Daniel Doktor

Reduction of Radiometric Miscalibration—Applications to Pushbroom Sensors

The analysis of hyperspectral images is an important task in Remote Sensing. Foregoing radiometric calibration results in the assignment of incident electromagnetic radiation to digital numbers and reduces the striping caused by slightly different responses of the pixel detectors. However, due to uncertainties in the calibration some striping remains. This publication presents a new reduction framework that efficiently reduces linear and nonlinear miscalibrations by an image-driven, radiometric recalibration and rescaling. The proposed framework—Reduction Of Miscalibration Effects (ROME)—considering spectral and spatial probability distributions, is constrained by specific minimisation and maximisation principles and incorporates image processing techniques such as Minkowski metrics and convolution. To objectively evaluate the performance of the new approach, the technique was applied to a variety of commonly used image examples and to one simulated and miscalibrated EnMAP (Environmental Mapping and Analysis Program) scene. Other examples consist of miscalibrated AISA/Eagle VNIR (Visible and Near Infrared) and Hawk SWIR (Short Wave Infrared) scenes of rural areas of the region Fichtwald in Germany and Hyperion scenes of the Jalal-Abad district in Southern Kyrgyzstan. Recovery rates of approximately 97% for linear and approximately 94% for nonlinear miscalibrated data were achieved, clearly demonstrating the benefits of the new approach and its potential for broad applicability to miscalibrated pushbroom sensor data.

Extraction of Plant Physiological Status from Hyperspectral Signatures Using Machine Learning Methods

The machine learning method, random forest (RF), is applied in order to derive biophysical and structural vegetation parameters from hyperspectral signatures. Hyperspectral data are, among other things, characterized by their high dimensionality and autocorrelation. Common multivariate regression approaches, which usually include only a limited number of spectral indices as predictors, do not make full use of the available information. In contrast, machine learning methods, such as RF, are supposed to be better suited to extract information on vegetation status. First, vegetation parameters are extracted from hyperspectral signatures simulated with the radiative transfer model, PROSAIL. Second, the transferability of these results with respect to laboratory and field measurements is investigated. In situ observations of plant physiological parameters and corresponding spectra are gathered in the laboratory for summer barley (Hordeum vulgare). Field in situ measurements focus on winter crops over several growing seasons. Chlorophyll content, Leaf Area Index and phenological growth stages are derived from simulated and measured spectra. RF performs very robustly and with a very high accuracy on PROSAIL

The use of airborne hyperspectral data for tree species classification in a species-rich Central European forest area

Abstract The success of remote sensing approaches to assess tree species diversity in a heterogeneously mixed forest stand depends on the availability of both appropriate data and suitable classification algorithms. To separate the high number of in total ten broadleaf tree species in a small structured floodplain forest, the Leipzig Riverside Forest, we introduce a majority based classification approach for Discriminant Analysis based on Partial Least Squares (PLS-DA), which was tested against Random Forest (RF) and Support Vector Machines (SVM). The classifier performance was tested on different sets of airborne hyperspectral image data (AISA DUAL) that were acquired on single dates in August and September and also stacked to a composite product. Shadowed gaps and shadowed crown parts were eliminated via spectral mixture analysis (SMA) prior to the pixel-based classification. Training and validation sets were defined spectrally with the conditioned Latin hypercube method as a stratified random sampling procedure. In the validation, PLS-DA consistently outperformed the RF and SVM approaches on all datasets. The additional use of spectral variable selection (CARS, “competitive adaptive reweighted sampling”) combined with PLS-DA further improved classification accuracies. Up to 78.4% overall accuracy was achieved for the stacked dataset. The image recorded in August provided slightly higher accuracies than the September image, regardless of the applied classifier.

Combining Messy Phenological Time Series

We describe a method for combining phenological time series and outlier detection based on linear models as presented in Schaber and Badeck (Tree Physiol, 22, 973–982, 2002). We extend the outlier detection method based on Gaussian Mixture Models as proposed by Doktor et al. (Geostatistics for environmental applications, Springer, Berlin, 2005) in order to take into account year-location interactions. We quantify the effect of the extension of the outlier detection algorithm using Gaussian Mixture Models. The proposed methods are adequate for the analysis of messy time series with heterogeneous distribution in time and space as well as frequent gaps in the time series. We illustrate the use of combined time series for the generation of geographical maps of phenological phases using station effects. The algorithms discussed in the current paper are publicly available in the updated R – package “pheno”.

Optimising Phenological Metrics Extraction for Different Crop Types in Germany Using the Moderate Resolution Imaging Spectrometer (MODIS)

Phenological metrics extracted from satellite data (phenometrics) have been increasingly used to access timely, spatially explicit information on crop phenology, but have rarely been calibrated and validated with field observations. In this study, we developed a calibration procedure to make phenometrics more comparable to ground-based phenological stages by optimising the settings of Best Index Slope Extraction (BISE) and smoothing algorithms together with thresholds. We used a six-year daily Moderate Resolution Imaging Spectrometer (MODIS) Normalized Difference Vegetation Index (NDVI) time series and 211 ground-observation records from four major crop species (winter wheat/barley, oilseed rape, and sugar beet) in central Germany. Results showed the superiority of the Savitzky–Golay algorithm in combination with BISE. The satellite-derived senescence dates matched ripeness stages of winter crops and the dates with maximum NDVI were closely related to the field-observed heading stage of winter cereals. We showed that the emergence of winter crops corresponded to the dates extracted with a threshold of 0.1, which translated into 8.89 days of root-mean-square error (RMSE) improvement compared to the standard threshold of 0.5. The method with optimised settings and thresholds can be easily transferred and applied to areas with similar growing conditions. Altogether, the results improve our understanding of how satellite-derived phenometrics can explain in situ phenological observations.

Simulation of forest tree species’ bud burst dates for different climate scenarios: chilling requirements and photo-period may limit bud burst advancement

This study investigates whether the assumed increase of winter and spring temperatures is depicted by phenological models in correspondingly earlier bud burst (BB) dates. Some studies assume that rising temperatures lead to an earlier BB, but even later BB has been detected. The phenological model PIM (promoter-inhibitor-model) fitted to the extensive phenological database of the German Weather Service was driven by several climate scenarios. This model accounts for the complicated mechanistic interactions between chilling requirements, temperature and photo-period. It predicts BB with a r2 between 0.41 and 0.62 and a RMSE of around 1 week, depending on species. Parameter sensitivities depict species dependent interactions between growth and chilling requirements as well as photo-period. A mean trend to earlier BB was revealed for the period 2002– 2100, varying between −0.05 and −0.11 days per year, depending on species. These trends are lower than for the period 1951– 2009. Within climate scenario period, trends are decreasing for beech and chestnut, stagnating for birch and increasing for oak. Results suggest that not fulfilled chilling requirements accompanied by an increasing dependency on photo-period potentially limit future BB advancement. The combination of a powerful phenological model, a large scale phenological database and several climate scenarios, offers new insights into the mechanistic comprehension of spring phenology.

Phase difference analysis of temperature and vegetation phenology for beech forest: a wavelet approach

Temporal shifts in phenology or vegetation period of plants are seen as indicators of global warming with potentially severe impacts on ecosystem functioning. In spite of increasing knowledge on drivers, it is of utmost importance to disentangle the relationship between air temperatures, phenological events, potential temporal lags (phase shifts) and time scale for certain plant species. Assessing the phase shifts as well as the scale-dependent relationship between temperature and vegetation phenology requires the development of a nonlinear temporal model. Therefore, we use wavelet analysis and present a framework for identifying scale-dependent cross-phase coupling of bivariate time series. It allows the calculation of (a) scale-dependent decompositions of time series, (b) phase shifts of seasonal components in relation to the annual cycle, and (c) inter-annual phase differences between seasonal phases of different time series. The model is applied to air temperature data and remote sensing phenology data of a beech forest in Germany. Our study reveals that certain seasonal changes in amplitude and phase with respect to the normal annual rhythm of temperature and beech phenology are coupled time-delayed components, which are characterized by a time shift of about one year.

Validating MODIS and Sentinel-2 NDVI Products at a Temperate Deciduous Forest Site Using Two Independent Ground-Based Sensors

Quantifying the accuracy of remote sensing products is a timely endeavor given the rapid increase in Earth observation missions. A validation site for Sentinel-2 products was hence established in central Germany. Automatic multispectral and hyperspectral sensor systems were installed in parallel with an existing eddy covariance flux tower, providing spectral information of the vegetation present at high temporal resolution. Normalized Difference Vegetation Index (NDVI) values from ground-based hyperspectral and multispectral sensors were compared with NDVI products derived from Sentinel-2A and Moderate-resolution Imaging Spectroradiometer (MODIS). The influence of different spatial and temporal resolutions was assessed. High correlations and similar phenological patterns between in situ and satellite-based NDVI time series demonstrated the reliability of satellite-based phenological metrics. Sentinel-2-derived metrics showed better agreement with in situ measurements than MODIS-derived metrics. Dynamic filtering with the best index slope extraction algorithm was nevertheless beneficial for Sentinel-2 NDVI time series despite the availability of quality information from the atmospheric correction procedure.

Radiometric Correction of Simultaneously Acquired Landsat-7/Landsat-8 and Sentinel-2A Imagery Using Pseudoinvariant Areas (PIA): Contributing to the Landsat Time Series Legacy

The use of Pseudoinvariant Areas (PIA) makes it possible to carry out a reasonably robust and automatic radiometric correction for long time series of remote sensing imagery, as shown in previous studies for large data sets of Landsat MSS, TM, and ETM+ imagery. In addition, they can be employed to obtain more coherence among remote sensing data from different sensors. The present work validates the use of PIA for the radiometric correction of pairs of images acquired almost simultaneously (Landsat-7 (ETM+) or Landsat-8 (OLI) and Sentinel-2A (MSI)). Four pairs of images from a region in SW Spain, corresponding to four different dates, together with field spectroradiometry measurements collected at the time of satellite overpass were used to evaluate a PIA-based radiometric correction. The results show a high coherence between sensors (r2 = 0.964) and excellent correlations to in-situ data for the MiraMon implementation (r2 > 0.9). Other methodological alternatives, ATCOR3 (ETM+, OLI, MSI), SAC-QGIS (ETM+, OLI, MSI), 6S-LEDAPS (ETM+), 6S-LaSRC (OLI), and Sen2Cor-SNAP (MSI), were also evaluated. Almost all of them, except for SAC-QGIS, provided similar results to the proposed PIA-based approach. Moreover, as the PIA-based approach can be applied to almost any image (even to images lacking of extra atmospheric information), it can also be used to solve the robust integration of data from new platforms, such as Landsat-8 or Sentinel-2, to enrich global data acquired since 1972 in the Landsat program. It thus contributes to the program’s continuity, a goal of great interest for the environmental, scientific, and technical community.

Measuring β-diversity by remote sensing: a challenge for biodiversity monitoring

Biodiversity includes multiscalar and multitemporal structures and processes, with different levels of functional organization, from genetic to ecosystemic levels. One of the mostly used methods to infer biodiversity is based on taxonomic approaches and community ecology theories. However, gathering extensive data in the field is difficult due to logistic problems, overall when aiming at modelling biodiversity changes in space and time, which assumes statistically sound sampling schemes. In this view, airborne or satellite remote sensing allow to gather information over wide areas in a reasonable time. Most of the biodiversity maps obtained from remote sensing have been based on the inference of species richness by regression analysis. On the contrary, estimating compositional turnover (beta-diversity) might add crucial information related to relative abundance of different species instead of just richness. Presently, few studies have addressed the measurement of species compositional turnover from space. Extending on previous work, in this manuscript we propose novel techniques to measure beta-diversity from airborne or satellite remote sensing, mainly based on: i) multivariate statistical analysis, ii) the spectral species concept, iii) self-organizing feature maps, iv) multi- dimensional distance matrices, and the v) Raos Q diversity. Each of these measures allow to solve one or several issues related to turnover measurement. This manuscript is the first methodological example encompassing (and enhancing) most of the available methods for estimating beta-diversity from remotely sensed imagery and potentially relate them to species diversity in the field.