1,720,977 research outputs found
Methodology for combining optical and microwave remote sensing in agricultural crop monitoring : the sugar beet crop as special case
No full textAccurate and up-to-date information on agricultural production is a vital component in running present market economies. In Europe considerable differences between c es in their agricultural production have led to a complex system of rules and subsidies which all rely on a certain level of accuracy regarding agricultural statistics (such as acreage and yield). At national level and regional level, such statistics have been collected so far by using conventional methods, which are mostly based on knowledge and experience from the past. Before using this information on a European level, there is a growing need for combining new information techniques and present knowledge to provide realistic estimates of crop yield and production on a lower scale level.Yield prediction is an important tool for industry, fanners and policy makers, facilitating logistic planning of transportation and production, storage and sale at national level and planning at farm level. In this thesis, the study is concentrated on the application of observation or remote sensing (RS) techniques to crop growth monitoring of agricultural crops in the Netherlands. A common crop in the Netherlands is sugar beet crop and this crop served as a perfect illustration for validation of the developed methodology in this study. The objective of this study is to understand how optical and microwave remote sensing may be used in a synergetic way in order to develop a methodology, that can be used to monitor crop growth and predict crop yield together with existing knowledge.More specifically, the study presented in this thesis aims to reveal (1) how useful information on biophysical properties of agricultural crops estimated with airborne remote sensing is for crop growth monitoring and yield prediction, (2) how successful this information can be utilized in the developed methodology for combining crop growth and remote sensing and (3) whether there are possibilities to apply this methodology for operational crop growth monitoring and yield prediction procedures by using airborne and to some lesser extent the current available spaceborne sensors.The thesis work is subdivided in three parts. Part I outlines the theory and background supporting the thesis methodology and the combination methodology itself In Part IL the test data are presented and, for the case study, the synergy of the combination of information is studied, especially for the multi-sensor airborne campaign MAC Europe 1991. Here the research questions I and 2 are being studied. The application of the methodology (research question 3) described in this thesis is evaluated in Part 111, accompanied by concluding remarks and recommendations.In Chapter 2, an inventory of the information estimated with RS is made in order to prepare the development of a methodology to monitor growth and production of agricultural crops with RS techniques. The major objective of this study is the investigation of the possibilities of a synergistic use of both optical and microwave RS data. Therefore, a review of the state of the art in modelling in the reflective optical and microwave region of the electromagnetic (EM) spectrum is performed. Furthermore, the most suitable models are selected and validated with the data from campaigns held in the Flevoland Province of the Netherlands as good as possible. It appeared that semi-empirical RS models, describing the observation of crops in a simplified physical way, could be calibrated and validated better than the complex radiative transfer models. The CLAIR model in the optical region has proven to be applicable over the different growing seasons, while the semi-empirical Cloud model in the microwave region revealed an unstable behaviour. Both models are calibrated with campaign data and were applied under strict conditions in this study in order to supply actual crop status information on respectively leaf area index (LAI) from the optical and biomass in the microwave model by inversion. From sensitivity analysis of the more complex radiative transfer (RS) models canopy structure appeared to be another important factor in the observation of crops as well in the optical as in the microwave region. Canopy structure information is not clearly incorporated in the semi-empirical RS models and therefore difficult to estimate. Changes in canopy structure have been recognised as specific features in time series of RS observation of the crop during the growing season, especially in microwave RS observations. The sugar beet crop revealed some characteristic features during the growing season, but not as clear as the vertically structured cereal crops, like winter wheat. Crop development related to changes in canopy structure in the case of winter wheat showed more potential for detection in RS time series as for sugar beet.In Chapter 3, a general methodology is proposed for combining the information (RS data, field data and models) of different sources in order to monitor crop growth and predict the yield. The underlying physiological processes of crop growth are studied for linkage of crop growth models with RS information. The SUCROS-type of crop growth model for the sugar beet crop from the School of de Wit from Wageningen appeared to be very suitable for this study, because of its detailed description of crop growth modelling and its status of being well initialised for crop growth conditions for sugar beet in the Flevoland Province. In this chapter, different methods were developed to calibrate the crop growth model with the actual information estimated by RS. The combination methods are:· Direct modelling method: Calibration of crop growth model with a forward RS model. By comparing the simulated RS signal with the observed RS data optimization of the most important variables of the crop growth model is performed.· Inverse modelling method: Calibration of crop growth model with an inverse RS model. In this method crop variables estimated with an inverse RS model are compared with crop variables of the crop growth model and used for optimization of the most important crop growth variables of the crop growth model.· Feature modelling method: Calibration of crop growth model by using characteristic information from RS time-series, which is mostly related to a change in structure of the canopy owing to changes in development stage of the crop.The direct model-based approach is only used for reference for the other methods and is developed in former research.The inverse model-based approach combines LAI and biomass estimated by optical and microwave RS model inversion with the crop growth model. The crop growth model was calibrated with this information and their estimation accuracies by using the reciproke of the standard deviation, which reflects the 'state of the art' in the RS modelling.The feature-based approach completes the methodology by detection of features in RS time series information on changes in canopy structure possibly related to crop development stages, which provide another source of information to calibrate the crop growth model as well. The overall methodology comprises the combination of the two approaches.Chapter 4 comprises a brief overview of data sets from campaigns at the Flevoland test site held in the past. In order to study the effect of synergism of optical and microwave RS data, conditioned data sets were required and aspects of quality and quantity of data in campaigns were discussed. The criteria for the synergy study were best met by the data set of the MAC Europe 1991 campaign compared to the other available data sets. For testing the combination methodologies of Chapter 3, the data from the airborne MAC Europe 1991 campaign were selected for the synergy study. This campaign was held at the time of the thesis study, so specific additional measurements could be collected like measurements on canopy structure. For RS model calibration and validation as well as for crop growth model initialization the Agriscatt 1987 and 1988 campaigns proved particularly suitable, because of the highly detailed information on field measurements. The ROVE data set from the late seventies provided measurements of high temporal frequency and were used for study of the impact of canopy structure on microwave backscatter and with that to illustrate specific radar features. The spaceborne ERS-1 time series from 1992 and 1993 were selected in order to discuss the potential of microwave satellite RS for operational crop growth monitoring in the last chapter and were not explicitly used in the study. A total processing line and a database for RS data interpretation was set up to prepare the study.In Chapter 5, the proposed combination methods of Chapter 3 were applied with contemporaneous (simultaneous) and non-contemporaneous recordings of airborne optical and microwave sensors of the MAC Europe 1991 campaign. The configuration of the airborne RS data was selected for this study on basis of the current optical (SPOT and Landsat) and microwave (ERS-1/2 and JERS-1) satellite configurations. The performance of the methods was measured by comparing the simulated yield as a result of the calibrated crop growth model and the actual measured yield figures at a specific harvest date.The inverse method is tested on the selected data set. The inverse RS model estimates LAI and biomass with a certain accuracy. The accuracy depends on the success of calibration of the (direct) RS model. It appeared that estimation of LAI from the optical model 'CLAIR' is at least twice as good as estimation of LAI from microwave model 'Cloud'. The combination of the crop growth model with optical data only gave good results. The added value of microwave data to this is present when no optical data are available (e.g. bad weather conditions). Using the information from both the airborne optical and microwave sensors weighted with the reciproke of the standard deviations the combination methods yielded success especially when the RS data was acquired in the beginning of the growing season. In this period the LAI can be well estimated, especially with optical RS models. Later in the growing season other information was found in RS time-series. With special attention to microwave time-series information on changes in canopy structure has been found and validated with field measurements of leaf angle distributions with respect to sugar beet. In the case of sugar beet these changes in structure are not clearly related to development of the crop. However, this is more pronounced in the case of cereals (e.g. winter wheat). This is information is also a source of calibration of the crop growth model. However, the accuracy of the feature found in the time-series is not high enough to calibrate the already well initialized crop growth model. When the observation frequency is high enough (weekly) then this information could be used for estimating the moment of sowing by using the meteorological information during the growing season.Chapter 6 discusses the practical application of the methodology. An important aspect is that the level of study is translated from field to regional level in order to find practical use for the method in conventional prediction strategies of the present (food) processing industry. The generalization step appeared to give new information. An example is that e.g. the minimum. in standard deviation in backscatter time-series from ERS-1 for all sugar beet fields in the Southern part of the Flevoland Province appeared to be related to a regional crop closure of the sugar beet for two different years (1992 and 1993). It is obvious that information estimated with RS models for each specific crop is valuable for crop growth monitoring when the moment and density of the RS measurement is well chosen during the growing season. This imposes high requirements to the present available satellite systems (ERS-1/2, JERS-1, Radarsat, SPOT, Landsat, etc.)Chapter 7 presents the main conclusions and recommendations for further research. More research is needed to calibrate and validate RS models for application in crop growth monitoring. The present generation crop growth models evolve towards reliable tools for yield forecasting and impose high requirements on quality of the RS information in order to be valuable for operational purposes. The methods used in this study gave good results in the case for airborne RS data on field level. Airborne optical and microwave RS information appeared to give synergetic results when combining with a crop growth model. The step towards a more operational monitoring method is expected to be difficult. The latter should be studied into more detail by using a more simple crop growth model and regional information from RS
The potential of UAV-based sun-induced chlorophyll fluorescence in understanding crop photosynthesis
No full textMonitoring photosynthesis is essential to understand the photosynthetic activity in crops for sustainable yield. Sun-induced chlorophyll fluorescence (SIF) is a byproduct of the photosynthesis process and is considered a direct measurement of the dynamic behaviour of photosynthesis. Studies on SIF at the ground, airborne, and satellite level have made important achievements in understanding the dynamic SIF-photosynthesis relationship. There is an urgent need to bridge the gap in SIF measurements between temporally continuous ground retrievals and high-altitude airborne or spatially coarse satellites and further explore the potential of SIF at field level in the context of precision agriculture. Unmanned Aerial Vehicle (UAV)-based measurements allow studying temporal SIF variation at the field scale and can potentially close the mentioned spatial gap. To minimize the risk of data artifacts and correctly understand SIF values, the ability of UAV-based SIF observations to provide reliable information within agricultural fields needs a robust evaluation. Due to the physiological connection between photosynthetic changes and fluorescence emission, UAV-based SIF has the potential to support the timely detection of water stress at the field scale. However, the direct effect of water stress on the SIF response in crops at the field level still needs further research to clearly understand the involved mechanisms. As SIF retrieval is jointly affected by multiple factors, elucidation of the confounding factors of SIF is also highly needed for a reliable interpretation of the physiological variations caused by water stress. The objectives of this research are i) to evaluate the ability of a novel UAV-based system to acquire reliable SIF under field conditions, and ii) to interpret UAV-based SIF response to water stress for a better understanding of photosynthetic activities.Chapter 2 presented the system set-up and the processing chain of a novel UAV-based system, FluorSpec, for SIF acquisition at the field level and evaluated the potential of this system and the diurnal SIF patterns for different arable crops. To test the reliability of FluorSpec diurnal SIF measurements, canopy diurnal SIF was firstly monitored over two crops using the ground-based FluorSpec. SIF from the two crops had a pronounced and expected diurnal SIF development similar to the photosynthetically active radiation (PAR). UAV-based SIF exhibited a clear diurnal pattern similar to the proximal canopy SIF measurements. Clear spatial variation within different crop fields was observed within FluorSpec footprints. The obtained results showed the ability of the FluorSpec system to reliably measure plant fluorescence at ground and field level.Chapter 3 further evaluated the ability of the UAV-based FluorSpec to measure reliable SIF by comparing FluorSpec with a high-performance airborne imaging spectrometer, HyPlant. Airborne HyPlant and the UAV-based FluorSpec acquired diurnal SIF measurements almost simultaneously during a clear sky day. The FluorSpec and HyPlant SIF measurements, their diurnal developments, and spatial distributions for different crop types were compared. A high linear correlation was found between UAV-based FluorSpec SIF and HyPlant SIF. Both UAV and airborne-based SIF show similar and pronounced diurnal patterns for most crops. Consistent spatial patterns of SIF over different crops for both systems were also clearly observed. These findings confirm that the UAV-based FluorSpec system is able to measure meaningful SIF values at the field scale and facilitate bridging the gap in SIF monitoring between ground and ecosystem scales.Chapter 4 assessed the value of UAV-based SIF for water stress detection in a crop field. SIF measurements by the UAV-based FluorSpec were acquired over irrigated and water-stressed sugar beet plots in June 2019 under water stress and in July 2019 under combined water stress and heat stress. SIF indices were applied to detect water stress under the two different conditions. Additionally, UAV-based hyperspectral and thermal data were acquired to assist the interpretation of SIF indices. The SIF indices showed a significant response to the recovery of sugar beet after irrigation when sugar beet plants were exposed to water stress. However, only some selected SIF indices weakly tracked the changes induced by the irrigation when the crop was under severe combined stress. SIF at 687 nm and 760 nm and their indices reacted differently to the irrigation. This study confirms the capacity of SIF acquired by a UAV system to detect water stress at the field level, but its value might be limited for severe water stress detection. Further investigations are necessary to give a comprehensive understanding of the potential of UAV-based SIF to detect crop stress at different levels.Chapter 5 presented a modelling approach to disaggregate the induced physiological and non-physiological effects by water stress on SIF variations for a better understanding of the photosynthetic dynamics under stressed conditions. UAV-based SIF measurements were acquired over irrigated and non-irrigated sugar beet plots. Fluorescence emission yield (ΦF) and biochemical and structural factors jointly controlled TOC SIF. SIF variation both at 687 nm and 760 nm caused by water stress was strongly affected by the physiological factor ΦF and positively correlated well with SIF variations only caused by ΦF. At 687 nm, non-physiological changes had a weak effect on SIF variations, while at 760 nm non-physiological changes negatively and non-significantly mediated SIF response to water stress. The combination of RTMs, TOC reflectance, and TOC SIF measurements enables the physiological information quantification from SIF observations and supports the scalable quantitative use of SIF from leaf to ecosystem level.From the results in this thesis, it can be concluded that 1) the UAV-based FluorSpec observations can explore crop SIF and photosynthetic activities at field level and upscale the SIF measuring from the ground level to the field level by providing accurate, high resolution, and flexible spectral measurements; 2) UAV-based SIF is capable of detecting water stress at an early stage in a crop while its potential in severe stress detection needs further research; 3) the physiological and non-physiological changes both contribute to SIF variation caused by water stress, and the physiological changes had a strong effect on SIF variations in the presented case
UAV-based multi-angular measurements for improved crop parameter retrieval
Get PDFOptical remote sensing enables the estimation of crop parameters based on reflected light through empirical-statistical methods or inversion of radiative transfer models. Natural surfaces, however, reflect light anisotropically, which means that the intensity of reflected light depends on the viewing and illumination geometry. Therefore, reflectance anisotropy can be considered as an unwanted effect since it may lead to inaccuracies in parameter estimations. However, it can also be considered as information source due to its unique response to the optical and structural properties of the observed surface. In the past, reflectance anisotropy was studied by multi-angular reflectance measurements from space-borne or ground-based sensors. In this research, the opportunities of Unmanned Aerial Vehicles (UAVs) to collect multi-angular measurements were explored. The main results of this research show that multi-angular measurements can be done with UAVs and that the reflectance anisotropy signal can be used to improve the retrieval of crop parameters
Combining conventional ground-based and remotely sensed forest measurements
No full textThe world’s natural ecosystems are under pressure due to land conversion and climate change. Forests are a major part of these natural ecosystems and cover up to 30% of the earth’s surface. Trees are crucial for timber, store carbon, and provide other ecosystem functions. Assessing and predicting forest ecosystem responses based on global environmental changes is an important task for scientists. For many decades monitoring of forest ecosystems has been implemented using various well established (conventional) methods, but in more recent decades remote sensing techniques have made steep developments and provide opportunities in this respect. Although conventional monitoring has proven its value in many cases, monitoring forest ecosystems for decision making often requires large scale monitoring. In this sense, remote sensing (RS) offers a solution and has been successfully used for monitoring forest disturbance at regional and global scales. However, remote sensing also has made advances at plot scale, for example using near-sensing terrestrial laser scanning (TLS). Despite the potential for close collaborations between the remote sensing and forest ecology communities, there is still a disconnect (e.g. spatial / temporal resolution of data) between the two fields of expertise, which means that combining data from the two fields is difficult. To better understand for example tree physiological processes using remote sensing, further synchronisation of the two fields is vital for improving the potential for satellite data. In this thesis, I explored how to link conventional ground-based methods with remote sensing techniques, using both satellite and novel near-sensing TLS in order to investigate aspects of forest change. To do this, I looked at four representative cases with different forest types, which were selected to address different current environmental challenges with indirect (reduced tree vitality) and direct (change in forest structure) impacts. In chapter 2, we attempted to upscale ground-based conventional forest canopy measurements at plot level to remote sensing derived indices of the canopy in the Pampa del Tamarugal aquifer, in the hyper-arid Atacama desert of Chile. We assessed the foliage loss (dry branches) of the Prosopis tamarugo Phil. (a native tree) by ground-based visual assessment and digital pictures over three groundwater depletion conditions. These pictures were segmented and classified into green and brown canopy to derive the GCF (green canopy fraction). The GCF was then related to NDVIw (NDVI in winter time) from the WorldView2 satellite data, and NDVIw was used to estimate and thus upscale the GCF to all P. tamarugo trees in the aquifer. NDVIw derived from the Landsat archive allowed us to not only assess the current status of the P. tamarugo trees, but also changes over time. In this study we could successfully link ground-based conventional forest canopy measurements to remote sensing derived indices of the canopy. This allowed us, in combination with the groundwater level grids, to assess the tree vitality of the whole aquifer and determine a critical groundwater depth of 20 m for the P. tamarugos survival. Chapter 3 and 4 link ring width (RW) data at plot level to remote sensing derived plot level indices of the canopy. In both chapters the aim was to better understand the effect of environmental factors (e.g. water shortage) on the growth of trees both from the stem (wood) and canopy perspective. In Chapter 3 we used the GCF (current situation) and NDVI-based indices (historical situation) from satellite data derived from chapter 2, and assessed the correlation between NDVI-based satellite indices with ground-based tree-ring increments in two contrasting sites (low and high groundwater depletion). Time-series analysis (over a period of 26 years) and NDVI-derived parameters showed significant negative trends in the high-depletion site, indicating drought stress. Ring width of P. tamarugo trees was 48% lower in the high-depletion site. At the tree level, the GCF in the highly depleted site also indicated drought stress since a larger percentage of trees fell within lower GCF classes. In this case monitoring water shortage over time was straightforward since the stand was monospecific, and water shortage happened gradually. This was not the case in chapter 4 where we addressed the effect of climate on tree growth by combining tree-ring data of 25 locations in Slovenia with remote sensing derived EVI indices (enhanced vegetation index) from MODIS (Moderate Resolution Imaging Spectroradiometer) satellite data. We attempted to upscale the results at plot scale to national level for the tree species Fagus sylvatica L. (Beech) in the temperate forests of Slovenia. We were not able to find any relations of both RW and EVI based anomalies with climate parameters, nor was there a relation between RW and EVI based anomalies. Reasons might be: (i) time-series length (i.e. overlap between the data types), (ii) complexity of the environmental stress, such as the interplay of climatic conditions with other factors such as topography (also depending on the timing and duration of a climate event within the year), (iii) a satellite pixel might consist of other tree species less sensitive to drought, and (iv) empirical linkage between parameters – uncertainty about the direct relationship between stem and canopy derived parameters. We did find indications that both RW and EVI based anomalies were negatively affected by the extreme climate events in Slovenia, in particular the effect of the ice storm of 2014. Combining dendrochronology and remote sensing allowed us to understand the effects of drought stress on two different carbon pools (crown and stem, respectively), providing more insights on the physiological response of the species to drought. In Chapter 5 we investigated forest structure in tropical forests in Ethiopia. Here we combined conventional forest inventory measurements such as biomass, tree density, and tree species, with near-sensing TLS measurements such as PAVD (plant area vegetation density) and canopy openness. Differences between four forest types (intact forest, coffee forest, silvopasture, and plantation) for both conventional and TLS measurements were assessed. Results showed that the 3D vegetation structure (i.e. PAVD) and canopy parameters could be used to differentiate between forest types. TLS as tool for monitoring forest structure showed potential as it can capture the 3D position of the vegetation volume and open spaces at all heights. To quantify changes in different forest types, consistent monitoring of 3D structure is needed and here TLS is an add-on or an alternative to conventional forest structure monitoring. This thesis contributes to the exploration of the advantages of combining conventional ground-based data and remote sensing derived data. Combining both data types can mutually enhance the potential capabilities of each other. Remote sensing can upscale plot data to large spatial scales, while near sensing tools such as TLS can provide detailed forest structural data. Conventional ground-based data provide insight into the ecology of stands, e.g. RW or ecophysiological processes, and can help to understand remote sensing derived canopy indices. Advances in remote sensing are moving towards higher spatial, temporal, and spectral detail, but without the in-situ ecological data these advances do not reach their full potential. I, therefore, strongly advocate closer collaborations between the two research fields in the set-up of monitoring campaigns (e.g. different spatial scales). In this thesis I explored the value of combing the two fields in an empirical way. However, not all issues have been solved and more research is needed. Future research could focus on an integrated and tree-centred approach that can help to understand climate-growth interaction and the connections between stem and canopy derived indices. Future challenges also lay in improving the data operability and data processing. For remote sensing to become a conventional method, it has to become more ecologically (ecosystem) driven in an operational and cost-effective way.</p
Tropical deforestation monitoring using Landsat time series and breakpoint detection
No full textLandsat time series Breaks For Additive Season and Trend (BFAST) breakpoint detection was identified as an accurate and generic deforestation monitoring device for the tropics. Suitability for time series pre-processing while varying sites and response signals was researched across different study sites in south America, Asia and Africa. Extensive reference data composed of ground truth data and very high resolution data was used to describe error sources and calculate performance accuracies. Machine learning was used to fuse results to create best maps
Change detection with remote sensing: relating NOAA-AVHRR to environmental impact of agriculture in Europe
No full textAgricultural production in the European Union sharply rose during the second half of the 20 <sup>th</SUP>century. As a side-effect environmental impact increased as well, and resulted in widespread environmental problems, which policymakers now seek to reduce. Therefore, up-to-date, standardised information on environmental impact of agriculture is required covering the entire area of the Union. NOAA-AVHRR images seem well suited to provide part of this information, because 1) one image covers a large area, 2) so significant time series are available, and 3) they contain two relevant spectral bands for vegetation and crop studies. The objective of this study is to develop a change detection method to locate changes in environmental impact using NOAA-AVHRR images.The required spatial observation units were defined such that they match both agriculture and NOAA-AVHRR. For this purpose a method was developed to determine the correspondence in geometry between two polygon sets. It was shown that polygons formed by bio-physical variables match patterns in AVHRR images better than those formed by socio-economic variables. The selected units were obtained from the soil map.Once the spatial units were defined, measures could be sought that characterise environmental impact in terms of land cover, so they might be observable in the AVHRR images. A suitable measure was found in change in agricultural area, which will result in changed environmental impact if other factors remain unchanged. For changes in agricultural intensity, which will lead to changed environmental impact as well, no suitable measures exist.Finally, three different change detection methods were proposed to detect changes in agricultural area using NOAA-AVHRR images. The methods aim at enhancing the information regarding agricultural change while minimising classification inaccuracy, spatial misregistration and radiometric effects. None of these methods proved successful in locating regions with changes in agricultural areas. The conclusion is that NOAA-AVHRR images seem not suited to detect changes in European agriculture.Besides these aspects related to a change detection method, methods to solve cloud contamination of NOAA-AVHRR images were studied. Clouds often reduce the useful area in AVHRR images. Seven procedures, including conventional and geostatistical methods, to replace small clouds by estimated land radiation values were compared. The estimates from the geostatistical methods led to the best estimates of reflection values from the landscape underlying the clouds.Next, the suitability of near-future remote-sensing systems was assessed for detecting changes in environmental impact of agriculture. MERIS is a sensor mounted on ENVISAT, a European satellite that will be launched in November 2001. Its announced specifications make it seem a promising information source for land applications at the continental scale. To estimate the value of its 300m pixel a new method is proposed, which is referred to as the Stained Glass Procedure. This method relates pixel size to discernible detail, and predicts the level of detail detectable in another (here non-existing yet) image. According to the Stained-Glass Procedure, MERIS images will show twice as much detail as NOAA-AVHRR images, which is a significant improvement. Unfortunately, it will take quite some years before time series useful for change detection have been collected
Mapping and monitoring forest remnants : a multiscale analysis of spatio-temporal data
Get PDFKEYWORDS : Landsat, time series, machine learning, semideciduous Atlantic forest, Brazil, wavelet transforms, classification, change detectionForests play a major role in important global matters such as carbon cycle, climate change, and biodiversity. Besides, forests also influence soil and water dynamics with major consequences for ecological relations and decision-making. One basic requirement to quantify and model these processes is the availability of accurate maps of forest cover. Data acquisition and analysis at appropriate scales is the keystone to achieve the mapping accuracy needed for development and reliable use of ecological models.The current and upcoming production of high-resolution data sets plus the ever-increasing time series that have been collected since the seventieth must be effectively explored. Missing values and distortions further complicate the analysis of this data set. Thus, integration and proper analysis is of utmost importance for environmental research. New conceptual models in environmental sciences, like the perception of multiple scales, require the development of effective implementation techniques.This thesis presents new methodologies to map and monitor forests on large, highly fragmented areas with complex land use patterns. The use of temporal information is extensively explored to distinguish natural forests from other land cover types that are spectrally similar. In chapter 4, novel schemes based on multiscale wavelet analysis are introduced, which enabled an effective preprocessing of long time series of Landsat data and improved its applicability on environmental assessment.In chapter 5, the produced time series as well as other information on spectral and spatial characteristics were used to classify forested areas in an experiment relating a number of combinations of attribute features. Feature sets were defined based on expert knowledge and on data mining techniques to be input to traditional and machine learning algorithms for pattern recognition, viz . maximum likelihood, univariate and multivariate decision trees, and neural networks. The results showed that maximum likelihood classification using temporal texture descriptors as extracted with wavelet transforms was most accurate to classify the semideciduous Atlantic forest in the study area.In chapter 6, a multiscale approach to digital change detection was developed to deal with multisensor and noisy remotely sensed images. Changes were extracted according to size classes minimising the effects of geometric and radiometric misregistration.Finally, in chapter 7, an automated procedure for GIS updating based on feature extraction, segmentation and classification was developed to monitor the remnants of semideciduos Atlantic forest. The procedure showed significant improvements over post classification comparison and direct multidate classification based on artificial neural networks.</p
Advancing forest structure product validation with ground, space and unmanned aerial vehicle sensors
No full textForests play a crucial role in the functioning of the Earth’s climate system, through their role in the carbon, energy and water cycles. The accurate description and quantification of their physical structure is essential to understand these roles, predict their behaviour under future climate change and adapt management practices accordingly. Remote sensing in particular from space-borne platforms is attractive for large area assessment of forest structure due to its cost-effectiveness, repeatability and objectiveness. However, the remote sensing signal is by nature ambiguous and needs to be interpreted with solid understanding of the underlying radiative mechanisms and uncertainties need to be rigorously quantified with independent ground data. The remote sensing community has produced a range of biophysical products describing vegetation and forest structure as well as best practice guidelines for their validation. However, the full implementation of anticipated products, including systematic repetition of validation across multiple sites (Committee on Earth Observing Satellites (CEOS) Land Product Validation (LPV) stage 4), is still to be concluded. A major challenge in this context is the provision of long-term validation data sets, which need to be cost-effective, repeatable and fast to acquire in the field. This thesis aims to investigate new ways of validation that meet the temporal and/or spatial scales of global forest structure products from space-borne missions with hectometric resolution. The particular focus is on Leaf Area Index (LAI) and Above-Ground Biomass (AGB) as metrics of physical forest structure. For the purpose of this thesis, the Speulderbos Reference site in the Veluwe forest area (The Netherlands) was established, where ground and Unmanned Aerial Vehicle (UAV)-borne sensors were tested. In Chapter 2, the automatic, passive optical sensor PAI Autonomous System from Transmittance Sensors at 57° (PASTiS-57) was tested for its suitability to monitor forest phenology and Plant Area Index (PAI), the total one-sided area of plant material per unit ground. For this, Radiative Transfer Model (RTM) experiments with turbid media and heterogeneous scenes were employed. PASTiS-57 generally meets the CEOS LPV requirement of 20% accuracy over a wide range of biochemical and illumination conditions for turbid medium canopies. However, canopy non-randomness in discrete tree models led to strong biases. In a field experiment, PASTiS-57 compared well in terms of phenological timing with Terrestrial Laser Scanning (TLS)-based PAI time series. PASTiS-57 represents a cost-effective way to continuously monitor PAI in forests. In Chapter 3, decametric resolution Sentinel-2 and Landsat 7/8 observations were analysed with hybrid LAI retrieval algorithms, which combine RTMs with Machine Learning Regression Algorithms (MLRAs). Several combinations of RTMs, MLRAs, and modifications to the processing chain were tested in order to assess their performance to predict a ground-based LAI time series, created from combined TLS and litter trap data. Most important for the success of the processing chain was the addition of a certain level of Gaussian noise to the RTM-produced database prior to MLRA training. With this processing chain, decametric resolution optical missions can produce reference LAI products for inter-comparison with hectometric products. Alternatively, the higher resolution can help to scale up small plot-based ground validation data. In Chapter 4, a novel Unmanned Aerial Vehicle Laser Scanning (UAV-LS), the RiCOPTER with VUX-1UAV laser scanner, was used to estimate canopy height and Diameter at Breast Height (DBH). TLS was used to derive reference datasets for both variables. Canopy height was comparable between both sensors with a slight underestimation for TLS, which was expected due to occlusion of the upper canopy when seen from below and hence lower TLS canopy heights. DBH was derived for the first time from UAV-LS data and compared well with TLS derived DBH. However, a part of the UAV-LS samples could not produce a meaningful estimate of DBH based on the extracted point cloud segment due to low point density. Repeated overpasses could counteract this to some degree. In this context, UAV-LS can support fast, plot-scale assessment of these two variables. In Chapter 5, the capabilities of UAV-LS are further explored in terms of explicit 3D modelling in order to estimate tree volume, which is the first step to retrieve tree AGB. For this purpose, 3D cylinder models were fitted to the segmented single trees with the TreeQSM routine. The resulting models were compared with TLS-based models and analysed separately for five different stands with varying architectures, including deciduous and coniferous species. UAV-LS was generally very successful in modelling large, deciduous trees, while coniferous trees with low branches and foliage as well as small trees proved more difficult. If successful, UAV-LS can provide the means to produce plot-scale assessment of woody volume and subsequently AGB at a fraction of time needed for TLS surveys. This thesis investigates new ways of forest structure product validation with techniques and sensors that meet the temporal and/or spatial resolution of hectometric space-borne missions.</p
Controls of forest age and ecological memory effects on biosphere-atmosphere CO2 exchange
Get PDFUnderstanding the dynamics of terrestrial ecosystems in a changing environment is critical because of their fundamental role in the global carbon (C) cycle. Climate extremes, ecological disturbances, and anthropogenic activities are currently altering the functioning of terrestrial ecosystems. As a result, there is a need to improve the monitoring of the terrestrial ecosystem’s and the role of extreme events (i.e. natural and human-induced disturbances) in the biogeochemical cycles for better quantifying regional and global C dynamics. In recent years, there has been an intensive global effort to measure and model carbon dioxide (CO2) exchanges between the terrestrial biosphere and the atmosphere. The integration of multiple modeling methods, remote sensing data, climate data, and a global network of eddy-covariance (EC) flux towers has provided unprecedented insights in understanding the mechanisms controlling CO2 fluxes from ecosystem to regional scales. However, current bottom-up approaches do not explicitly account for the effects of site history (e.g. forest age) and the ecological memory effects of both vegetation and climate dynamics on CO2 fluxes. Although most scientists agree on the importance of forest age and ecological memory effects in controlling the CO2 flux variability, there is still a debate about the quantitative role of forest age and ecological memory effects in estimating CO2 fluxes. In my thesis, I explored ways both to integrate forest age as well as ecological memory effects and to quantify their relevance when estimating the spatiotemporal variability of the CO2 fluxes. This was approached from two directions. First, a statistical method based on a combination of climate, ancillary, and EC data was developed to quantify the role of forest age towards the terrestrial net CO2 fluxes. Second, the application of a deep learning (DL) method was explored for understanding the contribution of vegetation and climate’s ecological memory effects on CO2 fluxes. In Chapter 2, I carried-out an observational synthesis to determine to what extent environmental conditions and site history (i.e. forest age) influence the spatiotemporal variability of forest annual net ecosystem production (NEP) across a set of forest EC flux sites globally. The proposed empirical model yielded a substantial capacity for reproducing the spatiotemporal (Nash-Sutcliffe model efficiency (NSE) of 0.62) and across-site variability (NSE of 0.71) of annual forest NEP. By investigating the model structure, I found that forest age was the main driver of NEP spatiotemporal variability in both space and time (decrease in NSE of 0.42 and 0.50 for spatiotemporal and across-site variability, respectively). These results confirmed the importance of forest age in quantifying spatiotemporal variation in NEP using data-driven approaches and paved the way towards further developments in upscaling EC data. Based on the findings of Chapter 2, I provided new global estimates of forest C balance by accounting for both forest age and climate spatial variations (chapter 3). Gridded estimates of forest NEP inferred from a new forest age map and environmental gridded global products (i.e. air temperature, gross primary production, and nitrogen deposition) at 0.5 spatial resolution for the period 2000-2013 were produced. This approach estimated the global forest NEP as a sink of around +50.2 PgC yr-1 and the net biome production (NBP) of forests of around +30.3 PgC yr-1. Forest NBP estimates matched results of independent forest inventories globally, while discrepancies were found at biome level (i.e. temperate, boreal, and tropical regions). Overall, this first attempt to include forest age for estimating the forest C balance globally provided new insights on both the location and the magnitude of the global land C sink. Furthermore, I investigated the relevance of capturing the vegetation and climate temporal properties, the so-called ecological memory effects, for predicting net ecosystem exchange (NEE) at 185 forest and woodland FLUXNET sites (Chapter 4). To answer this question, I used a data-driven DL model that translates the response of net CO2 fluxes to past climate and vegetation fluctuations: Long-Short-Term Memory (LSTM) model. The findings of the experiments were two folds: (1) an LSTM approach with embedded climate and vegetation ecological memory effects outperforms a non-dynamic statistical model (i.e. Random Forest) and (2) the vegetation mean seasonal cycle embeds most of the information content to realistically explain the spatial and seasonal variations in NEE. To further explore the contribution of vegetation and climate’s temporal dynamic properties to CO2 fluxes globally (Chapter 5), I expanded the approach developed in Chapter 4. A bottom-up approach and a series of experiments provided evidence with respect to the geographical distribution and magnitude of vegetation and climate’s ecological memory effects on NEE for the 2001-2018 period. The spatial patterns of ecological memory effects as well as the controls of the ecosystem properties and climatic conditions on the observed ecological memory effects’ spatial patterns were explored. The results depicted widespread and substantial ecological memory effects across the globe, confirming the importance of explicitly capturing the vegetation and climate temporal properties to accurately reproduce CO2 flux spatiotemporal patterns. Finally, I explored to what degree vegetation and climate’s ecological effects control the biosphere-atmosphere CO2 responses to a specific climate extreme event (i.e. 2018 European heatwave). Chapter 6 summarized the main findings of the thesis and provided additional reflections as well as outlooks for future research. Overall, this thesis strengthened the role of ecosystem history in understanding the biosphere-atmosphere CO2 exchange. Methodologically, my works have demonstrated the potential of new modeling approaches in the Earth system science, such as DL. Yet, the accommodation of new data streams in the presented modeling schemes and the development of new model frameworks are of relevance. Thereby, the potential of integrating new datasets (e.g. biomass time-series, soil moisture) for investigating the control of C stocks and soil moisture stress on CO2 fluxes for providing more reliable global CO2 fluxes products were briefly presented in Chapter 6. In addition, the potential application of a data-driven method (i.e. transfer learning) for overcoming the problem of extrapolation when modeling CO2 fluxes from site to globe was introduced. This chapter also pointed at the design of modeling schemes that are not only data-adaptive but also embed physical ecosystem properties, the so-called hybrid models. Finally, a brief reflection on the feasibility of implementing operational systems for CO2 flux monitoring was provided
- …
