Mapping the Earth’s surface and its rapid changes with remotely sensed data is a crucial tool to un-derstand the impact of an increasingly urban world population on the environment. However, the impressive amount of freely available Copernicus data is only marginally exploited in common clas-sifications. One of the reasons is that measuring the properties of training samples, the so-called ‘fea-tures’, is costly and tedious. Furthermore, handling large feature sets is not easy in most image clas-sification software. This often leads to the manual choice of few, allegedly promising features. In this Master’s thesis degree project, I use the computational power of Google Earth Engine and Google Cloud Platform to generate an oversized feature set in which I explore feature importance and analyse the influence of dimensionality reduction methods. I use Support Vector Machines (SVMs) for object-based classification of satellite images - a commonly used method. A large feature set is evaluated to find the most relevant features to discriminate the classes and thereby contribute most to high clas-sification accuracy. In doing so, one can bypass the sensitive knowledge-based but sometimes arbi-trary selection of input features.Two kinds of dimensionality reduction methods are investigated. The feature extraction methods, Linear Discriminant Analysis (LDA) and Independent Component Analysis (ICA), which transform the original feature space into a projected space of lower dimensionality. And the filter-based feature selection methods, chi-squared test, mutual information and Fisher-criterion, which rank and filter the features according to a chosen statistic. I compare these methods against the default SVM in terms of classification accuracy and computational performance. The classification accuracy is measured in overall accuracy, prediction stability, inter-rater agreement and the sensitivity to training set sizes. The computational performance is measured in the decrease in training and prediction times and the compression factor of the input data. I conclude on the best performing classifier with the most effec-tive feature set based on this analysis.In a case study of mapping urban land cover in Stockholm, Sweden, based on multitemporal stacks of Sentinel-1 and Sentinel-2 imagery, I demonstrate the integration of Google Earth Engine and Google Cloud Platform for an optimised supervised land cover classification. I use dimensionality reduction methods provided in the open source scikit-learn library and show how they can improve classification accuracy and reduce the data load. At the same time, this project gives an indication of how the exploitation of big earth observation data can be approached in a cloud computing environ-ment.The preliminary results highlighted the effectiveness and necessity of dimensionality reduction methods but also strengthened the need for inter-comparable object-based land cover classification benchmarks to fully assess the quality of the derived products. To facilitate this need and encourage further research, I plan to publish the datasets (i.e. imagery, training and test data) and provide access to the developed Google Earth Engine and Python scripts as Free and Open Source Software (FOSS). / Kartläggning av jordens yta och dess snabba förändringar med fjärranalyserad data är ett viktigt verktyg för att förstå effekterna av en alltmer urban världsbefolkning har på miljön. Den imponerande mängden jordobservationsdata som är fritt och öppet tillgänglig idag utnyttjas dock endast marginellt i klassifikationer. Att hantera ett set av många variabler är inte lätt i standardprogram för bildklassificering. Detta leder ofta till manuellt val av få, antagligen lovande variabler. I det här arbetet använde jag Google Earth Engines och Google Cloud Platforms beräkningsstyrkan för att skapa ett överdimensionerat set av variabler i vilket jag undersöker variablernas betydelse och analyserar påverkan av dimensionsreducering. Jag använde stödvektormaskiner (SVM) för objektbaserad klassificering av segmenterade satellitbilder – en vanlig metod inom fjärranalys. Ett stort antal variabler utvärderas för att hitta de viktigaste och mest relevanta för att diskriminera klasserna och vilka därigenom mest bidrar till klassifikationens exakthet. Genom detta slipper man det känsliga kunskapsbaserade men ibland godtyckliga urvalet av variabler.Två typer av dimensionsreduceringsmetoder tillämpades. Å ena sidan är det extraktionsmetoder, Linjär diskriminantanalys (LDA) och oberoende komponentanalys (ICA), som omvandlar de ursprungliga variablers rum till ett projicerat rum med färre dimensioner. Å andra sidan är det filterbaserade selektionsmetoder, chi-två-test, ömsesidig information och Fisher-kriterium, som rangordnar och filtrerar variablerna enligt deras förmåga att diskriminera klasserna. Jag utvärderade dessa metoder mot standard SVM när det gäller exakthet och beräkningsmässiga prestanda.I en fallstudie av en marktäckeskarta över Stockholm, baserat på Sentinel-1 och Sentinel-2-bilder, demonstrerade jag integrationen av Google Earth Engine och Google Cloud Platform för en optimerad övervakad marktäckesklassifikation. Jag använde dimensionsreduceringsmetoder som tillhandahålls i open source scikit-learn-biblioteket och visade hur de kan förbättra klassificeringsexaktheten och minska databelastningen. Samtidigt gav detta projekt en indikation på hur utnyttjandet av stora jordobservationsdata kan nås i en molntjänstmiljö.Resultaten visar att dimensionsreducering är effektiv och nödvändig. Men resultaten stärker också behovet av ett jämförbart riktmärke för objektbaserad klassificering av marktäcket för att fullständigt och självständigt bedöma kvaliteten på de härledda produkterna. Som ett första steg för att möta detta behov och för att uppmuntra till ytterligare forskning publicerade jag dataseten och ger tillgång till källkoderna i Google Earth Engine och Python-skript som jag utvecklade i denna avhandling.

No description available.

Agriculture and food security are interlinked. New technologies and instruments are making the agricultural system easy to operate and increasing the food production. Remote sensing technology is widely used as a non-destructive method for crop growth monitoring, climate analysis, and forecasting crop yield. The objectives of this study are to (1) monitor crop growth remotely, (2) identify climate impacts on crop yield, and (3) forecasting crop yield. This study proposed methods to improve crop growth monitoring and yield predictions by using remote sensing technology. In this study, we developed crop vegetative growth metrics (VGM) from the MODIS (Moderate Resolution Imaging Spectroradiometer) 250m NDVI (Normalized Difference Vegetation Index) and EVI (Enhanced Vegetation Index) data. We developed 19 NDVI and EVI based VGM metrics for soybean crop from a time series of 2000 to 2018, but the methods are applicable to other crops as well. We found VGMmax, VGM70, VGM85, VGM98T are about 95% crop yield predictable. However, these metrics are independent of climatic events. We modelled the climatic impacts on soybean crop from the time series data from1980-2019 collected from NOAA's National Climatic Data Center (NCDC). Therefore, we estimated the impacts of increase and decrease of temperature (maximum, mean, and minimum) and precipitation (average) pattern on crop yields which will be helpful to monitor climate change impacts on crop production. Lastly, we made crop yield forecasting statistical model across different climatic regions in USA using Google Earth Engine. We used remotely sensed MODIS Terra surface reflectance 8-day global 250m data to calculate VGM metrics (e.g. VGM70, VGM85, VGM98T, VGM120, VGMmean, and VGMmax), MODIS Terra land surface temperature and Emissivity 8-Day data for average day-time and night-time temperature and CHIRPS (Climate Hazards Group Infra-red Precipitation with station data) data for precipitation, from a time series data of 2000-2019. Our predicted models showed a NMPE (Normalized Mean Prediction error) with in a range of -0.002 to 0.007. These models will be helpful to get an overall estimate of crop production and aid in national agricultural strategic planning. Overall, this study will benefit farmers, researchers, and management system of U.S. Department of Agriculture (USDA).

Hul’qumi’num communities on south eastern Vancouver Island have concerns about the status and safety of marine foods potentially impacted by environmental change and the urbanization and industrialization of their territories. Collaborative research undertaken with the Hul’q’umi’num’ Lands and Resources Society is part of a broader effort to revitalize cultural practices, language, and food systems. Lhuq’us (the Hul’q’umi’num’ language term for pohrpyra/pyropia spp. (commonly known as red laver or black gold)) is a flavourful and nutritious intertidal seaweed that grows on rocky beaches across the Pacific Northwest. Hul’q’umi’num’ language, cultural values, teachings, and family histories are all interwoven into the harvesting and consumption of lhuq’us in Hul’qumi’num territories. Lhuq’us is one of the species that have been persistently mentioned in conversations with state regulatory agencies and though these concerns have been raised for at least two decades there has been no systematic monitoring of the species. There are two broad streams of inquiry taken by thesis thesis. The first, employing ethnographic methodology including interviews and observant participation, seeks to both document the cultural values, oral histories, lived experiences associated with lhuq’us as well as concerns for the future collaborators have for lhuq’us and lhuq’us beaches. The second stream, based in a geographic approach, asks whether Unoccupied Aerial Vehicle (UAV) technologies could be employed to record the status of lhuq’us as a baseline for monitoring. Two study sites in the Salish sea were surveyed using UAV techniques: ȾEL,IȽĆ and St’utl’qulus. The overall accuracies of the UAV imagery classifications and the particular accuracies of the class representing lhuq’us suggest that UAV technologies paired with Google Earth Engine (GEE) object based image analysis (OBIA) methodologies can effectively detect lhuq’us. There are serious concerns and cultural values and practices deeply interconnected with culturally important species like lhuq’us. Through holding these concerns and values side by side with systematic observation and analyses maps and materials were created which communities can use to assert their rights, enact their own monitoring of territories and re-prioritize environmental decision-making done by federal, provincial, and municipal management agencies. / Graduate

Our planet is traversing the age of human-induced climate change and biodiversity loss. Projected global warming of 1.5 ºC above pre-industrial levels and related greenhouse gas emission pathways will bring about detrimental and irreversible impacts on the interconnected natural and human ecosystem. A global warming of 2 ºC could further exacerbate the risks across the sectors of biodiversity, energy, food, and water. Time- and cost-effective solutions and strategies are required for strengthening humanity’s response to the present environmental and societal challenges. Coastal seascape ecosystems including seagrasses, corals, mangrove forests, tidal flats, and salt marshes have been more recently heralded as nature-based solutions for mitigating and adapting to the climate-related impacts. This is due to their ability to absorb and store large quantities of carbon from the atmosphere. Focusing on seagrass habitats, although occupying only 0.2% of the world’s oceans, they can sequestrate up to 10% of the total oceanic carbon pool, all the while providing important food security, biodiversity, and coastal protection. But seagrass ecosystems, as all of their blue carbon seascape neighbors, are losing 1.5% of their extent per year due to anthropogenic activities. This has adverse implications for global carbon stocks, coastal protection, and marine biodiversity. Seagrass and seascape recession necessitates their science and policy-based management, protection, conservation which will ensure that our planet will remain within its sustainable boundaries in the age of climate change. The present PhD Thesis and research aim is to develop algorithms for seagrass mapping and monitoring leveraging the recent emergences in remote sensing technology―new satellite image archives, machine learning frameworks, and cloud computing―with field data from multiple sources. The main PhD findings are the demonstration of the suitability of Sentinel-2, RapidEye, and PlanetScope satellite imagery for regional to large-scale seagrass mapping; the introduction and incorporation of machine learning frameworks in the context of seagrass remote sensing and data analytics; the development of a semi-analytical model to invert the bottom reflectance of seagrasses; the design and implementation of multi-temporal satellite image approaches in coastal aquatic remote sensing; and the introduction, design and application of a scalable cloud-based tool to scale up seagrass mapping across large spatial and temporal dimensions. The approaches of the present PhD cover the gaps of the existing scientific literature of seagrass mapping in terms of the lack of spatial and temporal scalability and adaptability; the infancy in seagrass and seascape-related artificial intelligence endeavours; the restrictions of local server and mono-temporal approaches; and the absence of new methodological developments and applications using new (mainly open) satellite image archives. I anticipate and envisage that the near-future steps after the completion of my PhD will address the scalability of the designed cloud-native, data-driven mapping tool to standardise, automate, commercialise and democratise mapping and monitoring of seagrass and seascape ecosystems globally. The synergy of the developed momentum around the global seascape with the technological potential of Earth Observation can contribute to humanity’s race to adapt to and mitigate the climate change impacts and avoid cross tipping points in climate patterns, and biodiversity and ecosystem functions.

No description available.

Hydrology and water quality are affected by land use and climate changes. Mississippi’s diverse agro–ecosystem comprises of a range of land use land cover (LULC) including agriculture, forest, wetlands, urban, and grasslands. The objectives of this study were to investigate the impacts of various factors such as Best Management Practices (BMPs), wetlands, LULC, and climate changes on water quality and quantity. The hydrologic and water quality responses to dynamic LULC input in Soil and Water Assessment Tool (SWAT) were evaluated. Results showed that agricultural and forest expansion were major drivers of hydrologic and water quality changes in Big Sunflower River Watershed (BSRW), with agricultural expansion increasing runoff, sediments, and nutrients and forest expansion reducing these variables. The results showed that the integration of dynamic LULC and agricultural management operations in SWAT enables a more realistic representation of agricultural watersheds. Similarly, this study investigated the effects of wetland area changes overtime on surface and groundwater. Results demonstrated that 26% increase in wetland areas, reduced streamflow, sediments, total nitrogen, and total phosphorus by 2%, 37%, 13%, and 4% respectively as well as increased groundwater storage by 90 mm in selected sub–watershed. This highlighted the importance of preservation and restoration of wetlands to enhance the agro–ecosystem resilience to LULC change. Likewise, the effectiveness of BMPs in reducing sediment yield from critical areas within BSRW was assessed. Results demonstrated that BMPs reduced sediments by up to 50%, suggesting their usefulness in mitigating high sediment yield from agricultural areas. This study also assessed the impacts of climate change on streamflow and sediment loads and the role of waterbodies in mitigating those impacts. Results depicted a significant increase in future streamflow and sediment loads due to potential increase in precipitation and temperature. When waterbodies were simulated, projected change in annual streamflow was < 1%. However, the projected annual sediment loads reduced substantially by 44–46%, highlighting the role of waterbodies on watershed resilience to climate change. Overall, this dissertation study provides insights about the complex interactions between LULC, climate, anthropogenic activities, and water resources that can help to develop watershed management strategies to promote agricultural sustainability.

An investigation into using attention mechanisms for better feature extraction in wildfire spread prediction models. This research examines the U-net architecture to achieve image segmentation, a process that partitions images by classifying pixels into one of two classes. The deep learning models explored in this research integrate modern deep learning architectures, and techniques used to optimize them. The models are trained on 12 distinct observational variables derived from the Google Earth Engine catalog. Evaluation is conducted with accuracy, Dice coefficient score, ROC-AUC, and F1-score. This research concludes that when augmenting U-net with attention mechanisms, the attention component improves feature suppression and recognition, improving overall performance. Furthermore, employing ensemble modeling reduces bias and variation, leading to more consistent and accurate predictions. When inferencing on wildfire propagation at 30-minute intervals, the architecture presented in this research achieved a ROC-AUC score of 86.2% and an accuracy of 82.1%.

Detecting water under vegetation is critical to tracking the status of geological ecosystems like wetlands. Researchers use different methods to estimate water presence, avoiding costly on-site measurements. Optical satellite imagery allows the automatic delineation of water using the concept of the Normalised Difference Water Index (NDWI). Still, optical imagery is subject to visibility conditions and cannot detect water under the vegetation, a typical situation for wetlands. Synthetic Aperture Radar (SAR) imagery works under all visibility conditions. It can detect water under vegetation but requires deep network algorithms to segment water presence, and manual annotation work is required to train the deep models. This project uses DEEPAQUA, a cross-modal knowledge distillation method, to eliminate the manual annotation needed to extract water presence from SAR imagery with deep neural networks. In this method, a deep student model (e.g., UNET) is trained to segment water in SAR imagery. The student model uses the NDWI algorithm as the non-parametric, cross-modal teacher. The key prerequisite is that NDWI works on the optical imagery taken from the exact location and simultaneously as the SAR. Three different deep architectures are tested in this project: UNET, SegNet, and UNET++, and the Otsu method is used as the baseline. Experiments on imagery from Swedish wetlands in 2020-2022 show that cross-modal distillation consistently achieved better segmentation performances across architectures than the baseline. Additionally, the UNET family of algorithms performed better than SegNet with a confidence of 95%. The UNET++ model achieved the highest Intersection Over Union (IOU) performance. However, no statistical evidence emerged that UNET++ performs better than UNET, with a confidence of 95%. In conclusion, this project shows that cross-modal knowledge distillation works well across architectures and removes tedious and expensive manual work hours when detecting water from SAR imagery. Further research could evaluate performances on other datasets and student architectures. / Att upptäcka vatten under vegetation är avgörande för att hålla koll på statusen på geologiska ekosystem som våtmarker. Forskare använder olika metoder för att uppskatta vattennärvaro vilket undviker kostsamma mätningar på plats. Optiska satellitbilder tillåter automatisk avgränsning av vatten med hjälp av konceptet Normalised Difference Water Index (NDWI). Optiska bilder fortfarande beroende av siktförhållanden och kan inte upptäcka vatten under vegetationen, en typisk situation för våtmarker. Synthetic Aperture Radar (SAR)-bilder fungerar under alla siktförhållanden. Den kan detektera vatten under vegetation men kräver djupa nätverksalgoritmer för att segmentera vattennärvaro, och manuellt anteckningsarbete krävs för att träna de djupa modellerna. Detta projekt använder DEEPAQUA, en cross-modal kunskapsdestillationsmetod, för att eliminera det manuella annoteringsarbete som behövs för att extrahera vattennärvaro från SAR-bilder med djupa neurala nätverk. I denna metod tränas en djup studentmodell (t.ex. UNET) att segmentera vatten i SAR-bilder semantiskt. Elevmodellen använder NDWI, som fungerar på de optiska bilderna tagna från den exakta platsen och samtidigt som SAR, som den icke-parametriska, cross-modal lärarmodellen. Tre olika djupa arkitekturer testas i detta examensarbete: UNET, SegNet och UNET++, och Otsu-metoden används som baslinje. Experiment på bilder tagna på svenska våtmarker 2020-2022 visar att cross-modal destillation konsekvent uppnådde bättre segmenteringsprestanda över olika arkitekturer jämfört med baslinjen. Dessutom presterade UNET-familjen av algoritmer bättre än SegNet med en konfidens på 95%. UNET++-modellen uppnådde högsta prestanda för Intersection Over Union (IOU). Det framkom dock inga statistiska bevis för att UNET++ presterar bättre än UNET, med en konfidens på 95%. Sammanfattningsvis visar detta projekt att cross-modal kunskapsdestillation fungerar bra över olika arkitekturer och tar bort tidskrävande och kostsamma manuella arbetstimmar vid detektering av vatten från SAR-bilder. Ytterligare forskning skulle kunna utvärdera prestanda på andra datamängder och studentarkitekturer.