Rapid urbanization is taking place in Low-and middle-income countries (LMICs). Often there is not sufficient data monitoring the quick urban change. This study explores the use of machine learning classification within remote sensing to foster sustainable urban practices in a secondary city in an LMIC. The aim is to extract spatially detailed land cover data and investigate its temporal evolution from 2018 to 2021. Furthermore, targeted interviews with residents were conducted to gain an in-situ understanding of the land cover changes. The research reveals a trend of increased impervious surface in Sekondi-Takoradi, especially around the urban outskirts. Some patterns of densification can also be identified, predominantly in urban areas with a mix of impervious surfaces and vegetation. These findings reveal similar land cover change patterns as previous remote sensing studies, a decrease in vegetation, and an increase in impervious surfaces.  The used method can be applied at a larger scale to monitor the urbanization of secondary cities in LMICs, a field that often is neglected. These insights can contribute to achieving the UN's 11th Sustainable Development Sustainable Cities and Communities.

Karlstads kommun innehar en så kallad solkarta vars syfte är att ge kommunens invånare en översikt över hur mycket solenergi som infaller på varje individuell takyta och kan användas som underlag för beslut om installation av solpaneler på byggnadstak. Kartan är interaktiv och sträcker sig över hela kommunen. Tyvärr brister den i detaljnivå utanför Karlstad tätort och skulle behöva en uppdatering. Syftet med studien är att undersöka metoder att utvinna tvådimensionella takplansytor enligt LOD2 utifrån byggnadsfotavtryck och punktmolnsdata, som sedan ska kunna lägga grunden till en interaktiv solkarta. Tre metoder för takplanssegmentering valdes ut som baserades på tre olika GIS-mjukvaror: ArcGIS Pro, Whitebox Tools och TerraScan. Studieområdet, beläget på industriområdet Våxnäs i Karlstad, bestod av 68 byggnader med varierande taktyper av olika hög komplexitet. En ytterligare dimension till studien var att två olika indatamängder jämfördes för varje segmenteringsmetod: ett högupplöst fotogrammetriskt framställt punktmoln utifrån bilder tagna med UAV (Unmanned Aerial Vehicle) samt ett lägre upplöst punktmoln insamlat med flygburen laserskanning. Totalt erhölls sex olika resultat som utvärderades efter fullständighet och utseende. Den högsta medelfullständigheten för varje metod var: 99,6 % för metoden baserad på TerraScan, 90,2 % för metoden baserad på Whitebox Tools och 82,0 % för metoden baserad på ArcGIS Pro. Gällande indatamängder gav UAV-datamängden ca 6 procentenheter bättre medelfullständighet för de två bästa metoderna och 10 procentenheter lägre för metoden som funkade sämst. Gällande användbarheten av resultaten är kontentan att TerraScan-metoden hade lagt en bra grund för en solkarta. Whitebox Tools-metoden hade sannolikt också kunnat vara användbar om en förbättrad generaliseringsalgoritm i efterbearbetningen hade applicerats. Studien diskuterar skillnader, felkällor samt nämner några ytterligare beprövade metoder som aldrig färdigställdes på grund av odugliga resultat. Problem som återstår att lösa är hantering av hål i punktmolnsdata inför takplanssegmentering. / Karlstad Municipality has what is known as a solar radiation map with the purpose of providing an overview of how much solar irradiance that is received by individual roof planes. It serves as a basis for decisions regarding the installation of solar panels on building roofs. The map is interactive and covers the entire municipality. Unfortunately, it lacks detail outside the city of Karlstad and would benefit from an update. The study aims to explore methods for extracting two-dimensional roof planes according to LOD2 (Level of Detail 2) using building footprints and point cloud data. The roof planes could later form the foundation for an interactive solar map. Three methods for roof segmentation were examined, each based on different software: ArcGIS Pro, Whitebox Tools, and TerraScan. The study area, located in the Våxnäs industrial area in Karlstad consisted of 68 buildings with varying roof types and complexities. An additional dimension to the study involved comparing two different input datasets for each segmentation method: a high-resolution photogrammetric point cloud generated from UAV (Unmanned Aerial Vehicle) images and a lower-resolution point cloud collected with airborne laser scanning. In total, six different results were evaluated based on completeness and appearance. The highest mean completeness for each method was: 99,6 % for the TerraScan-based method, 90,2 % for the Whitebox Tools-based method and 82,0 % for the ArcGIS Pro-based method. Regarding input datasets, the two best methods showed an increase of approximately 6 percentage points in mean completeness for the UAV dataset, while the least effective method showed a decrease of 10 percentage points. In terms of practicality, the TerraScan method provided a solid basis for a solar map. The Whitebox Tools method could most likely be usable if a better generalization algorithm in post-processing is cultivated. The study also discusses differences, potential sources of error, and mentions some additional methods that were not fully developed due to inadequate results. Remaining challenges include addressing gaps of missing data in point clouds before roof plane segmentation.

This thesis is carried out in collaboration with a private company, DANSiC AB This study is an extension of a research work started by DANSiC AB in 2019 to classify a source. This study is about classifying a source into two classes with the sensitivity of one source higher than the other as one source has greater importance. The data provided for this thesis is based on sensor measurements on different temperature cycles. The data is high-dimensional and is expected to have a drift in measurements. Principal component analysis (PCA) is used for dimensionality reduction. “Differential”, “Relative” and “Fractional” drift compensation techniques are used for compensating the drift in data. A comparative study was performed using three different classification algorithms, which are “Linear Discriminant Analysis (LDA)”, “Naive Bayes classifier (NB)” and “Random forest (RF)”. The highest accuracy achieved is 59%,Random forest is observed to perform better than the other classifiers. / <p>This work is done with DANSiC AB in collaboration with Linkoping University.</p>

Remote sensing data has great potential for wildfire detection and monitoring with enhanced spatial resolution and temporal coverage. Earth Observation satellites have been employed to systematically monitor fire activity over large regions in two ways: (i) to detect the location of actively burning spots (during the fire event), and (ii) to map the spatial extent of the burned scars (during or after the event). Active fire detection plays an important role in wildfire early warning systems. The open-access of Sentinel-2 multispectral data at 20-m resolution offers an opportunity to evaluate its complementary role to the coarse indication in the hotspots provided by MODIS-like polar-orbiting and GOES-like geostationary systems. In addition, accurate and timely mapping of burned areas is needed for damage assessment. Recent advances in deep learning (DL) provides the researcher with automatic, accurate, and bias-free large-scale mapping options for burned area mapping using uni-temporal multispectral imagery. Therefore, the objective of this thesis is to evaluate multispectral remote sensing data (in particular Sentinel-2) for wildfire detection, including active fire detection using a multi-criteria approach and burned area detection using DL models.        For active fire detection, a multi-criteria approach based on the reflectance of B4, B11, and B12 of Sentinel-2 MSI data is developed for several representative fire-prone biomes to extract unambiguous active fire pixels. The adaptive thresholds for each biome are statistically determined from 11 million Sentinel-2 observations samples acquired over summertime (June 2019 to September 2019) across 14 regions or countries. The primary criterion is derived from 3 sigma prediction interval of OLS regression of observation samples for each biome. More specific criteria based on B11 and B12 are further introduced to reduce the omission errors (OE) and commission errors (CE).        The multi-criteria approach proves to be effective in cool smoldering fire detection in study areas with tropical &amp; subtropical grasslands, savannas &amp; shrublands using the primary criterion. At the same time, additional criteria that thresholds the reflectance of B11 and B12 can effectively decrease the CE caused by extremely bright flames around the hot cores in testing sites with Mediterranean forests, woodlands &amp; scrub. The other criterion based on reflectance ratio between B12 and B11 also avoids the effects of CE caused by hot soil pixels in sites with tropical &amp; subtropical moist broadleaf forests. Overall, the validation performance over testing patches reveals that CE and OE can be kept at a low level  (0.14 and 0.04) as an acceptable trade-off. This multi-criteria algorithm is suitable for rapid active fire detection based on uni-temporal imagery without the requirement of multi-temporal data. Medium-resolution multispectral data can be used as a complementary choice to the coarse resolution images for their ability to detect small burning areas and to detect active fires more accurately.        For burned area mapping, this thesis aims to expound on the capability of deep DL models for automatically mapping burned areas from uni-temporal multispectral imagery. Various burned area detection algorithms have been developed using Sentinel-2 and/or Landsat data, but most of the studies require a pre-fire image, dense time-series data, or an empirical threshold. In this thesis, several semantic segmentation network architectures, i.e., U-Net, HRNet, Fast- SCNN, and DeepLabv3+ are applied to Sentinel-2 imagery and Landsat-8 imagery over three testing sites in two local climate zones. In addition, three popular machine learning (ML) algorithms (LightGBM, KNN, and random forests) and NBR thresholding techniques (empirical and OTSU-based) are used in the same study areas for comparison.        The validation results show that DL algorithms outperform the machine learning (ML) methods in two of the three cases with the compact burned scars,  while ML methods seem to be more suitable for mapping dispersed scar in boreal forests. Using Sentinel-2 images, U-Net and HRNet exhibit comparatively identical performance with higher kappa (around 0.9) in one heterogeneous Mediterranean fire site in Greece; Fast-SCNN performs better than others with kappa over 0.79 in one compact boreal forest fire with various burn severity in Sweden. Furthermore, directly transferring the trained models to corresponding Landsat-8 data, HRNet dominates in the three test sites among DL models and can preserve the high accuracy. The results demonstrate that DL models can make full use of contextual information and capture spatial details in multiple scales from fire-sensitive spectral bands to map burned areas. With the uni-temporal image, DL-based methods have the potential to be used for the next Earth observation satellite with onboard data processing and limited storage for previous scenes.    In the future study, DL models will be explored to detect active fire from multi-resolution remote sensing data. The existing problem of unbalanced labeled data can be resolved via advanced DL architecture, the suitable configuration on the training dataset, and improved loss function. To further explore the damage caused by wildfire, future work will focus on the burn severity assessment based on DL models through multi-class semantic segmentation. In addition, the translation between optical and SAR imagery based on Generative Adversarial Network (GAN) model could be explored to improve burned area mapping in different weather conditions. / Fjärranalysdata har stor potential för upptäckt och övervakning av skogsbränder med förbättrad rumslig upplösning och tidsmässig täckning. Jordobservationssatelliter har använts för att systematiskt övervaka brandaktivitet över stora regioner på två sätt: (i) för att upptäcka placeringen av aktivt brinnande fläckar (under brandhändelsen) och (ii) för att kartlägga den brända ärrens rumsliga omfattning ( under eller efter evenemanget). Aktiv branddetektering spelar en viktig roll i system för tidig varning för skogsbränder. Den öppna tillgången till Sentinel-2 multispektral data vid 20 m upplösning ger en möjlighet att utvärdera dess kompletterande roll i förhållande till den grova indikationen i hotspots som tillhandahålls av MODIS-liknande polaromloppsbanesystem och GOES-liknande geostationära system. Dessutom krävs en korrekt och snabb kartläggning av brända områden för skadebedömning. Senaste framstegen inom deep learning (DL) ger forskaren automatiska, exakta och förspänningsfria storskaliga kartläggningsalternativ för kartläggning av bränt område med unitemporal multispektral bild. Därför är syftet med denna avhandling att utvärdera multispektral fjärranalysdata (särskilt Sentinel- 2) för att upptäcka skogsbränder, inklusive aktiv branddetektering med hjälp av ett multikriterietillvägagångssätt och detektering av bränt område med DL-modeller. För aktiv branddetektering utvecklas en multikriteriemetod baserad på reflektionen av B4, B11 och B12 i Stentinel-2 MSI data för flera representativa brandbenägna biom för att få fram otvetydiga pixlar för aktiv brand. De adaptiva tröskelvärdena för varje biom bestäms statistiskt från 11 miljoner Sentinel-2 observationsprover som förvärvats under sommaren (juni 2019 till september 2019) i 14 regioner eller länder. Det primära kriteriet härleds från 3-sigma-prediktionsintervallet för OLS-regression av observationsprover för varje biom. Mer specifika kriterier baserade på B11 och B12 införs vidare för att minska utelämningsfel (OE) och kommissionsfel (CE). Det multikriteriella tillvägagångssättet visar sig vara effektivt när det gäller upptäckt av svala pyrande bränder i undersökningsområden med tropiska och subtropiska gräsmarker, savanner och buskmarker med hjälp av det primära kriteriet. Samtidigt kan ytterligare kriterier som tröskelvärden för reflektionen av B11 och B12 effektivt minska det fel som orsakas av extremt ljusa lågor runt de heta kärnorna i testområden med skogar, skogsmarker och buskage i Medelhavsområdet. Det andra kriteriet som bygger på förhållandet mellan B12 och B11:s reflektionsgrad undviker också effekterna av CE som orsakas av heta markpixlar i områden med tropiska och subtropiska fuktiga lövskogar. Sammantaget visar valideringsresultatet för testområden att CE och OE kan hållas på en låg nivå (0,14 och 0,04) som en godtagbar kompromiss. Algoritmen med flera kriterier lämpar sig för snabb aktiv branddetektering baserad på unika tidsmässiga bilder utan krav på tidsmässiga data. Multispektrala data med medelhög upplösning kan användas som ett kompletterande val till bilder med kursupplösning på grund av deras förmåga att upptäcka små brinnande områden och att upptäcka aktiva bränder mer exakt. När det gäller kartläggning av brända områden syftar denna avhandling till att förklara hur djupa DL-modeller kan användas för att automatiskt kartlägga brända områden från multispektrala bilder i ett tidsintervall. Olika algoritmer för upptäckt av brända områden har utvecklats med hjälp av Sentinel-2 och/eller Landsat-data, men de flesta av studierna kräver att man har en förebränning. bild före branden, täta tidsseriedata eller ett empiriskt tröskelvärde. I den här avhandlingen tillämpas flera arkitekturer för semantiska segmenteringsnätverk, dvs. U-Net, HRNet, Fast- SCNN och DeepLabv3+, på Sentinel- 2 bilder och Landsat-8 bilder över tre testplatser i två lokala klimatzoner. Dessutom används tre populära algoritmer för maskininlärning (ML) (Light- GBM, KNN och slumpmässiga skogar) och NBR-tröskelvärden (empiriska och OTSU-baserade) i samma undersökningsområden för jämförelse. Valideringsresultaten visar att DL-algoritmerna överträffar maskininlärningsmetoderna (ML) i två av de tre fallen med kompakta brända ärr, medan ML-metoderna verkar vara mer lämpliga för kartläggning av spridda ärr i boreala skogar. Med hjälp av Sentinel-2 bilder uppvisar U-Net och HRNet jämförelsevis identiska prestanda med högre kappa (omkring 0,9) i en heterogen brandplats i Medelhavet i Grekland; Fast-SCNN presterar bättre än andra med kappa över 0,79 i en kompakt boreal skogsbrand med varierande brännskadegrad i Sverige. Vid direkt överföring av de tränade modellerna till motsvarande Landsat-8-data dominerar HRNet dessutom på de tre testplatserna bland DL-modellerna och kan bevara den höga noggrannheten. Resultaten visade att DL-modeller kan utnyttja kontextuell information fullt ut och fånga rumsliga detaljer i flera skalor från brandkänsliga spektralband för att kartlägga brända områden. Med den unika tidsmässiga bilden har DL-baserade metoder potential att användas för nästa jordobservationssatellit med databehandling ombord och begränsad lagring av tidigare scener. I den framtida studien kommer DL-modeller att undersökas för att upptäcka aktiva bränder från fjärranalysdata med flera upplösningar. Det befintliga problemet med obalanserade märkta data kan lösas med hjälp av en avancerad DL-arkitektur, lämplig konfiguration av träningsdatasetet och förbättrad förlustfunktion. För att ytterligare utforska de skador som orsakas av skogsbränder kommer det framtida arbetet att fokusera på bedömningen av brännskadornas allvarlighetsgrad baserat på DL-modeller genom semantisk segmentering av flera klasser. Dessutom kan översättningen mellan optiska bilder och SAR-bilder baserad på en GAN-modell (Generative Adversarial Network) undersökas för att förbättra kartläggningen av brända områden under olika väderförhållanden. / <p>QC 20210525</p>

Abstract A wildfire is an uncontrollable fire in an area of combustible fuel that occurs in the wild or countryside area. Wildfires are becoming a deadly and frequent event in Europe due to extreme weather conditions. In 2018, wildfires profoundly affected Sweden, Finland, and Norway, which were not big news before. In Norway, although there is well–organized fire detection, warning, and mitigation systems, mapping wildfire risk areas before the fire occurrence with georeferenced spatial information, are not yet well-practiced. At this moment, there are freely available remotely sensed spatial data and there is a good possibility that analysing wildfire hazard areas with geographical information systems together with multicriteria decision analysis (GIS–MCDA) and frequency ratio models in advance so that subsequent wildfire warning, mitigation, organizational and post resilience activities and preparations can be better planned.  This project covers eight counties of Norway: Oslo, Akershus, Østfold, Vestfold, Telemark, Buskerud, Oppland, and Hedmark. These are the counties with the highest wildfire frequency for the last ten years in Norway. In this study, GIS-MCDA integrated with analytic hierarchy process (AHP), and frequency ratio models (FR) were used with selected sixteen–factor criteria based on their relative importance to wildfire ignition, fuel load, and other related characteristics. The produced factor maps were grouped under four main clusters (K): land use (K1), climate (K2), socioeconomic (K3), and topography (K4) for further analysis. The final map was classified into no hazard, low, medium, and high hazard level rates. The comparison result showed that the frequency ratio model with MODIS satellite data had a prediction rate with 72% efficiency, followed by the same model with VIIRS data and 70% efficiency. The GIS-MCDA model result showed 67% efficiency with both MODIS and VIIRS data. Those results were interpreted in accordance with Yesilnacar’s classifications such as the frequency ratio model with MODIS data was considered a good predictor, whereas the GIS-MCDA model was an average predictor. When testing the model on the dependent data set, the frequency ratio model showed 72% with MODIS &amp; VIIRS data, and the GIS-MCDA model showed 67% and 68% performance with MODIS and VIIRS data, respectively. In the hazard maps produced, the frequency ratio models for both MODIS and VIIRS showed that Hedmark and Akershus counties had the largest areas with the highest susceptibility to wildfires, while the GIS-MCDA method resulted to Østfold and Vestfold counties. Through this study, the best independent wildfire predictor criteria were selected from the highest to the lowest of importance; wildfire constraint and criteria maps were produced; wildfire hazard maps with high-resolution georeferenced data using three models were produced and compared; and the best, reliable, robust, and applicable model alternative was selected and recommended. Therefore, the aims and specific objectives of this study should be considered and fulfilled.

This paper describes a unique method for the semi-automatic detection of historic charcoal production sites in LiDAR-derived digital elevation models. Intensified iron production in the early 17th century has remarkably influenced ways of how the land in Sweden was managed. Today, the abundance of charcoal kilns embedded in the landscape survives as cultural heritage monuments that testify about the scale forest management for charcoal production has contributed to the uprising iron manufacturing industry. An arbitrary selected study area (54 km2) south west of Gävle city served as an ideal testing ground, which is known to consist of already registered as well as unsurveyed charcoal kiln sites. The proposed approach encompasses combined morphometric classification methods being subjected to analytical image processing, where an image that represents refined terrain morphology was segmented and further followed by Hough Circle transfer function applied in seeking to detect circular shapes that represent charcoal kilns. Sites that have been identified manually and using the proposed method were only verified within an additionally established smaller validation area (6 km2). The resulting outcome accuracy was measured by calculating harmonic mean of precision and recall (F1-Score). Along with indication of previously undiscovered site locations, the proposed method showed relatively high score in recognising already registered sites after post-processing filtering. In spite of required continual fine-tuning, the described method can considerably facilitate mapping and overall management of cultural resources.

With a changing climate and an increased competition over water resources for agricultural irrigation, the need to improve crop water productivity using time and cost-efficient methodologies have become critically important. The Malwathu Oya river basin in Sri Lanka is struggling with water scarcity, which threatens food security and the income of farmers. In this study, freely available remote sensed land- and water productivity data from FAO’s WaPOR database was evaluated. The evaluation consisted of a comparison of the WaPOR data and primary collected field data using the crop water model, AquaCrop, for three irrigation schemes in the Malwathu Oya river basin. Additionally, the spatio-temporal variability in crop water use within and across these three irrigation schemes was assessed using indicators derived from the WaPOR portal. The evaluation was conducted for the main cultivation season, called Maha, between 2010 and 2021.  The WaPOR and AquaCrop actual evapotranspiration (ETa) values were found to be in relatively good agreement (312–537 and 400–465 mm respectively). WaPOR yield values (2.5–2.9 ton/ha) were however lower compared to the AquaCrop simulated yield values and historical yield data (4.6–5.7 and 4.4–5.6 ton/ha respectively). Difference in calculation methodology, possible sources of error in WaPOR conversion calculations and limitations in accuracy caused by cloud coverage when collecting satellite data could be explanations for this. Prior knowledge and accurate allocation of the crop type and parameters used in conversion calculations in WaPOR is therefore of significant influence. From the spatio-temporal variation assessment with WaPOR indicators, a fair uniformity of the water distribution within the irrigation schemes was shown (CV 11–19 %). The beneficial water use (BWU) in the irrigation schemes showed lower values (50–90 % allocated to T) for years when the available water amount was higher, which could be explained by the higher rate of water lost through soil evaporation. Crop water productivity (CWP) values showed higher values (about 0.70 kgDM/m3) when the available water amount was higher, indicating that yield production is sensitive to water-scarce environments. Applying a yield boundary function, representing the best attainable yield in relation to water resource, showed that there is potential to achieve the same yield with less amount of water. There are thus possibilities for improved water productivity in the three irrigation schemes investigated. For future research it is recommended to perform a sensitivity analysis for WaPOR and ground truth with yield data to obtain a better understanding of potential limitations. To obtain more precise site descriptions it is also recommended to ground truth AquaCrop with yield and soil data.