Movement ecology of small wild animals is often reliant on radio‐tracking methods due to the size and weight restrictions of available transmitters. In manual radio telemetry, large errors in spatial position and infrequent relocations prevent the effective analysis of small‐scale movement patterns and dynamic aspects of habitat selection. Automatic radio‐tracking systems present a potential solution for overcoming these drawbacks. However, existing systems use customized electronics and commercial software or exclusively record presence/absence data instead of triangulating the position of tagged individuals. We present a low‐cost automatic radio‐tracking system built from consumer electronic devices that can locate the position of radio transmitters under field conditions. We provide information on the hardware components, describe mobile and stationary set‐up options, and offer open‐source software solutions. We describe the workflow from hardware setup and antenna calibration, to recording and processing the data and present a proof of concept for forest‐dwelling bats using a mixed forest as study area. With an average bearing error of 6.8° and a linear error of 21 m within a distance ranging from 65 m to 190 m, the accuracy of our system exceeds that of both traditional methods as well as manual telemetry. This affordable and easy‐to‐use automatic radio‐tracking system complements existing tools in movement ecology research by combining the advantages of lightweight and cost‐efficient radio telemetry with an automatic tracking set‐up.

Unmanned aerial systems (UAS) are cost-effective, flexible and offer a wide range of applications. If equipped with optical sensors, orthophotos with very high spatial resolution can be retrieved using photogrammetric processing. The use of these images in multi-temporal analysis and the combination with spatial data imposes high demands on their spatial accuracy. This georeferencing accuracy of UAS orthomosaics is generally expressed as the checkpoint error. However, the checkpoint error alone gives no information about the reproducibility of the photogrammetrical compilation of orthomosaics. This study optimizes the geolocation of UAS orthomosaics time series and evaluates their reproducibility. A correlation analysis of repeatedly computed orthomosaics with identical parameters revealed a reproducibility of 99% in a grassland and 75% in a forest area. Between time steps, the corresponding positional errors of digitized objects lie between 0.07 m in the grassland and 0.3 m in the forest canopy. The novel methods were integrated into a processing workflow to enhance the traceability and increase the quality of UAS remote sensing.

Previous studies of clouds over the Tibetan Plateau (TiP) were subject to limitations. Surface observations are scarce, and satellite retrievals are not well adapted to the peculiar conditions of the TiP. For the most comprehensive existing cloud data set, provided by the International Satellite Cloud Climatology Project (ISCCP), issues were reported for the TiP. It also lacks sufficient spatiotemporal resolution for this topographically complex region. With the Indian Ocean Data Coverage service, European Organisation for the Exploitation of Meteorological Satellites provides a Meteosat data set between 1998 and 2008. The resolution of around 6 km at the study area is sufficient even for complex terrain. Based on this data set and on products of the active sensor onboard CloudSat, we develop a novel gross‐cloud retrieval for the TiP using logistic regression models. The approach maintains the original Meteosat resolution. Validation against independent CloudSat data reveals good performance. The approach also outperforms the ISCCP pixel level (DX) data set. The resulting data set is the first for the TiP that provides cloud information with sufficient resolution for both day and night. Patterns of cloud frequencies during winter, premonsoon, and monsoon seasons are analyzed. Strong diurnal forcing is found for the plateau. Peaks of cloud frequencies above the slopes occur during afternoon, while they are delayed in the valleys, where high cloud frequencies persist throughout the nights. Above the lower parts of the southern foothills of the Himalayas cloud frequencies were for the first time found to increase until the early morning. Katabatic flows are suspected to be responsible for this pattern by initiating the formation of mesoscale convective systems.

In the realm of solar photovoltaic system image segmentation, existing deep learning networks focus almost exclusively on single image sources both in terms of sensors used and image resolution. This often prevents the wide deployment of such networks. Our research introduces a novel approach to train a network on a diverse range of image data, spanning UAV, aerial, and satellite imagery at both native and aggregated resolutions of 0.1 m, 0.2 m, 0.3 m, 0.8 m, 1.6 m, and 3.2 m. Using extensive hyperparameter tuning, we first determined the best possible parameter combinations for the network based on the DeepLabV3 ResNet101 architecture. We then trained a model using the wide range of different image sources. The final network offers several advantages. It outperforms networks trained with single image sources in multiple test applications as measured by the F1-Score (95.27%) and IoU (91.04%). The network is also able to work with a variety of target imagery due to the fact that a diverse range of image data was used to train it. The model is made freely available for further applications.

Deep learning (DL)—in particular convolutional neural networks (CNN)—methods are widely spread in object detection and recognition of remote sensing images. In the domain of DL, there is a need for large numbers of training samples. These samples are mostly generated based on manual identification. Identifying and labelling these objects is very time-consuming. The developed approach proposes a partially automated procedure for the sample creation and avoids manual labelling of rooftop photovoltaic (PV) systems. By combining address data of existing rooftop PV systems from the German Plant Register, the Georeferenced Address Data and the Official House Surroundings Germany, a partially automated generation of training samples is achieved. Using a selection of 100,000 automatically generated samples, a network using a RetinaNet-based architecture combining ResNet101, a feature pyramid network, a classification and a regression network is trained, applied on a large area and post-filtered by intersection with additional automatically identified locations of existing rooftop PV systems. Based on a proof-of-concept application, a second network is trained with the filtered selection of approximately 51,000 training samples. In two independent test applications using high-resolution aerial images of Saarland in Germany, buildings with PV systems are detected with a precision of at least 92.77 and a recall of 84.47.

Successful conservation strategies require frequent observations and assessments of the landscape. Although expert surveys provide a great level of detail, the trade-off is the limited spatial coverage and repetition with which they are executed. Remote sensing technology can partially resolve these issues; nevertheless, it still requires experts’ experience to create conservation planning and reaction options. Nature 4.0 seeks to address these shortcomings by developing a prototype of a modular environmental monitoring system for high-resolution observation of species, habitats, and processes. The project combines expert surveys by nature conservationists, remote sensing, and a network of environmental sensors, which are integrated into stationary units as well as attached to unmanned aerial vehicles, rovers, or animals. By utilizing powerful data integration and analysis methods, Nature 4.0 will enable researchers to effectively observe landscapes through a set of diverse lenses. Time series data from the project will also inform the development of early warning indicators. Following the open-source principle, as much of the project as possible will be made publicly available, including, for instance, schematics for sensor units, algorithms for data integration or information on species occurrence. In summary, Nature 4.0 will establish new methods and protocols in the field of comprehensive environmental monitoring by combining traditional sampling, remote sensing, and automated measurement stations. The prototype system is being developed in the Marburg Open Forest, an open research, education, and development platform for environmental monitoring methods. The Marburg Open Forest brings a cross-disciplinary group of scientists together with nature conservation experts from the private sector and the state government, as well as local schools and private citizens to collaborate and bridge the gap between basic and applied environmental research. After one year, we will present the results of the initial phase and share our experience with developing Nature 4.0.