Дисертації з теми "Fire prevention Remote sensing"

An experiment was designed to capture micrometeorological observations during a fire spread on a simple slope. Three towers equipped with a variety of instrumentation, an array of fire-sensing packages, and a Doppler lidar was deployed to measure various aspects of the fire. Pressure and temperature perturbations were analyzed for each of the grid packages to determine if the fire intensity could be observed in the covariance of the two variables. While two of the packages measured a covariance less than –15 °C hPa, there was no clear trend across the grid. The fire front passage at each of the three towers on the slope yielded extreme swings in observed turbulent kinetic energy and sensible heat flux. Vertical velocity turbulence spectra showed that the high-intensity fire front passage at the bottom tower was 2 to 3 orders of magnitude larger than the low-intensity fire front passages at the top two towers. Opposing wind regimes on the slope caused a unique L-shaped pattern to form in the fire front. A vorticity estimation from the sonic anemometers showed that vorticity reached a maximum just as a fire whirl formed in the bend of the L-shaped fire front, leading to a rapid increase in fire spread.

Forest fire policies are changing. Managers now face conflicting imperatives to re-establish pre-suppression fire regimes, while simultaneously preventing resource destruction. They must, therefore, understand the spatial patterns of fires. Geographers can facilitate this understanding by developing new techniques for mapping fire behavior. This dissertation develops such techniques for mapping recent fires and using these maps to calibrate models of potential fire hazards. In so doing, it features techniques that strive to address the inherent complexity of modeling the combinations of variables found in most ecological systems. Image processing techniques were used to stratify the elements of terrain, slope, elevation, and aspect. These stratification images were used to assure sample placement considered the role of terrain in fire behavior. Examination of multiple stratification images indicated samples were placed representatively across a controlled range of scales. The incorporation of terrain data also improved preliminary fire hazard classification accuracy by 40%, compared with remotely sensed data alone. A Kauth-Thomas transformation (KT) of pre-fire and post-fire Thematic Mapper (TM) remotely sensed data produced brightness, greenness, and wetness images. Image subtraction indicated fire induced change in brightness, greenness, and wetness. Field data guided a fuzzy classification of these change images. Because fuzzy classification can characterize a continuum of a phenomena where discrete classification may produce artificial borders, fuzzy classification was found to offer a range of fire severity information unavailable with discrete classification. These mapped fire patterns were used to calibrate a model of fire hazards for the entire mountain range. Pre-fire TM, and a digital elevation model produced a set of co-registered images. Training statistics were developed from 30 polygons associated with the previously mapped fire severity. Fuzzy classifications of potential burn patterns were produced from these images. Observed field data values were displayed over the hazard imagery to indicate the effectiveness of the model. Areas that burned without suppression during maximum fire severity are predicted best. Areas with widely spaced trees and grassy understory appear to be misrepresented, perhaps as a consequence of inaccuracies in the initial fire mapping.

The use of active and passive remote sensing systems for relating forest spatial patterns to fire history was tested over one of the Arizona Sky Islands. Using Landsat Thematic Mapper (TM), Shuttle Imaging Radar (SIR-C), and data fusion I examined the relationship between landscape metrics and a range of fire history characteristics. Each data type (TM, SIR-C, and fused) was processed in the following manner: each band, channel, or derived feature was simplified to a thematic layer and landscape statistics were calculated for plots with known fire history. These landscape metrics were then correlated with fire history characteristics, including number of fire-free years in a given time period, mean fire-free interval, and time since fire. Results from all three case studies showed significant relationships between fire history and forest spatial patterns. Data fusion performed as well or better than Landsat TM alone, and better than SIR-C alone. These comparisons were based on number and strength of significant correlations each method achieved. The landscape metric that was most consistent and obtained the greatest number of significant correlations was Shannon's Diversity Index. Results also agreed with field-based research that has linked higher fire frequency to increased landscape diversity and patchiness. An additional finding was that the fused data seem to detect fire-related spatial patterns over a range of scales.

Doutoramento em Engenharia Florestal e dos Recursos Naturais - Instituto Superior de Agronomia
In tropical savannas, one of the most fire-prone biomes on Earth, fire management is a continuous and iterative process that can only be effectively achieved with thorough consideration of fire regimes. Based on remotely sensed imagery and in-situ field data, key fire regime components were assessed for tropical savannas of northern Australia (frequency, patchiness, intensity, and severity) and Brazil (frequency). The discrete lognormal model was found to be the best method for modelling fire frequency in tropical savannas, and demonstrated that fire frequency is very high in both countries. In northern Australia, fire patchiness was lower in the late dry season, characterized by shorter and fewer unburned patches, than in the early dry season. Fire intensity and severity were highest in the late dry season. The observed temporal differences are consistent with the hypothesis that climate is the main driver of fire regime seasonality. Fuel load and fuel continuity explained fire regime differences between vegetation types. Fire season was bimodal, with peaks in May and October, related to periods of anthropogenic fire and optimal fire weather conditions. Prescribed burning in the early dry season can increase the patchiness and reduce the intensity of late dry season fires, with substantial benefits for biodiversity and a reduction in greenhouse gas emissions

Thesis (M.S.) -- University of Montana, 2009.
Title from author supplied metadata. Description based on contents viewed on October 14, 2009. Author supplied keywords: Fire, precipitation, season ending event, Remote Automated Weather Stations, North American Regional Reanalysis, Northern Rockies, climate change. Includes bibliographical references.

Our understanding of natural fire regimes in human-dominated landscapes is limited. Fire regimes operating in the pyrogenic ecosystems of Florida have been altered by fire suppression and fuel fragmentation. This is especially true of North Merritt Island, Florida, where human impacts have led to an incomplete knowledge of current fire regimes. We know that growing season fires frequently occurred within general return intervals and that many native terrestrial species require fire to remain viable. A 20-year plus period of fire suppression caused structural and compositional changes to vegetation/fuels that led to catastrophic fires and the decline of native species populations such as the Florida Scrub-Jay. Fire has been reintroduced as a means to reduce fuels and maintain habitat requirements for native species. Scientific studies have documented the effects and benefits of prescribed burning on KSC/MINWR habitat/fuels structure. The necessity for fire to maintain vegetation/fuels structure and composition on the landscape is clear so fire is being applied to the landscape despite our imperfect knowledge of the native fire regime. It is imperative for the survival of many native species that fire managers be able to mimic the results of the native fire regime. Fire regime research is difficult in shrublands, and using dendrochronologic techniques are often not possible in flatwoods communities. I therefore used a process of remote sensing, GIS mapping, and spatial modeling to quantify lightning fire ignition properties, the current managed fire regime, and the natural fire regime. Chapter one develops a new remote sensing technique to accurately map burned areas in Florida scrub and pine flatwoods dominated communities on Kennedy Space Center, Merritt Island National Wildlife Refuge, Canaveral National Seashore, and Cape Canaveral Air Force Station, Florida. At the center of this technique is a newly developed separation index (SI) that was used to evaluate each individual satellite image band for its power to discriminate unburned and burned areas. Burned areas were classified and found to be highly accurate in relation to empirical fire records. This chapter addressed a number of issues relevant to the classification of burned areas including: the effect of geographic extent of remote sensing data on classification, determining the best bands for classification, and cleaning classification results by using GIS masking. It also serves as the first published effort to map fire scars in the Florida scrub and flatwoods vegetative communities of the southeastern U.S. using image processing techniques. Chapter two quantified a managed fire regime on John F. Kennedy Space Center, Florida and surrounding federal properties by mapping all fires between 1983 to 2005 using the image processing technique developed in chapter one, time series satellite imagery, and GIS techniques. The goals were to: (1) determine if an image processing technique designed for individual fire scar mapping could be applied to an image time series for mapping a managed fire regime in a rapid re-growth pyrogenic system; (2) develop a method for labeling mapped fire scar confidence knowing that a formal accuracy analysis was not possible; and (3) compare results of the managed fire regime with regional information on natural fire regimes to look for similarities/differences that might help optimize management for persistence of native fire-dependent species. The area burned by managed fire peaked when the drought index was low and was reduced when the drought index was high. This contrasts with the expectations regarding the natural fire regime of this region. Chapter three quantified the natural lightning ignition regime on Kennedy Space Center, Merritt Island National Wildlife Refuge, Canaveral National Seashore, and Cape Canaveral Air Force Station, Florida. Lightning is the natural ignition source in Florida, substantiating the need for understanding lightning fire incidence. Sixteen years of lightning data (1986-2003, excluding 1987 and 2002 due to missing data) from the NASA Cloud to Ground Lightning Surveillance System and fire ignition records were used to quantify the relationship between lightning incidence and fire ignition. Precipitation influenced the efficiency of lightning ignitions, particularly July precipitation. Negative polarity strikes caused the majority of ignitions. Pine flatwoods was ignited more frequently than expected given equal chance of ignition among landcover types. About half (51%) of detected fires were instantaneous ignitions and the other 49% were delayed an average of two days. Summer lightning ignitions were dominant, especially during July, with only one winter lightning ignition. Chapter four used an existing fire regime model (HFire) to simulate the natural fire regime (prior to European settlement) on Kennedy Space Center, Merritt Island National Wildlife Refuge, Canaveral National Seashore, and Cape Canaveral Air Force Station, Florida. A sensitivity analysis was performed to establish which parameters were most important and the range of variation surrounding empirically derived model information from the same model. A mosaic pattern of small fires dominated this fire regime with extremely large fires occurring during dry La Nina periods. Dead fuel moisture and wind speed had the largest influence on model outcome. The majority of variability was found to be in the largest fires. The research presenter here provides a comprehensive perspective on current and historic fire regimes that may be useful for optimizing land management on Kennedy Space Center, Merritt Island National Wildlife Refuge, Canaveral National Seashore, and Cape Canaveral Air Force Station, Florida and throughout the southeastern United States. Native fire dependent species are suffering from many changes imposed from human alteration. The success of conservation efforts protecting native fire dependent species hinge on my factors. Two of the largest factors are first protecting native habitat and then secondly managing that protected habitat to mimic natural maintenance processes for suitable structure and composition which may favor their demography. This study focuses on developing techniques necessary for producing information that can aid the optimization of fire management on these properties and within the southeastern United States, but may be useful in other fire maintained ecosystems globally.
Ph.D.
Department of Biology
Sciences
Conservation Biology PhD

Fire activity, in terms of intensity, frequency, and total area burned, is expected to increase with a changing climate. A challenge for landscape-level assessment of fire effects, often termed burn severity, is that current remote sensing assessments provide very little information regarding tree/vegetation physiological performance and recovery, limiting our understanding of fire effects on ecosystem services such as carbon storage/cycling. In this paper, we evaluated whether spectral indices common in vegetation stress and burn severity assessments could accurately quantify post-fire physiological performance (indicated by net photosynthesis and crown scorch) of two seedling species, Larix occidentalis and Pinus contorta. Seedlings were subjected to increasing fire radiative energy density (FRED) doses through a series of controlled laboratory surface fires. Mortality, physiology, and spectral reflectance were assessed for a month following the fires, and then again at one year post-fire. The differenced Normalized Difference Vegetation Index (dNDVI) spectral index outperformed other spectral indices used for vegetation stress and burn severity characterization in regard to leaf net photosynthesis quantification, indicating that landscape-level quantification of tree physiology may be possible. Additionally, the survival of the majority of seedlings in the low and moderate FRED doses indicates that fire-induced mortality is more complex than the currently accepted binary scenario, where trees survive with no impacts below a certain temperature and duration threshold, and mortality occurs above the threshold.

The Cienega de Santa Clara is a 5600 ha, anthropogenic wetland in the delta of the Colorado River in Mexico. It is the inadvertent creation of the disposal of brackish agricultural waste water from the U.S. into the intertidal zone of the river delta in Mexico, but has become an internationally important wetland for resident and migratory water birds. The marsh is dominated by Typha domengensis with Phragmites australis as a sub-dominant species in shallower marsh areas. The most important factor controlling vegetation density was fire. The second significant (P<0.01) factor controlling NDVI was flow rate of agricultural drain water from the U.S. into the marsh. Reduced summer flows in 2001 due to canal repairs, and in 2010 during the YDP test run, produced the two lowest NDVI values of the time series from 2000 to 2011 (P<0.05). Salinity is a further determinant of vegetation dynamics as determined by greenhouse experiments, but was nearly constant over the period 2000 to 2011, so it was not a significant variable in regression analyses. Evapotranspiration (ET) and other water balance components were measured in Cienega de Santa Clara; we used a remote sensing algorithm to estimate ET from meteorological data and Enhanced Vegetation Index values from the Moderate Resolution Imaging Spectrometer (MODIS) sensors on the Terra satellite. We used Landsat NDVI imagery from 1978-2011 to determine the area and intensity of vegetation and to estimate evapotranspiration (ET) to construct a water balance. Remote sensing data was supplemented with hydrological data, site surveys and literature citations. The vegetated area increased from 1978 to 1995 and has been constant at about 4200 ha since then. The dominant vegetation type is Typha domingensis (southern cattail), and peak summer NDVI since 1995 has been stable at 0.379 (SD = 0.016), about half of NDVI(max). About 30% of the inflow water is consumed in ET, with the remainder exiting the Cienega as outflow water, mainly during winter months when T. domingensis is dormant.

Thesis (MA.)--Stellenbosch University, 2001.
ENGLISH ABSTRACT: There are many reasons to conserve fynbos. Not only does fynbos form part of the Cape floral kingdom, one of the richest floral kingdoms in the world, but the contribution that it makes to the regional economy through utilisation, education, recreation and tourist opportunities is immeasurable. Fire plays an integral role in fynbos ecosystems. According to Van Wilgen, Richardson & Seydack (1994: 322) " ... managing fynbos equates to managing fire". Therefore managers need accurate fire information about a fynbos area to manage it properly. This is where satellite remote sensing can provide the manager with useful information about the fire regime. In other words, satellite remote sensing can help a manager establish where and when an area has burnt. Using readily available satellite data, this study attempts to establish (through comparison) what techniques would be most suitable and affordable to compile a fire information database. Landsat Thematic Mapper data from 1990 - 1996 of the southwestern Cape was used and compared with existing fire records of the area. The results show that techniques such as supervised and unsupervised classification are reliable in identifying burnt areas, but a major drawback of these techniques is that they require a large amount of user input and knowledge. They are thus not regarded as simple or easily repeatable. - The' more simple techniques like image differencing and image ratioing were also found to be reliable in identifying burnt areas. These techniques require less user input and in some instances less data (image bands) to produce similar (or better) results than supervised and unsupervised classification techniques. The results show that differencing temporally different Images, obtained from applying principle components analysis, produces reliable results with very little confusion and little user input. Using such a technique could enable users to procure only two bands of Landsat data and still produce reliable fire information for managing a fynbos ecosystem.
AFRIKAANSE OPSOMMING: Daar is verskeie redes waarom fynbos bewaar moet word. Nie net vorm dit deel van een van die rykste blommeryke in die wereld nie, maar die bydrae wat dit tot die streeksekonomie maak, deur die benutting van veldblomme en die geleenthede wat dit bied vir toerisme en ontspanning, is enorm. Vuur speel 'n belangrike rol in die bestuur van fynbos ekosisteme. Soos beklemtoon deur Van Wilgen, Richardson & Seydack (1994: 322) se stelling: " ... managing fynbos equates to managing fire". Om hierdie rede is dit belangrik dat 'n bestuurder akkurate inligting oor die verspreiding van veldbrande moet he. Satellietafstandwaarneming kan hier 'n belangrike rol speel deur sulke inligting te verskaf Deur gebruik te maak van maklik bekombare satellietdata, poog hierdie studie om te bepaal (d.m.v. vergelyking) watter tegnieke die mees geskikte is in terme van bekostigbaarheid en gebruikersvriendelikheid. Landsat Thematic Mapper data van 1990 tot 1996 van die suidwes-Kaap is gebruik en vergelyk met bestaande branddata van die studiegebied. Daar is gevind dat tegnieke soos gerigte en nie-gerigte klassifikasie in staat is om gebrande dele betroubaar uit te ken. Hierdie tegnieke verg egter baie insette en kennis van die gebruiker en is ook nie maklik om jaar na jaar te herhaal nie. Daarom word hierdie tegnieke nie aanbeveel nie. Daar is gevind dat die eenvoudiger tegnieke soos veranderingsanalise ook gebrande dele betroubaar kon uitken. Hierdie tegnieke het die voordeel dat die gebruiker nie baie' kennis van die gebied hoef te he nie en ook nie so baie insette hoef te lewer nie. Hierdie tegnieke word bo gerigte en nie-gerigte klassifikasie aanbeveel. - Die resultate dui daarop dat betroubare resultate verkry kan word deur tempo reel verskillende beeIde, verkry deur hoofkomponentanalise, van mekaar af te trek. Hierdie tegniek vereis relatief min gebruikersinsette en daar kan selfs met slegs twee Landsat bande gewerk word. So 'n tegniek kan beslis 'n bekostigbare en effektiewe manier wees om nodige inligting vir die bestuur van 'n fynbos ekosisteem te bekom.

An accurate fire model is a useful tool in predicting the behavior of a prescribed fire. Simulation of fire requires an extensive amount of data and can be accomplished best using GIS applications. This paper demonstrates integrative procedures of using of ArcGIS™, ERDAS Imagine™, GPS, and FARSITE© to predict prescribed fire behavior on the Kaibab-Paiute Reservation. ArcGIS was used to create a database incorporating all variables into a common spatial reference system and format for the FARSITE model. ArcGIS Spatial Analyst was then used to select optimal burn sites for simulation. Our predictions will be implemented in future interagency efforts towards vegetative restoration on the reservation.

Mestrado em Engenharia Florestal e dos Recursos Naturais / Instituto Superior de Agronomia. Universidade de Lisboa
The main goal of the study is to analyse the possibility of quantifying the loss of biomass in burned forest stands using Light Detection and Ranging (LiDAR) data. Since wildfires are not uncommon in Mediterranean areas, it is useful to quantify the magnitude of fire damage in forests. With the use of remote sensing, it is possible to plan post-fire recovery management and to quantify the losses of biomass and carbon stock. Mata Nacional de Leiria (MNL) was chosen, because, after the fire in October 2017, it showed areas with low and medium-high fire severity. MNL is divided in several rectangular management units (MU). To achieve our objective, it was necessary to find a MU with burned and unburned areas. In this selection process, we used Sentinel-2 images. The fire severity was estimated by deriving a spectral index related with the effects of fire and to compute the temporal difference (pre- minus post-fire) of this index, the delta normalized burn ratio (DNBR). Forest inventory was carried out in four plots installed in the selected MU. Allometric equations were used to estimate values of stand aboveground biomass. These values were used to fit a relationship with data extracted from LiDAR cloud metrics. The LiDAR data were acquired with a VLP-16 Velodyne LiDAR PUCK™ mounted on an Unmanned Aerial Vehicles (UAV) at an altitude of 60 m above the ground. The point clouds were then processed with the FUSION software until a cloud metrics was generated and then regression models were used to fit equations related to LiDAR-derived parameters. Two biomass equations were fit, one with the whole tree metrics having a R² = 0,95 and a second one only considering the tree crown metrics presenting a R² = 0,93. The state of the forest (unburned/burned) was significant on the final equation
N/A

The primary goal of this research is to advance methods for deriving fine-grained, scalable, wildland fuels attributes in 3-dimensions using terrestrial and airborne laser scanning technology. It is fundamentally a remote sensing research endeavor applied to the problem of fuels characterization. Advancements in laser scanning are beginning to have significant impacts on a range of modeling frameworks in fire research, especially those utilizing 3-dimensional data and benefiting from efficient data scaling. The pairing of laser scanning and fire modeling is enabling advances in understanding how fuels variability modulates fire behavior and effects.

This dissertation details the development of methods and techniques to characterize and quantify surface fuelbeds using both terrestrial and airborne laser scanning. The primary study site is Eglin Airforce Base, Florida, USA, which provides a range of fuel types and conditions in a fire-adapted landscape along with the multi-disciplinary expertise, logistical support, and prescribed fire necessary for detailed characterization of fire as a physical process. Chapter 1 provides a research overview and discusses the state of fuels science and the related needs for highly resolved fuels data in the southeastern United States. Chapter 2, describes the use of terrestrial laser scanning for sampling fuels at multiple scales and provides analysis of the spatial accuracy of fuelbed models in 3-D. Chapter 3 describes the development of a voxel-based occupied volume method for predicting fuel mass. Results are used to inform prediction of landscape-scale fuel load using airborne laser scanning metrics as well as to predict post-fire fuel consumption. Chapter 4 introduces a novel fuel simulation approach which produces spatially explicit, statistically-defensible estimates of fuel properties and demonstrates a pathway for resampling observed data. This method also can be directly compared to terrestrial laser scanning data to assess how energy interception of the laser pulse affects characterization of the fuelbed. Chapter 5 discusses the contribution of this work to fire science and describes ongoing and future research derived from this work. Chapters 2 and 4 have been published in International Journal of Wildland Fire and Canadian Journal of Remote Sensing, respectively, and Chapter 3 is in preparation for publication.

Obtaining quantitative information about the recovery of fire affected ecosystems is of utmost importance from the management and decision-making point of view. Nowadays the concern about natural environment protection and recovery is much greater than in the past. However, the resources and tools available for its management are still not sufficient. Thus, attention and precision are needed when decisions must be taken. Quantitative estimates on how the vegetation is recovering after a fire can be of help for evaluating the necessity of human intervention on the fire-affected ecosystem, and their importance will grow as the problem of forest fires, climate change and desertification increase. This thesis presents a comparison of methods to extract quantitative estimates of vegetation cover regrowth in burned areas with remote sensing data. In order to eliminate possible sources of error, a thorough pre-processing was carried out, including a careful geometric correction (reaching RMSE lower than 0.3 pixels), a topographic correction by means of a constrained Minneart model and a combination of absolute and relative atmospheric correction methods. Pseudo Invariant Features (PIF) were identified either by visual inspection methods or by a new automated selection method based in temporal Principal Component Analysis (PCA), which has been called multi-Temporal n-Dimensional Principal Component Analysis (mT-nD-PCA). This automated method demonstrated its capability in selecting accurate and objective PIFs within the satellite images. Spectral Mixture Analysis (SMA) was compared against quantitative vegetation indices which are based on well known traditional vegetation indices like Normalised Difference Vegetation Index (NDVI) and Modified Soil Adjusted Vegetation Index (MSAVI). Accuracy assessment was performed by regressing vegetation cover results obtained with each method, against field data gathered during the field work campaigns. Results obtained showed how vegetation cover fractions obtained with the NDVI based quantitative index were the most accurate, being superior to the rest of the techniques applied, including SMA.

La mayoría de los bosques del planeta, exceptuando, quizá, el cinturón más húmedo del trópico, han sufrido perturbaciones recurrentes por incendios, desde hace miles de años. Sin embargo, en el último siglo, la combinación de factores socio-económicos y ambientales han alterado la frecuencia y distribución de incendios en casi todos los ecosistemas forestales. Esta mala distribución del fuego ha conllevado la acumulación de este elemento perturbador en ecosistemas poco adaptados a su presencia (e.g. bosques tropicales húmedos), mientras que otros ecosistemas han sufrido los efectos de las políticas de supresión de incendios (e.g. bosques templados norteamericanos). Entre las consecuencias de estas alteraciones, algunas se refieren a variaciones en el régimen de incendios, con implicaciones en términos de conservación forestal, impacto ecológico, económico y social. La presente tesis analiza la influencia de factores ambientales y socio-económicos en algunas características del régimen de incendios en ecosistemas tropicales y mediterráneos. En el ámbito tropical, la zona de estudio es el Estado mexicano de Chiapas, y los objetivos son básicamente dos: 1) caracterizar el régimen de incendios en el Estado, en términos de frecuencia de incendios, distribución, áreas y ecosistemas afectados, tipología de incendios, causalidad, y su interacción con algunos factores como el fenómeno de El Niño, la distribución de la tenencia de la tierra o la variación de las precipitaciones. 2) Analizar la influencia relativa de factores ambientales frente a factores socio-económicos, en años de condiciones climáticas normales y años de El Niño. El objetivo último era determinar variaciones en la afectación de los principales ecosistemas y listar los factores más importantes a considerar, a nivel de gestión forestal contra incendios. A nivel de ecosistemas mediterráneos, se seleccionó el incendio del Solsonés 1998, como caso estudio, siendo los principales objetivos: 3) Seleccionar la metodología de clasificación post-incendio más eficiente entre varias técnicas analizadas (teledetección y muestreo de campo). 4) Analizar la influencia de diversos factores ambientales en la formación de islas de vegetación en un gran incendio, empleando las laderas como unidad espacial. 5) Verificar la importancia de la estructura forestal y la continuidad del combustible en la heterogeneidad final de las severidades de afectación de un incendio. En relación a estos objetivos, las principales aportaciones de este estudio han sido, en lo referente a ecosistemas tropicales: La descripción de los incendios en Chiapas como mayoritariamente superficiales, afectando principalmente a estratos no arbóreos, de grandes dimensiones, de origen antrópico y con cierta propensión a afectar zonas protegidas. La aportación clave sin embargo, hace referencia a la demostración de la importancia de El Niño en la afectación de los ecosistemas tropicales húmedos, principalmente influenciado por la presencia de pastos alrededor de bosques fragmentados. El cambio de importancia relativa de los factores ambientales en años de no-ENSO frente a la mayor importancia de los factores socio-económicos en años de ENSO. También se puede remarcar el cambio de ecosistema afectado condicionado por el tipo de año (condiciones climáticas normales o extremas, ENSO o no ENSO: pino-encino versus bosques tropicales húmedos). En relación a los bosques mediterráneos, las técnicas más sencillas de teledetección resultaron las adecuadas para la identificación de islas. La formación de estas islas se ve condicionada por el tamaño y orientación y pendiente de las laderas, así como la cantidad y madurez de los rodales afectados, siendo las laderas más extensas, de orientación noroeste, de mayor pendiente, con mayores extensiones forestales y bosques más maduros (en términos de mayores tamaños), las más propicias a formar islas. Las características estructurales de los rodales fue de gran importancia para predecir la supervivencia forestal en zonas de condiciones climáticas moderadas.
The great majority of the forests of the world, excepting perhaps, the wettest belt of the tropics, have been burned over, at more or less frequent intervals, for many thousands of years. In the last century however, a combination of human and environmental factors have altered the frequency and distribution of fire, almost everywhere. This maldistribution of fire has resulted in an accumulation of this element in ecosystems not adapted to high frequencies of fire (i.e. tropical and sub-tropical areas), while other ecosystems have suffered the effects of fire suppression policies (i.e. north American temperate forests). Among the consequences of these alterations, some of them refer to variations in fire regimes, with implications in terms of forest conservation, ecological, economic and social impact. The present thesis analyses the influence of environmental and socio-economic factors for certain elements of the fire regime in tropical and Mediterranean ecosystems. In tropical areas, the study case relates to the tropical Mexican State of Chiapas, and the objectives are basically two: 1) to characterize the fire regime in the State, in terms of frequency of fires, fire distribution, areas and ecosystems affected, type of fires, causality, and their interaction with some factors like the phenomenon of El Niño, the land distribution, or the variation of rainfall values. 2) To analyse the relative influence of environmental factors versus socio-economic factors, in years of normal climatic conditions and years of El Niño. The final objective is to detect variations in the affectation of the major ecosystems and to list the most important factors to consider for forest fire management. For the Mediterranean ecosystems, the fire of Solsonés 1998 was selected as case study. Main objectives are: 3) To select the most efficient methodology for post-fire classification among several techniques: image classification and field survey. 4) To determine the influence of diverse environmental factors in the formation of vegetation islands inside a large forest fire, using "slope", as the spatial unit. 5) To verify the importance of forest structure and fuel continuity in the final heterogeneity of fire severities in a large fire. In relation to these objectives, the main contributions of this study are, with respect to tropical ecosystems: The characterization of fires in Chiapas, as in other tropical areas, mainly as superficial fires, frequently affecting non-arboreal layers. Major problems focus on large forest fires, which are responsible for the major burned areas. Fires are mainly human related and display certain propensity to affect protected zones. The key contribution nevertheless, refers to the major importance of El Niño in the affectation of humid tropical ecosystems. This is specially enhanced by the presence of cattle pasture surrounding the fragmented forests. The shift on the importance of environmental factors in years of no-ENSO versus the greater importance of socio-economic factors in years of ENSO, is also a major achievement. Moreover, shifts in ecosystem affectation have also been observed depending on the climatic conditions: pine-oak in no-ENSO years versus humid tropical forests in ENSO years. Regarding the Mediterranean forests, the simplest techniques of teledetection were the best performing ones for island identification. The formation of these islands is conditioned by the size and aspect of the slopes, as well as by the amount and maturity of the affected patches. Most extensive slopes, north-western orientations, more abrupt, larger forest extensions per slope, and more mature forests (in terms of greater sizes), are the most suitable combination of factors to form islands. This formation of islands does not have, therefore, a random distribution. Besides, the structural characteristics of the forest patches was confirmed to be of great importance to predict the forest survival in areas were fire burned under moderate climatic conditions. Larger patches and larger trees are the ones that will survive fire the best.

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 & subtropical grasslands, savannas & 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 & 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 & 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.

QC 20210525

Doctor of Philosophy
Department of Horticulture, Forestry, and Recreation Resources
Charles J. Barden
Recurrent, landscape-level fires played an integral part in the development and persistence of eastern oak (Quercus spp.) forests of the United States. These periodic surface fires helped secure a competitive position for oaks in the regeneration pool by maintaining a desirable species composition and forest structure. This historical fire regime was altered with the European settlement of North America, and fire suppression within forestlands became a standard practice since 1930s. With decades of fire suppression, mature oak-dominated woodlands have widely converted to shade-tolerant tree species. Prescribed fire has successfully been used to enhance oak regeneration in eastern forests. However, oak woodland restoration within the forest-prairie ecotone of the Central plains has not been systematically studied. Fuel beds under shade-tolerant species are often less conducive to fire. Therefore, monitoring fuel loading (FL) and its changes are essential to inform management decisions in an oak regeneration project. Rapid expansion of eastern redcedar (Juniperus virginiana/ERC) is another ecological issue faced by land managers throughout North America’s midcontinent forest-prairie ecotone. Hence, it is worthy to monitor ERC expansion and effects on deciduous forests, to inform oak ecosystem restoration interventions within this region. Therefore, the main objectives of this dissertation were three-fold: (1) understand the effects of prescribed burning and mechanical thinning to encourage oak regeneration; (2) investigate the initial effects of an oak regeneration effort with prescribed fire and mechanical thinning on FL; and (3) monitor the spatio-temporal dynamics of ERC expansion in the forest-prairie ecotone of Kansas, and understand its effects on deciduous forests. The first two studies were conducted on a 90-acre oak dominated woodland, north of Manhattan, Kansas. The experimental design was a 2 (burn) x 2 (thin) factorial in a repeated measures design. The design structure allowed four treatment combinations: burn only (B), thin only (T), burn and thin combined (BT), and a control (C). Burning and thinning treatments were administered in spring 2015. Changes in the FL estimates after the burn treatment revealed that the BT treatment combination consumed more fuel and burned more intensely compared to the B treatment. This observation was reflected in vegetation responses. The thinning reduced the canopy cover significantly, but under enhanced light environments, both oaks and competitive species thrived when no burn was incorporated. In contrast, burn treatments controlled the competitive vegetation. Hence, the most promising results were obtained when both fire and thinning were utilized. The remote sensing study documented the expansion of ERC in three areas of eastern Kansas over 30 years. The use of multi-seasonal layer-stacks with a Support Vector Machines (SVM) supervised classification was found to be the most effective approach to map ERC distribution. Total ERC cover increased by more than 6000 acres in all three study areas investigated in this study between 1986 and 2017. Much of the ERC expansion was into deciduous woodlands. Therefore, ERC control measures should be incorporated into oak woodland restoration efforts within the forest-prairie ecotone of Kansas.

Thesis (Ph. D.) -- University of Maryland, College Park, 2007.
Thesis research directed by: Geography. Title from t.p. of PDF. Includes bibliographical references. Published by UMI Dissertation Services, Ann Arbor, Mich. Also available in paper.

The aim of this work was to use remote sensing data from the MODIS instrument of the Terra satellite to detect bush fires in Australia. This included preprocessing the demodulator output, bit synchronization and reassembly of data packets. IMAPP was used to do the geolocation and data calibration. The fire detection used a combination of fixed threshold techniques with difference tests and background comparisons. The results were projected in a rectangular latidue/longitude map to remedy the bow tie effect. Algorithms were implemented in C and Matlab. It proved to be possible to detect fires in the available data. The results were compared with fire detection done done by NASA and fire detections based on other sensors and found to be very similar
Das Ziel dieser Arbeit war die Nutzung von Fernerkundungsdaten des MODIS Instruments an Bord des Satelliten Terra zur Erkennung von Buschfeuern in Australien. Das schloss die Vorverarbeitung der Daten vom Demodulator, die Bitsynchronisation und die Umpacketierung der Daten ein. IMAPP wurde genutzt um die Daten zu kalibrieren und zu geolokalisieren. Die Feuererkennung bedient sich einer Kombination von absoluten Schwellwerttests, Differenztests und Vergleichen mit dem Hintergrund. Die Ergebnisse wurden in eine rechteckige Laengen/Breitengradkarte projiziert um dem BowTie Effekt entgegenzuwirken. Die benutzten Algrorithmen wurden in C und Matlab implementiert. Es zeigte sich, dass es moeglich ist in den verfuegbaren Daten Feuer zu erkennen. Die Ergebnisse wurden mit Feuererkennungen der NASA und Feuererkennung die auf anderen Sensoren basieren verglichen und fuer sehr aehnlich befunden

Fire is a natural and essential component of forests in western North America. Fire maintains biodiversity through the creation of different habitat types, and regular fire rotations reduce the accumulation of woody fuels and thick understory plant densities that give rise to catastrophic fire. The practice of fire exclusion has altered western forests and increased the risk of widespread change under rising temperatures projected for the 21st century. To manage for the reintroduction of fire it is critical that we understand the interactions between fire and forest biota in recently fire-suppressed forests. In Chapter 2, I studied the formation and impact of small-scale fire refugia. Fire refugia are areas within burned forest that experienced relatively little change, and are recognized as important places that offer protection for forest biota (vegetation, wildlife) during and after the fire. Very few studies, however, have examined small-scale fire refugia despite their importance to many organisms (e.g., small mammals, understory plants). In a long-term forest monitoring plot in Yosemite National Park, I mapped all unburned areas ≥ 1 m2 the first year after fire. I found small fire refugia were abundant, somewhat predictable, and fostered increased survival and diversity of nearby plant life. My results suggest that small fire refugia are an important component of burned forests that should be included in management considerations. In Chapter 3, I examined possible fisher habitat in burned areas. Fishers are forest carnivores of high conservation concern due to widespread declines since European settlement and the risk of habitat loss due to fire. An isolated population remains in the Sierra National Forest, where managers are weighing the need to reintroduce fire against possible detrimental impacts to current habitat. My research examined the forest structural characteristics (vegetative cover, heights of different forest layers) surrounding fisher dens. I found suitable thresholds of these structural characteristics in recently burned areas in Yosemite, particularly after low-severity fire. My results suggest that burned areas may offer suitable denning habitat for fishers, though more research is needed to determine if this conclusion holds for all fisher activities (e.g., foraging, resting) and scales of selection.

Forest structure is an important indicator of ecosystem function and carbon storage in above-ground biomass, yet observations of forest structure are scarce across Canada’s unmanaged bo-real. To reduce uncertainties in global carbon budgets, an improved understanding of spatial and temporal variability in forest structure is required across unmanaged boreal forests. The objective of this dissertation is to investigate how fire history and forest productivity together shape the structure of Canada’s boreal forests, and to develop methods to assess these relationships over large forested areas. Transects of airborne light detection and ranging (lidar) data, totaling 25,000 km in length, were collected across northern Canada in 2010, providing a unique opportunity to study spatial varia-bility in forest structure. To elucidate on the relationships between fire, productivity, and struc-ture, lidar measures of forest structure were combined with optical satellite indicators of disturb-ance history and forest productivity. Specifically, a 25-year chronosequence of forest regenera-tion following fire was developed, and the relationship between forest structure and productivity was assessed as a function of time since fire. In addition, the relationship between structure and productivity was assessed in stands with no recorded disturbances. Satellite-derived estimates of forest productivity were an important predictor of early stand de-velopment following fire, as lidar-derived estimates of canopy cover varied strongly along re-gional gradients of productivity after 15 years following fire (r = 0.63 – 0.72, p < 0.01). Additionally, pre-disturbance conditions were a strong indicator of stand development following fire, as patches classified as dense forest (> 50% canopy cover) prior to burning displayed faster growth and recovery compared to patches classified as open forest (20 – 50% canopy cover). Further, this research highlights the importance of monitoring multiple aspects of forest recovery, as lidar-derived estimates of canopy cover and stand height showed contrasting relationships to productivity in recently burned stands (1985 – 2009) as well as in stands with no recent disturbance. The results of this dissertation demonstrate the value of the airborne lidar transects for describing stand-level variability in forest structure over large areas, and demonstrate the need for lidar to validate wall-to-wall indicators of disturbance, productivity, and structure.
Forestry, Faculty of
Graduate

Biological invasions are driving environmental state changes on a global scale. Exotic plant species must be successful at passing several abiotic and biotic filters to establish and disrupt the native plant community assembly. Understanding where exotic plants are on a regional scale and being able to characterize how exotic plants are generally interacting with their environment is crucial information for exotic species management (chapter 1). In the western United States human-related activities are augmenting the spread of exotic plant species by increasing the ignitions of wildfire. Wildfire can lead to nutrient pulses through the removal of intact native communities and returning some mineral content into the soil. Exotic plant species that have traits that efficiently acquire nutrients accompanied by rapid growth rates may outcompete native plants. In chapters 2, 3, and 4 experimental fires demonstrated that the direct effect of fire may not be as critical as the potential indirect effects of fire such as altering the behavior of consumers (chapter 2) and reducing competition (chapters 3 and 4). In the Mojave desert, rodent consumers can have strong top-down effects on plant community assembly through foraging selection preferences. Life history traits such as seedling and seed size can lead to differential herbivory and positively benefit some plant species while inhibiting others (chapter 1) which could indirectly alter plant-plant interactions. Plant competition is a biotic filter than can determine establishment success or failure. Species that with rapid growth rates and plastic growth responses are likely to be able to capitalize on fluctuations in available resources. In the Great Basin, forecasts in climate change models predict that precipitation timing will lead to heavier fall rains and more rain than snow in the winter. Water availability is one of the main limiting factors in semi-arid and arid ecosystems where native plants have adaptive traits to maximize resource use. The interaction of wildfire and changes in climate, specifically timing of precipitation is critical to understand to be able to predict and protect against increasing wildfire frequency and severity. In chapter three, the responses by a key exotic annual grass, Bromus tectorum, and keystone native perennial shrub Artemisia tridentata subsp. wyomingensis, were positive for increased early fall precipitation but much more pronounced for B. tectorum. Exotic annual plants are able to respond to changes in timing of fall precipitation and have extreme growth which leads to superior competitive abilities through interference and priority effects (chapter 4). Native plants can compete with exotics but the magnitude of the effects are diminished compared to the negative interaction from exotics. Together these findings demonstrate that across several regions exotic annual grasses are capable of passing through abiotic filters and disrupting biotic interactions of the native plant community. This is likely to lead to increased spread of exotic annual species and may indicate potential and availability of fine fuel production supporting increases in size and frequency of wildfires in the western United States.

The goal of this study is to develop a smoke detecting algorithm using digital image processing techniques on multi-spectral (visible & infrared) video. By utilizing principal component analysis (PCA) followed by spatial filtering of principal component images the location of smoke can be accurately identified over a period of exposure time with a given frame capture rate. This result can be further analyzed with consideration of wind factor and fire detection range to determine if a fire is present within a scene. Infrared spectral data is shown to contribute little information concerning the smoke signature. Moreover, finalized processing techniques are focused on the blue spectral band as it is furthest away from the infrared spectral bands and because it experimentally yields the largest footprint in the processed principal component images in comparison to other spectral bands. A frame rate of .5 images/sec (1 image every 2 seconds) is determined to be the maximum such that temporal variance of smoke can be captured. The study also shows eigenvectors corresponding to the principal components that best represent smoke and are valuable indications of smoke temporal signature. Raw video data is taken through rigorous pre-processing schemes to align frames from respective spectral band both spatially and temporally. A multi-paradigm numerical computing program, MATLAB, is used to match the field of view across five spectral bands: Red, Green, Blue, Long-Wave Infrared, and Mid-Wave Infrared. Extracted frames are aligned temporally from key frames throughout the data capture. This alignment allows for more accurate digital processing for smoke signature. v Clustering analysis on RGB and HSV value systems reveal that color alone is not helpful to segment smoke. The feature values of trees and other false positives are shown to be too closely related to features of smoke for in solely one instance in time. A temporal principal component transform on the blue spectral band eliminates static false positives and emphasizes the temporal variance of moving smoke in images with higher order. A threshold adjustment is applied to a blurred blue principal component of non-unity principal component order and smoke results can be finalized using median filtering. These same processing techniques are applied to difference images as a more simple and traditional technique for identifying temporal variance and results are compared.

Since 1990, thirteen fires over 100,000 acres in size have burned in California seven of which were recorded to be some of the most destructive wildfires of all time (California Department of Forestry & Fire Protection 2013). To aid the development of policy that reduces the destruction caused by wildfires, it is important to evaluate how risk changes through time in communities that are expanding into fire-prone areas. The objective of this study is to discover how the likelihood of structural loss is changing in WUI as newer; more fire resilient structures replace older structures on the edges of the WUI. Geographical Information Systems and remote sensing techniques were used to observe changes in urbanization, structural materials, housing density and defensible space over time in the communities of Rancho Santa Fe, Ramona and Julian in San Diego County. Fire Risk ratings were calculated using the equation Fire Risk= Hazard – Mitigation. Mitigation scores for each structure were informed using a binary logistic regression of variables influencing home loss in the Witch Creek Fire. Fire Risk Ratings were given to the 11,747 structures in the three communities for the years 2005, 2009, 2010 and 2012. The study found that the initial 0-1.5m zone around the home is the most critical for defensible space. In this zone, increased tree cover increases the odds of structure loss by over double that of grass cover. In Rancho Santa Fe and Julian, the majority of very high risk homes were located in high income communities despite moderate mitigation due to very high fire hazard levels. In Ramona most very high fire risk homes were located in lower income areas due to poor mitigation levels. Rancho Santa Fe and Julian decreased their fire risk over the 7 year study period with improved mitigation, Rancho Santa Fe improved the most (1.7% decrease in Very High and High risk homes). The proportion of very high fire risk homes increased in Ramona by .5% over the 7 year study period. Development on the outskirts of the WUI could increase the risk of the overall community if proper construction standards are not met and defensible space is not implemented. If fire resistant communities are constructed and maintained to high standards of defensible space, they could potentially provide a buffer for older high fire risk homes.

Geographic Information Systems (GIS) are increasingly used to examine fire activity. This study uses GIS to determine fire likelihood probabilities at an intermediate scale ( I-kilometer) on a daily basis given readily available data. Layers used for the analysis included slope, aspect, elevation, fuel type, proximity to existing fires, maximum temperature, minimum temperature, relative humidity, average vapor pressure deficit, precipitation, 1-hour fuel moisture, and 10-hour fuel moisture. There were three objectives of this study: 1. Establish a correlation between burn perimeters and readily available topographic and environmental data, and map the spatial distribution of these as fire likelihood areas; 2. Compare each day's fire likelihood areas to fire perimeters from the next day to determine to what extent areas deemed to be high fire likelihood on a given day could be used to identify likelihood areas for the subsequent day; 3. Create a generalized model using the fewest and most frequent significant variables and test this model as a general predictive tool for fire likelihood over a given season. Redundant variables and variables determined not to be significant at this scale were removed from the model. Variables that best explained the fire activity were identified and used to spatially map fire likelihood for any given day. By comparing subsequent days fire activity to the previous days fire likelihood areas, it was determined that the previous fire likelihood areas can be used as an indication for the subsequent day' s fire likelihood areas with a reasonable level of accuracy. Although factors changed from day to day, the most significant variables tended to be slope, elevation, fuel type, and 1-hour fuel moisture. These variables were incorporated into a generalized model which, when mapped spatially, provided a method to compare increasing or decreasing levels of fire likelihood on a temporal scale. The results were coarser, but still indicated that a generalized model could be used to identify the next day' s fire likelihood areas given the previous days spatial plots. When compared to new fire starts, fires occurred in areas of moderate fire likelihood probabilities and very few occurred in areas of low probability.

Mestrado em Geomateriais e Recursos Geológicos
Remote Sensing has been used for decades, and more and more applications are added to its repertoire. With this study we aim to show the use of Remote Sensing in the field of vegetation recovery monitoring in burned areas and the added value of data with a high spatial resolution. This was done by analysing both Landsat 7 and 8 scenes, after the forest fire of summer 2012 in the parish of Calde, in the central region of Portugal, as well as an orthophoto produced with images acquired by an unmanned aerial vehicle.
A Deteção Remota tem sido utilizada durante décadas com novas aplicações a surgirem constantemente. Com este estudo pretende-se demonstrar o uso da Deteção Remota no campo da monotorização da recuperação de vegetação em áreas ardidas e o valor acrescentado da elevada resolução espacial dos dados utilizados. Para o efeito, foi feita a análise de áreas ardidas na freguesia de Calde, região central de Portugal, depois do incêndio florestal no verão de 2012, usando imagens Landsat 7 e 8 assim como uma ortofoto produzida com imagens adquiridas por um veículo aéreo não tripulado.

O uso de técnicas de sensoriamento remoto além de auxiliar na detecção, monitoramento ou identificação de áreas afetadas por queimadas e incêndios florestais, permite realizar análises voltadas para caracterização e avaliação de ocorrências de fenômenos numa paisagem em diferentes escalas temporais/espaciais. O presente estudo tem como objetivo analisar as áreas afetadas por queimadas e incêndios florestais no ano 2015, nas bacias e sub-bacias dos sistemas hidrográficos Araguaia-Tocantins no estado do Tocantins. A quantificação e delimitação de áreas afetadas por queimadas e incêndios florestais foi realizada com uso duas composições RGB (754; 653) oriundas de imagens do satélite Landsat 8 que correspondem ao período de seca no Estado. Além das variáveis relacionadas a caracterização de área afetada por queimadas e incêndios florestais em cada um dos 30 sistemas hidrográficos do estado do Tocantins, foram utilizadas métricas de paisagem, número de focos de calor e resultado da autocorrelação espacial nas análises de correlação e componentes principais. A área afetada por queimadas e incêndios florestais apresentou correlação positiva com as seguintes variáveis: área ocupada pelo sistema (0,72**; 0,87**); percentual de área queimada (0,78**; 0,86**). Por outro lado, foram observados valores negativos com o percentual da área queimada dentro de áreas agrícolas (-0,55**; -0,80**). A análise de correlação, ainda, revelou correlação da área afetada por queimadas e incêndios florestais com o número de focos de calor (0,75**; 0,90**) e números de focos de calor dentro de áreas queimadas (0,95**; 0,97**). Com menores magnitudes, também, foram observadas correlações significativas com o número de dias sem chuva (0,51**; 0,63**) e número de dias com chuva menor que 13 mm (0,55**; 0,81**). Com relação a análise de componentes principais, foi possível extrair quatro componentes com primeiro explicando 44,05% da variância, o segundo, 16,69%, o terceiro, 12,67% e o quarto 9,66%, o que totaliza 83,03% da variabilidade do conjunto de dados original. Em nove das trinta sub-bacias do sistema hidrográfico do estado foram observadas área afetada por queimadas e incêndios florestais acima de 100 mil hectares no ano de 2015: A2 (> 501 mil ha); T1 (> 406 mil ha); T10 (> 248 mil ha); A1 (> 199 mil ha); A3 (> 176 mil ha); T12 (> 126 mil ha); T9 (> 117 mil ha); T5 (> 113 mil ha) e; T14 (> 101 mil ha), o que totaliza mais de 77% do mapeado no estado. As métricas de paisagem apresentaram as menores relações com a área afetada por queimada e incêndios florestais, no entanto, no sistema de segunda maior ocorrência da área queimada houve relação com maior fragmentação da paisagem. Os sistemas que apresentaram maior ocorrência de áreas queimadas podem ser caracterizados tanto por paisagens regulares quanto por formas mais complexas da paisagem, além da ocorrência agrupada do fogo.
The remote sensing techniques can be used for the detection, monitoring or identification of burned areas, as well for the analysis of the characterization and evaluation of occurrences of phenomena in a landscape at different temporal / spatial scales. The present study aims to analyse burned areas in the 2015 season, in the basins and sub-basins of the Araguaia-Tocantins hydrographic systems in the state of Tocantins. The quantification and delimitation of burned areas were carried out using two RGB compositions (754; 653) derived from Landsat8 satellite images and dry season. In addition to the variables related to a characterization of the burned area in each of the 30 hydrographic systems in the state of Tocantins, landscape metrics, number of hotspots and the results of the spatial autocorrelation were used too in analysis of correlation and principal components. The burned area had shown a positive correlation with the following variables: area occupied by the system (0.72 **; 0.87**) and the percentage of burned area (0.78**; 0.86 **). However, negative values were observed with the percentage of the burned area within the agricultural areas (-0.55 **; -0.80**). The correlation analysis also showed the relation of the burned area with the number of hotspots (0.75 **; 0.90**) and number of hotspots within the burned areas (0.95**; 0.97**). With smaller magnitudes, significant correlations were observed with the number of days without rain (0.51**; 0.63**) and number of days with rain less 13 mm (0.55 **; 0.81**). About the principal components analysis, it was possible to extract four components with first detail 44.05% of the variance, the second 16.69%, the third, 12.67% the fourth 9.66%, which they were represent the 83.03% of the variance of the original dataset variability. In nine sub-basins of the state were observed burned areas above 100 thousand hectares in 2015: A2 (> 501,000 ha); T1 (> 406,000 ha); T10 (> 248,000 ha); A1 (> 199,000 ha); A3 (> 176,000 ha); T12 (> 126,000 ha); T9 (> 117,000 ha); T5 (> 113,000 ha); T14 (> 101,000 ha), which they were represent more than 77% of the mapped in the state. The landscape metrics had shown the small relationships with burned area, however, in the system with the second largest occurrence of the burned area was observed relationship with the fragmentation of the landscape. The systems with the highest occurrence of burned areas can be characterized by regular or more complex landscapes, as well as the grouped occurrence of fire.

Remote sensing has been demonstrated an important tool in agricultural and natural resource management and research applications, however there are limitations that exist with traditional platforms (i.e., hand held sensors, linear moves, vehicle mounted, airplanes, remotely piloted vehicles (RPVs), unmanned aerial vehicles (UAVs) and satellites). Rapid technological advances in electronics, computers, software applications, and the aerospace industry have dramatically reduced the cost and increased the availability of remote sensing technologies.Remote sensing imagery vary in spectral, spatial, and temporal resolutions and are available from numerous providers. Appendix A presented results of a test project that acquired high-resolution aerial photography with a RPV to map the boundary of a 0.42 km2 fire area. The project mapped the boundaries of the fire area from a mosaic of the aerial images collected and compared this with ground-based measurements. The project achieved a 92.4% correlation between the aerial assessment and the ground truth data.Appendix B used multi-objective analysis to quantitatively assess the tradeoffs between different sensor platform attributes to identify the best overall technology. Experts were surveyed to identify the best overall technology at three different pixel sizes.Appendix C evaluated the positional accuracy of a relatively low cost UAV designed for high resolution remote sensing of small areas in order to determine the positional accuracy of sensor readings. The study evaluated the accuracy and uncertainty of a UAV flight route with respect to the programmed waypoints and of the UAV's GPS position, respectively. In addition, the potential displacement of sensor data was evaluated based on (1) GPS measurements on board the aircraft and (2) the autopilot's circuit board with 3-axis gyros and accelerometers (i.e., roll, pitch, and yaw). The accuracies were estimated based on a 95% confidence interval or similar methods. The accuracy achieved in the second and third manuscripts demonstrates that reasonably priced, high resolution remote sensing via RPVs and UAVs is practical for agriculture and natural resource professionals.

The capability to monitor bushfires on a large scale from space has long been identified as an important contribution to climate and atmospheric research as well as a tool an aid in natural hazard response. Since the work by Dozier (1981), fire monitoring from space has relied on the principles he described. His method of identifying fires within a pixel significantly larger than the fire by utilizing the different responses of the 3 μm and 11 μm channels has been applied to a number of sensors. Over the last decade a lot of work has been invested to refine and validate fire detections based on this approach. So far, the application of the method proposed by Dozier (1981) reached its peak with the launch of the MODIS instrument on board the Terra satellite. In contrast to earlier sensors, MODIS was equipped with spectral channels specifically designed for the detection of fires with algorithms based on the work by Dozier (1981). These channels were designed to overcome problems experienced with other platforms, the biggest of which is the saturation of the 3 μm channel caused by big, hot fires. Since its launch, MODIS has proven itself to be a capable platform to provide worldwide fire detection at a moderate resolution of 1 km on a daily basis.It is the intention of this work to open up new opportunities in remote sensing of fires from satellites by showing capabilities and limitations in the application of other spectral channels, in particular the 2.1 μm channel of MODIS, than the ones currently used. This channel is chosen for investigation as fires are expected to emit a significant amount of energy in this bandwidth and as it is available at a native resolution of 500 m on MODIS; double the resolution of the 3 μm and 11 μm channels. The modelling of blackbodies of typical bushfire temperatures shows that a fire detection method based on the 2.1 μm channel will not be able to replace the current methods. Blackbodies of temperatures around 600 to 700 K, that are common for smoldering fires, do not emit a great amount of energy at 2.1 μm. It would be hardly possible to detect those fires by utilizing the 2.1 μm channel. The established methods based on the 3 μm and 11 μm channels are expected to work better in these cases. Blackbodies of typically flaming fires (above 800 K) however show a very high emission around 2.1 μm that should make their detection using the 2.1 μm channel possible.In order to develop a fire detection method based on the 2.1 μm channel, it is necessary to differentiate between the radiance caused by a fire of sub pixel size and the radiance of a pixel caused by the reflection of sunlight. This is attempted by using time series of past observations to model a reflectance value for a given pixel expected in absence of a fire. A fire detection algorithm exploiting the difference between the expected and observed reflectance is implemented and its detection results are compared to high resolution ASTER fire maps, the standard MODIS fire detection algorithm (MOD14) and burnt area maps. The detections of the method based on the 2.1 μm channel are found to correspond very well with the other three datasets. However, the comparison showed detections that do not align with MOD14 active fire detections but are generally aligned with burn areas. This phenomena has to be investigated in the future.

I used Moderate-Resolution Imaging Spectroradiometer (MODIS) imagery to answer two ecological questions. In the first project, I investigated the interactions between bark beetle-caused tree mortality and fire occurrence in western U.S. forests. I used remotely sensed fire data detected by MODIS satellite and bark beetle-caused tree mortality data. I tested the hypothesis that there is an increased probability of fire incidence in bark beetle-damaged forests compared to healthy forests using conditional probability modeling across the national forests of the western U.S. regardless of forest type. My results did not show a consistent pattern (increase or decrease of conditional probability of fire occurrence, &#;CP) across all lag time periods considered. However, when &#;CP is averaged across the 5-year study period (2001-2005) fire probability increased at 2-year (16%) and 3-year (9%) lags with 0, 1, 4, and 5-year lags showing no positive effect of bark beetle activity on fire probability. Further, when I analyzed fire-bark beetle-caused tree mortality separately for persistent fires (fires that lasted for at least two 8-day composite periods per season) and transient fires (fires that lasted for only one 8-day composite period per season), the &#;CP increased in all lag periods except the 5-year lag for persistent fires. In the second stage of this project, I used a non-parametric modeling approach to test how important bark beetle-caused tree mortality is in influencing fire occurrence relative to other climate and topography-derived variables in spruce-fir, Douglas-fir, lodgepole, and ponderosa pine forests. My results showed that climate and topography-derived predictors were consistently selected as important predictors of fire occurrence while bark beetle-caused tree mortality showing the least importance. In the second project, I predicted the invasive potential of a Puerto Rican frog species in Hawaii using the following MODIS products: land surface temperature; normalized difference vegetation index and enhanced vegetation index; and leaf area index/fraction of photosynthetically active radiation absorbed by plant canopies. My predicted maps showed that the invasive frog species in Hawaii is likely to expand its current habitat. My results also showed that MODIS-derived biophysical variables are able to characterize the suitable habitats of the invasive frog species.

Thesis (MSc)--Stellenbosch University, 2013.
ENGLISH ABSTRACT: Current efforts to address Zimbabwe‘s decade long veld fire crisis has partly been hindered by a lack of financial resources and fire data. This study illustrates the potential of using the MODIS burned area product as an alternative cost- and time-effective method for reconstructing historical fire records in Zimbabwe. Two MODIS burned area products were evaluated, namely the MCD45A1 and WAMIS (Meraka Institute‘s MODIS burned area product). Both products yielded similar levels of accuracy despite the difference in algorithms. However, it is assumed that at certain thresholds, either in tree cover or fire intensity, WAMIS ceases to map fires as accurately as the MCD45A1. Ten years of fire data for Charara Safari Area (CSA) was extracted from the MCD45A1, and used as a basis to establish six parameters: fire incidence, extent, seasonality, fire size, frequency and fire return interval (FRI). It was observed that approximately 50% of CSA burned annually, with an average of 132 fires occurring every year. Although there was no overall increase or decrease in the extent of area burned over the 10 year study period, an increasing trend in fire incidence was noted. Through an assessment of effective fire size, it was established that more fires in CSA were gradually becoming smaller in size, while the extent of area burned remained fairly constant. Hence, the increase in fire incidences and lack of a corresponding increase in area burned. This study was also used to identify areas in the fire regime that may be a potential ecological risk to the miombo woodland in CSA. Three points of concern were revealed: firstly, a high prevalence of late season fires was observed in the northern bounds of CSA. Secondly, 64.2% of the total area burned in CSA burned between 6 and 10 times over the 10 year period, and lastly, 85% of the total area burned over the period 2001 and 2010 had a FRI of less than 2 years. The combination of late season fires, high fire frequency and short FRI in CSA is indicative of possible alterations in the state of the miombo woodlands, which may have negative socio-economic implications on CSA and its surrounding communities. This study has demonstrated that the MCD45A1 is a useful source of much needed fire information for Zimbabwe. Therefore, the possibility of integrating methods employed in this study into the current collection of fire data should be given due consideration.
AFRIKAANSE OPSOMMING: Huidige pogings om Zimbabwe se dekade lank veldbrand krisis aan te spreek is gedeeltelik belemmer deur 'n gebrek aan finansiële hulpbronne en vuurdata. Hierdie studie illustreer die potensiaal van die gebruik van die MODIS verbrande area produk as 'n alternatiewe koste-en tyd-effektiewe metode vir die rekonstruksie van historiese vuurrekords in Zimbabwe. Twee MODIS verbrande area produkte is geëvalueer, naamlik die MCD45A1 en WAMIS (Meraka Instituut se MODIS verbrand area produk). Beide produkte het soortgelyke vlakke van akkuraatheid opgelewer ten spyte van die verskil in die algoritmes. Dit word egter aanvaar dat op sekere drempels, óf in die boom bedekking, of brandintensiteit, WAMIS brande minder akkuraat karteer as die MCD45A1 produk. Tien jaar van vuurdata vir Charara Safari Area (CSA) is uit die MCD45A1 data onttrek, en gebruik as 'n basis om ses parameters vas te stel: vuurvoorkoms, omvang, seisoenaliteit, vuurgrootte, frekwensie en tyd tussen die terugkeer van vuur na ‗n spesifieke plek (nl. FRI). Dit is waargeneem dat ongeveer 50% van die CSA jaarliks gebrand word, met 'n gemiddeld van 132 brande wat elke jaar voorkom. Daar was nie 'n algehele toename of afname in die omvang van die totale verbrande area oor die 10 jaar studietydperk nie. Maar 'n toenemende neiging in die vuurvoorkoms was wel opgemerk. Deur middel van 'n assessering van effektiewe vuurgrootte, is daar vasgestel dat meer kleiner brande in CSA voorkom, terwyl die omvang van die verbrand area redelik konstant gebly het. Dus was daar ‗n toename in die aantal vuurvoorvalle al was daar nie 'n ooreenstemmende toename in die totale verbrande oppervlakte was nie. Hierdie studie is ook gebruik om gebiede in die vuurregime te identifiseer wat 'n potensiële ekologiese risiko vir die miombobosveld in CSA inhou. Drie punte van kommer word geopenbaar: eerstens, 'n hoë voorkoms van laatseisoen brande is waargeneem in die noordelike grense van CSA. Tweedens, 64,2% van die totale verbrande oppervlakte in die CSA brand tussen 6 en 10 keer bine die 10-jaar periode. Laastens, 85% van die totale verbrande oppervlakte oor die tydperk 2001 tot 2010 het 'n FRI van minder as twee jaar. Die kombinasie van laatseisoen brande, hoë vuurfrekwensie en kort FRI in CSA is 'n aanduiding van moontlike veranderinge in die toestand van die miomboveld, wat negatiewe sosio-ekonomiese implikasies op die CSA en die omliggende gemeenskappe kan uitoefen. Hierdie studie het getoon dat die MCD45A1 'n nuttige bron van broodnodige vuur inligting vir Zimbabwe is. Daarom, moet die moontlikheid van die integrasie van die metodes wat gebruik word in hierdie studie in die huidige versameling van vuurdata behoorlike oorweging gegee word.

Trockengebiete, die etwa 40% der globalen Landoberfläche abdecken, Lebensgrundlage für 2 Millionen Menschen bilden, einzigarte Biodiversität enthalten und wichtige Ökosystemdienstleistungen bereitstellen, sind beträchtlichem Druck durch Landnutzungs- und Klimawandel ausgesetzt. Die räumlichen Muster dieser Landnutzungsänderungen sowie deren Ursachen sind jedoch nur in Ansätzen verstanden. Fernerkundung kann Veränderungen großräumig beobachten, aber Methoden fehlen, um diese Veränderungen, die graduell und abrupt (z.B. Feuer) auftreten können, abzuleiten. Die Ziele dieser Dissertation waren die Entwicklung fernerkundlicher Methoden, um verschiedene Landnutzungsänderungen in mediterranen Ökosystemen zu quantifizieren und um den Einfluss verschiedener Landnutzungsprozesse auf das Feuerregime besser zu verstehen. Die griechische Insel Kreta wurde als Untersuchungsgebiet gewählt. Zuerst wurden basierend auf Trendanalysen von Landsatzeitreihen verschiedene Vegetationsmaße verglichen. Demnach führen einfache Vegetationsmaße zu ähnlich guten Ergebnissen und ermöglichen eine effiziente Beobachtung großer Gebiete. Anschließend wurden Veränderungstrajektorien abgeleitet, um graduelle und abrupte Prozesse zu unterscheiden und um zu analysieren, wie Landnutzungssysteme und Feuerregime sich auf Kreta ausgewirkt haben. Mittels einer statistischen Modellierung wurde der relative Einfluss von Landnutzungsprozessen auf das Feuerregime quantifiziert. Die Ergebnisse zeigten ein komplexes Muster von graduellen Vegetationsveränderungen und Feuern und deuten darauf hin, dass Feuer hauptsächlich vom Beweidungssystem abhängt. Feuer traten häufig am Fuße der Berge auf, wohingegen in den Berggebieten Vegetationszunahme vorherrscht. Die statistische Modellierung bestätigte, dass Extensivierung und Klima die Hauptursachen des kretischen Feuerregimes sind. So ist zu vermuten, dass sich das ehemals durch Brennmaterial limitierte Feuerregime zu einem trockenheitsgetriebenen entwickeln wird.
Drylands cover about 41% of the earth’s surface, sustain the livelihoods of 2 billion people, harbor unique biodiversity and provide important ecosystem services, but land use and climate change exert considerable pressure on these ecosystems. However, pattern and drivers of land use change are weakly understood. Remote sensing can monitor these changes for large areas but methods to detect both gradual and abrupt events like fires are missing. The main objectives of this thesis were to develop remote sensing based methods to better quantify the impact of land use change on fire-prone Mediterranean ecosystems, and to apply these methods to better understand the influence of different land use processes on fire regimes. The island of Crete (Greece) served as study region where diverging land use transformation are extensive, fires are frequent and environmental gradients are large. First, the trade-off between different vegetation estimates when using Landsat-based trend analyses was quantified. The results suggested that simple vegetation estimates perform equally well and thus, allow for effective mapping of large areas. Second, a trajectory change detection approach was applied to separate gradual changes from abrupt events and to answer the question how land use systems and fire regimes have affected Crete’s rangelands. Statistical modeling was then used to quantify the relative importance of land use processes in driving the fire regime. The results show that vegetation changes resulted in complex pattern of gradual changes and fires likewise. The fire regime appeared to be mainly driven by changing grazing systems. Fires were frequent in foothills whereas mountains showed increasing vegetation as a result of land abandonment. The statistical modeling confirmed that land extensification and climate are the primary drivers of fire regimes on Crete. The results suggest that the former fuel-limited fire regime will likely shift towards a drought-driven fire regime.

The capability to monitor bushfires on a large scale from space has long been identified as an important contribution to climate and atmospheric research as well as a tool an aid in natural hazard response. Since the work by Dozier (1981), fire monitoring from space has relied on the principles he described. His method of identifying fires within a pixel significantly larger than the fire by utilizing the different responses of the 3 μm and 11 μm channels has been applied to a number of sensors. Over the last decade a lot of work has been invested to refine and validate fire detections based on this approach. So far, the application of the method proposed by Dozier (1981) reached its peak with the launch of the MODIS instrument on board the Terra satellite. In contrast to earlier sensors, MODIS was equipped with spectral channels specifically designed for the detection of fires with algorithms based on the work by Dozier (1981). These channels were designed to overcome problems experienced with other platforms, the biggest of which is the saturation of the 3 μm channel caused by big, hot fires. Since its launch, MODIS has proven itself to be a capable platform to provide worldwide fire detection at a moderate resolution of 1 km on a daily basis.
It is the intention of this work to open up new opportunities in remote sensing of fires from satellites by showing capabilities and limitations in the application of other spectral channels, in particular the 2.1 μm channel of MODIS, than the ones currently used. This channel is chosen for investigation as fires are expected to emit a significant amount of energy in this bandwidth and as it is available at a native resolution of 500 m on MODIS; double the resolution of the 3 μm and 11 μm channels. The modelling of blackbodies of typical bushfire temperatures shows that a fire detection method based on the 2.1 μm channel will not be able to replace the current methods. Blackbodies of temperatures around 600 to 700 K, that are common for smoldering fires, do not emit a great amount of energy at 2.1 μm. It would be hardly possible to detect those fires by utilizing the 2.1 μm channel. The established methods based on the 3 μm and 11 μm channels are expected to work better in these cases. Blackbodies of typically flaming fires (above 800 K) however show a very high emission around 2.1 μm that should make their detection using the 2.1 μm channel possible.
In order to develop a fire detection method based on the 2.1 μm channel, it is necessary to differentiate between the radiance caused by a fire of sub pixel size and the radiance of a pixel caused by the reflection of sunlight. This is attempted by using time series of past observations to model a reflectance value for a given pixel expected in absence of a fire. A fire detection algorithm exploiting the difference between the expected and observed reflectance is implemented and its detection results are compared to high resolution ASTER fire maps, the standard MODIS fire detection algorithm (MOD14) and burnt area maps. The detections of the method based on the 2.1 μm channel are found to correspond very well with the other three datasets. However, the comparison showed detections that do not align with MOD14 active fire detections but are generally aligned with burn areas. This phenomena has to be investigated in the future.