@phdthesis{Asam2014,
  author    = {Asam, Sarah},
  title     = {Potential of high resolution remote sensing data for leaf area index derivation using statistical and physical models},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-108399},
  school      = {Universit{\"a}t W{\"u}rzburg},
  year      = {2014},
  abstract  = {Information on the state of the terrestrial vegetation cover is important for several ecological, economical, and planning issues. In this regard, vegetation properties such as the type, vitality, or density can be described by means of continuous biophysical parameters. One of these parameters is the leaf area index (LAI), which is defined as half the total leaf area per unit ground surface area. As leaves constitute the interface between the biosphere and the atmosphere, the LAI is used to model exchange processes between plants and their environment. However, to account for the variability of ecosystems, spatially and temporally explicit information on LAI is needed both for monitoring and modeling applications. Remote sensing aims at providing such information. LAI is commonly derived from remote sensing data by empirical-statistical or physical models. In the first approach, an empirical relationship between LAI measured in situ and the corresponding canopy spectral signature is established. Although this method achieves accurate LAI estimates, these relationships are only valid for the place and time at which the field data were sampled, which hampers automated LAI derivation. The physical approach uses a radiation transfer model to simulate canopy reflectance as a function of the scene's geometry and of leaf and canopy parameters, from which LAI is derived through model inversion based on remote sensing data. However, this model inversion is not stable, as it is an under-determined and ill-posed problem. Until now, LAI research focused either on the use of coarse resolution remote sensing data for global applications, or on LAI modeling over a confined area, mostly in forest and crop ecosystems, using medium to high spatial resolution data. This is why to date no study is available in which high spatial resolution data are used for LAI mapping in a heterogeneous, natural landscape such as alpine grasslands, although a growing amount of high spatial and temporal resolution remote sensing data would allow for an improved environmental monitoring. Therefore, issues related to model parameterization and inversion regularization techniques improving its stability have not yet been investigated for this ecosystem. This research gap was taken up by this thesis, in which the potential of high spatial resolution remote sensing data for grassland LAI estimation based on statistical and radiation transfer modeling is analyzed, and the achieved accuracy and robustness of the two approaches is compared. The objectives were an ecosystem-adapted radiation transfer model set-up and an optimized LAI derivation in mountainous grassland areas. Multi-temporal LAI in situ measurements as well as time series of RapidEye data from 2011 and 2012 over the catchment of the River Ammer in the Bavarian alpine upland were used. In order to obtain accurate in situ data, a comparison of the LAI derivation algorithms implemented in the LAI-2000 PCA instrument with destructively measured LAI was performed first. For optimizing the empirical-statistical approach, it was then analyzed how the selection of vegetation indices and regression models impacts LAI modeling, and how well these models can be transferred to other dates. It was shown that LAI can be derived with a mean accuracy of 80 \% using contemporaneous field data, but that the accuracy decreases to on average 51 \% when using these models on remote sensing data from other dates. The combined use of several data sets to create a regression which is used for LAI derivation at different points in time increased the LAI estimation accuracy to on average 65 \%. Thus, reduced field measurement labor comes at the cost of LAI error rates being increased by 10 - 30 \% as long as at least two campaigns are conducted. Further, it was shown that the use of RapidEye's red edge channel improves the LAI derivation by on average 5.4 \%. With regard to physical LAI modeling, special interest lay in assessing the accuracy improvements that can be achieved through model set-up and inversion regularization techniques. First, a global sensitivity analysis was applied to the radiation transfer model in order to identify the most important model parameters and most sensitive spectral features. After model parameterization, several inversion regularizations, namely the use of a multiple sample solution, the additional use of vegetation indices, and the addition of noise, were analyzed. Further, an approach to include the local scene's geometry in the retrieval process was introduced to account for the mountainous topography. LAI modeling accuracies of in average 70 \% were achieved using the best combination of regularization techniques, which is in the upper range of accuracies that were achieved in the few existing other grassland studies based on in situ or air-borne measured hyperspectral data. Finally, further physically derived vegetation parameters and inversion uncertainty measures were evaluated in detail to identify challenging modeling conditions, which was mostly neglected in other studies. An increased modeling uncertainty for extremely high and low LAI values was observed. This indicates an insufficiently wide model parameterization and a canopy deviation from model assumptions on some fields. Further, the LAI modeling accuracies varied strongly between the different scenes. From this observation it can be deduced that the radiometric quality of the remote sensing data, which might be reduced by atmospheric effects or unexpected surface reflectances, exerts a high influence on the LAI modeling accuracy. The major findings of the comparison between the empirical-statistical and physical LAI modeling approaches are the higher accuracies achieved by the empirical-statistical approach as long as contemporaneous field data are available, and the computationally efficiency of the statistical approach. However, when no or temporally unfitting in situ measurements are available, the physical approach achieves comparable or even higher accuracies. Furthermore, radiation transfer modeling enables the derivation of other leaf and canopy variables useful for ecological monitoring and modeling applications, as well as of pixel-wise uncertainty measures indicating the robustness and reliability of the model inversion and LAI derivation procedure. The established look-up tables can be used for further LAI derivation in Central European grassland also in other years. The use of high spatial resolution remote sensing data for LAI derivation enables a reliable land cover classification and thus a reduced LAI mapping error due to misclassifications. Furthermore, the RapidEye pixels being smaller than individual fields allow for a radiation transfer model inversion over homogeneous canopies in most cases, as canopy gaps or field parcels can be clearly distinguished. However, in case of unexpected local surface conditions such as blooming, litter, or canopy gaps, high spatial resolution data show corresponding strong deviations in reflectance values and hence LAI estimation, which would be reduced using coarser resolution data through the balancing effect of the surrounding surface reflectances. An optimal pixel size with regard to modeling accuracy hence depends on the canopy and landscape structure. Furthermore, a reduced spatial resolution would enable a considerable acceleration of the LAI map derivation. This illustration of the potential of RapidEye data and of the challenges associated to LAI derivation in heterogeneous grassland areas contributes to the development of robust LAI estimation procedures based on new and upcoming, spatially and temporally high resolution remote sensing imagery such as Landsat 8 and Sentinel-2.},
  subject      = {Optische Fernerkundung},
  language  = {en}
}
@article{RichardAbdelRahmanSubramanianetal.2017,
  author    = {Richard, Kyalo and Abdel-Rahman, Elfatih M. and Subramanian, Sevgan and Nyasani, Johnson O. and Thiel, Michael and Jozani, Hosein and Borgemeister, Christian and Landmann, Tobias},
  title     = {Maize cropping systems mapping using RapidEye observations in agro-ecological landscapes in Kenya},
  series = {Sensors},
  volume    = {17},
  journal   = {Sensors},
  number    = {11},
  doi       = {10.3390/s17112537},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-173285},
  year      = {2017},
  abstract  = {Cropping systems information on explicit scales is an important but rarely available variable in many crops modeling routines and of utmost importance for understanding pests and disease propagation mechanisms in agro-ecological landscapes. In this study, high spatial and temporal resolution RapidEye bio-temporal data were utilized within a novel 2-step hierarchical random forest (RF) classification approach to map areas of mono- and mixed maize cropping systems. A small-scale maize farming site in Machakos County, Kenya was used as a study site. Within the study site, field data was collected during the satellite acquisition period on general land use/land cover (LULC) and the two cropping systems. Firstly, non-cropland areas were masked out from other land use/land cover using the LULC mapping result. Subsequently an optimized RF model was applied to the cropland layer to map the two cropping systems (2nd classification step). An overall accuracy of 93\% was attained for the LULC classification, while the class accuracies (PA: producer's accuracy and UA: user's accuracy) for the two cropping systems were consistently above 85\%. We concluded that explicit mapping of different cropping systems is feasible in complex and highly fragmented agro-ecological landscapes if high resolution and multi-temporal satellite data such as 5 m RapidEye data is employed. Further research is needed on the feasibility of using freely available 10-20 m Sentinel-2 data for wide-area assessment of cropping systems as an important variable in numerous crop productivity models.},
  language  = {en}
}
@article{ForkuorConradThieletal.2014,
  author    = {Forkuor, Gerald and Conrad, Christopher and Thiel, Michael and Ullmann, Tobias and Zoungrana, Evence},
  title     = {Integration of Optical and Synthetic Aperture Radar Imagery for Improving Crop Mapping in Northwestern Benin, West Africa},
  doi       = {10.3390/rs6076472},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-113070},
  year      = {2014},
  abstract  = {Crop mapping in West Africa is challenging, due to the unavailability of adequate satellite images (as a result of excessive cloud cover), small agricultural fields and a heterogeneous landscape. To address this challenge, we integrated high spatial resolution multi-temporal optical (RapidEye) and dual polarized (VV/VH) SAR (TerraSAR-X) data to map crops and crop groups in northwestern Benin using the random forest classification algorithm. The overall goal was to ascertain the contribution of the SAR data to crop mapping in the region. A per-pixel classification result was overlaid with vector field boundaries derived from image segmentation, and a crop type was determined for each field based on the modal class within the field. A per-field accuracy assessment was conducted by comparing the final classification result with reference data derived from a field campaign. Results indicate that the integration of RapidEye and TerraSAR-X data improved classification accuracy by 10\%-15\% over the use of RapidEye only. The VV polarization was found to better discriminate crop types than the VH polarization. The research has shown that if optical and SAR data are available for the whole cropping season, classification accuracies of up to 75\% are achievable.},
  language  = {en}
}
@article{KhareLatifiRossietal.2019,
  author    = {Khare, Siddhartha and Latifi, Hooman and Rossi, Sergio and Ghosh, Sanjay Kumar},
  title     = {Fractional cover mapping of invasive plant species by combining very high-resolution stereo and multi-sensor multispectral imageries},
  series = {Forests},
  volume    = {10},
  journal   = {Forests},
  number    = {7},
  issn      = {1999-4907},
  doi       = {10.3390/f10070540},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-197250},
  year      = {2019},
  abstract  = {Invasive plant species are major threats to biodiversity. They can be identified and monitored by means of high spatial resolution remote sensing imagery. This study aimed to test the potential of multiple very high-resolution (VHR) optical multispectral and stereo imageries (VHRSI) at spatial resolutions of 1.5 and 5 m to quantify the presence of the invasive lantana (Lantana camara L.) and predict its distribution at large spatial scale using medium-resolution fractional cover analysis. We created initial training data for fractional cover analysis by classifying smaller extent VHR data (SPOT-6 and RapidEye) along with three dimensional (3D) VHRSI derived digital surface model (DSM) datasets. We modelled the statistical relationship between fractional cover and spectral reflectance for a VHR subset of the study area located in the Himalayan region of India, and finally predicted the fractional cover of lantana based on the spectral reflectance of Landsat-8 imagery of a larger spatial extent. We classified SPOT-6 and RapidEye data and used the outputs as training data to create continuous field layers of Landsat-8 imagery. The area outside the overlapping region was predicted by fractional cover analysis due to the larger extent of Landsat-8 imagery compared with VHR datasets. Results showed clear discrimination of understory lantana from upperstory vegetation with 87.38\% (for SPOT-6), and 85.27\% (for RapidEye) overall accuracy due to the presence of additional VHRSI derived DSM information. Independent validation for lantana fractional cover estimated root-mean-square errors (RMSE) of 11.8\% (for RapidEye) and 7.22\% (for SPOT-6), and R\(^2\) values of 0.85 and 0.92 for RapidEye (5 m) and SPOT-6 (1.5 m), respectively. Results suggested an increase in predictive accuracy of lantana within forest areas along with increase in the spatial resolution for the same Landsat-8 imagery. The variance explained at 1.5 m spatial resolution to predict lantana was 64.37\%, whereas it decreased by up to 37.96\% in the case of 5 m spatial resolution data. This study revealed the high potential of combining small extent VHR and VHRSI- derived 3D optical data with larger extent, freely available satellite data for identification and mapping of invasive species in mountainous forests and remote regions.},
  language  = {en}
}
@article{LoewDuveiller2014,
  author    = {L{\"o}w, Fabian and Duveiller, Gr{\´e}gory},
  title     = {Defining the Spatial Resolution Requirements for Crop Identification Using Optical Remote Sensing},
  doi       = {10.3390/rs6099034},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-113607},
  year      = {2014},
  abstract  = {The past decades have seen an increasing demand for operational monitoring of crop conditions and food production at local to global scales. To properly use satellite Earth observation for such agricultural monitoring, high temporal revisit frequency over vast geographic areas is necessary. However, this often limits the spatial resolution that can be used. The challenge of discriminating pixels that correspond to a particular crop type, a prerequisite for crop specific agricultural monitoring, remains daunting when the signal encoded in pixels stems from several land uses (mixed pixels), e.g., over heterogeneous landscapes where individual fields are often smaller than individual pixels. The question of determining the optimal pixel sizes for an application such as crop identification is therefore naturally inclined towards finding the coarsest acceptable pixel sizes, so as to potentially benefit from what instruments with coarser pixels can offer. To answer this question, this study builds upon and extends a conceptual framework to quantitatively define pixel size requirements for crop identification via image classification. This tool can be modulated using different parameterizations to explore trade-offs between pixel size and pixel purity when addressing the question of crop identification. Results over contrasting landscapes in Central Asia demonstrate that the task of finding the optimum pixel size does not have a "one-size-fits-all" solution. The resulting values for pixel size and purity that are suitable for crop identification proved to be specific to a given landscape, and for each crop they differed across different landscapes. Over the same time series, different crops were not identifiable simultaneously in the season and these requirements further changed over the years, reflecting the different agro-ecological conditions the crops are growing in. Results indicate that sensors like MODIS (250 m) could be suitable for identifying major crop classes in the study sites, whilst sensors like Landsat (30 m) should be considered for object-based classification. The proposed framework is generic and can be applied to any agricultural landscape, thereby potentially serving to guide recommendations for designing dedicated EO missions that can satisfy the requirements in terms of pixel size to identify and discriminate crop types.},
  language  = {en}
}