthogonal wavelet families were tested to find the optimal fusion and the corresponding parameters. The results showed that the Daubechies, Coiflet and Biorthogonal families did not differ significantly and that for this data set and analysis purpose more than one wavelet family fusion results showed satisfactory results. The correlation coefficient for these three families was all over 0,96 at decomposition level two.   Then change detection was performed using change vector analysis (CVA) and a supervised non-parametric classifier. A comparison is made between two inputs: one using only spectral information and the other adding textural information to the spectral information. The change detection analysis was undertaken in three steps: calculating texture measures from the original images, calculating change magnitude using Change Vector Analysis (CVA) and classifying change from no-change using Support Vector Machine (SVM). Three GLCM texture measures were chosen: Homogeneity, Mean and Entropy in the change detection analysis. These, as well as the spectral information, were input for change vector magnitude. Then SVM is used to classify changed pixels from no-change pixels. Two change results were obtained, the first using only spectral information, and the other using both spectral and textural information. The overall accuracy using only spectral information was rather high at 87, 86%. But the visual inspections indicate that using only spectral change magnitude is not sufficient for a good change detection result because there is an apparent overestimation of change. When adding the textural information the overall accuracy increase drastically to 97,01%, although at visual inspection there seem to be an underestimation of change.  Because of the high overall accuracy an independent validation was made causing the overall accuracy and kappa to decrease. Change detection using only multispectral data got an overall accuracy of 76, 12% and kappa coefficient 0,53. For change detection result with added texture measures the overall accuracy became 85,80%  and 0,72.  The results further confirm the general advantages using texture measure although the independent evaluation resulted in a lower accuracy than the author's evaluations.

Image classification based on remotely sensed data is the primary domain of automatic mapping research. With the increasing of urban development, keeping geographic database updating is imminently needed. Automatic mapping of land cover types in urban area is one of the most challenging problems in remote sensing. Traditional database updating is time consuming and costly. It has usually been performed by manual observation and visual interpretation, In order to improve the efficiency as well as the accuracy, new technique in the data collection and extraction becomes increasingly necessary. This paper studied an object-based decision tree classification based on orthophoto and lidar data, both alone and integrated. Four land cover types i.e. Forest, Water, Openland as well as Building were successfully extracted. Promising results were obtained with the 89.2% accuracy of orthophoto based classification and 88.6% accuracy of lidar data based classification. Both lidar data and orthophoto showed enough capacity to classify general land cover types alone. Meanwhile, the combination of orthophoto and lidar data demonstrated a prominent classification results with 95.2% accuracy. The results of integrated data revealed a very high agreement. Comparing the process of using orthophoto or lidar data alone, it reduced the complexity of land cover type discrimination. In addition, another classification algorithm, support vector machines (SVM) classification was preformed. Comparing to the decision tree classification, it obtained the same accuracy level as decision tree classification in orthophoto dataset (89.2%) and integration dataset (97.3%). However, the SVM results of lidar dataset was not satisfactory. Its overall accuracy only reached 77.1%. In brief, object-based land cover classification demonstrated its effectiveness in land cover map generation. It could exploit spectral and spatial features from input data efficiently and classifying image with high accuracy.

The Global L-band Observatory for Water Cycle Studies (GLOWS) is designed as a follow-on to the Soil Moisture Active Passive (SMAP) observatory launched in 2015. While GLOWS is essentially copying many aspects of the SMAP mission, a key change has been made in the antenna technology. SMAP uses a reflector antenna and to reduce mission costs GLOWS uses a metamaterial lens antenna. This type of antenna is less efficient, so it must be proven that GLOWS can achieve the same uncertainty levels in soil moisture measurements as SMAP. In this work, a unified framework for modeling and analyzing GLOWS' ability to meet all mission and measurement requirements is developed. A model for the uncertainty effects of the lens antenna is developed and used to show that so long as the lens efficiency is above a threshold determined by the accuracy of the lens physical temperature knowledge, GLOWS will also be able to achieve all measurement requirements. It is shown that GLOWS is able to copy the design parameters of SMAP and achieve the same mission requirements.