CWChristine I. B. Wallis

Christine I. B. Wallis

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Christine I. B. Wallis has a total of 72 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Remotely-sensed vegetation and habitat structure can serve as suitable surrogates to predict the distribution of a micro-endemic bird species

    by Edwin Zárate, Christine Wallis, Vinicio Santillán, Roland Brandl, Nina Farwig, Jörg Bendix

    Biodiversity is increasingly threatened by human land use and climate change, making predictive modeling crucial for conservation planning and the identification of priority conservation areas. Large-scale species distribution projections can be improved by integrating remotely sensed vegetation indices, as they reflect important vegetation and habitat characteristics. The Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI) indicate vegetation productivity, while texture metrics derived from these indices reveal habitat structure. When combined with climatic variables, these indicators can significantly improve the accuracy of species distribution models (SDMs). In this study, we evaluated the performance of SDMs using vegetation indices, texture metrics, and climatic variables to predict the distribution of the Violet-throated Metaltail (Metallura baroni), a microendemic hummingbird restricted to the environmentally complex high-altitude regions of the southern Ecuadorian Andes. Using a backward-selection and cross-validation approach for predictor selection and the Maximum Entropy (MaxEnt) algorithm, we compared model performance using AUC values. Our results demonstrate that incorporating habitat structure indicators (NDVI- and NDWI-derived texture metrics) together with climatic variables significantly improves SDM performance, allowing better discrimination of shrub ecosystems where M. baroni occurs. This approach highlights the importance of integrating key aspects of habitat structure as indicators of resource availability, which directly influence the distribution of bird species in complex landscapes.

  • Open Access

    A data-driven framework for forest conservation prioritisation: From citizen science data to protected area designation

    by Katharina H. Horn, Daniele Silvestro, Christine I.B. Wallis, Pedro J. Leitão, Ender Daldaban, K. Matthias B. Herkt, Birgit Kleinschmit, Annette Rudolph


  • Open Access

    Detecting forest disturbances in Germany from satellite time series using unsupervised LSTM autoencoders

    by Ines Grünberg, Michael Förster, Robert Jackisch, Christine Wallis

    Climate change and natural disturbances are posing an increasing pressure on European forests and have led to extensive forest losses over recent decades. Large-scale monitoring of forest dynamics is therefore essential and can be effectively supported by remote sensing techniques. Previous studies have demonstrated the potential of supervised and unsupervised machine learning approaches for detecting forest disturbances, typically characterising forest condition at specific points in time. At the same time, comprehensive reference data remain scarce at large spatial scales.Against this background, we investigate the potential of an unsupervised deep learning approach for large-scale detection of forest anomalies across Germany within the framework of the EO4Nature project. We apply a deep learning based Long Short-Term Memory (LSTM) autoencoder to model vegetation trajectories over multiple vegetation periods to capture gradual changes in forest vitality.The LSTM model is trained on stratified healthy forest pixels across Germany, selected based on a low disturbance probability derived from the European Forest Disturbance Atlas (EFDA). We compare multiple model configurations using different input feature sets based on Sentinel-2 data at a monthly temporal resolution for the period 2018-2025. Anomalies in forest vitality are detected based on the reconstruction error of the autoencoder, using adaptive thresholds that account for seasonal variation and forest type. This enables the identification of pixels with different levels of anomaly severity.We primarily evaluate the proposed approach using independent disturbance reference data at the local scale. High-resolution annual orthophotos from multiple disturbed forest sites in Germany are used to enable a detailed spatial assessment of detected anomalies.In addition, we conducted a preliminary large-scale consistency check by comparing areas exhibiting high anomaly scores with disturbed forest regions derived from the EFDA. These initial results indicate that the unsupervised LSTM autoencoder, trained on stable forest conditions using NDVI, NBR and abiotic variables, produces a continuous anomaly score that correlates with independently mapped disturbance patterns (Spearman’s ρ = 0.65, p < 0.001), demonstrating consistency with external disturbance probabilities.The results give insight into the disturbance intensities at which deviations from healthy forests dynamics become detectable and provide knowledge about the most relevant spectral features for large-scale monitoring of forest ecosystem stability.

  • Open Access

    Forest recovery analysis combining AI with multi-platform LiDAR and UAV-based hyperspectral imaging (KI-Recover)

    by Robert Jackisch, Caileigh Shoot, Christine Wallis, Jakob Ebenbeck, Max Mangold, Anna-Lena Thran, Marco Heurich

    Temperate forest ecosystems are increasingly pressured by climate warming and drought events that drive disturbances such as storms, wildfires and calamities. Forest dieback caused by windthrow and bark beetle infestations has increased significantly in severity and affected entire regions in Germany.With the progression of remote sensing and unoccupied aerial vehicles (UAV, drone) as sensor platform, precise monitoring of extensive forest areas at high resolution is feasible. The project KI-Recover integrates AI-driven multi-sensor data analysis of diverse sites following significant disturbances. Our surveys were conducted during summer 2025 in the Bavarian Forest National Park and the Harz National Park. Both regions are characterized by disturbance legacies, recent dieback and minimal forest management.Within these regions, we chose forest stands based on disturbance type and history to allow for natural regeneration, except for two sites with recent wildfire. Monitoring utilized UAV-hyperspectral scanning, multispectral and RGB mapping and UAV-LiDAR. An extensive ground campaign provided forest inventory adapted for remote sensing, vegetation species and deadwood mapping. Additionally, mobile laser scanning was employed to obtain fine-scaled 3D information of forest metrics, e.g. forest structural complexity.We present initial results of our integrated multi-modal geospatial modelling approach. Forest inventory at image level is conducted via instance segmentation of individual living trees, as well as standing and lying deadwood at various decay stages, using convolutional neural network (CNN) architectures. A large training and validation database was created by manual annotations and labelling of RGB and multispectral data. Detailed volumetric forest structure was extracted from fused mobile and UAV LiDAR, to overcome the scale gap between. Hyperspectral transect data is used i.e. to model species richness and measure plant vitality. All methods combined will inform indicators of successional development, stand dynamics, and species establishment.The overarching goal of this project is to couple the geospatial remote sensing surveys with a forest succession prediction and detailed AI-driven climate modelling to assess effects of extreme events, heat stress and drought, and to provide data-driven recommendations for forest management.

  • Open Access

    Scalable Mapping of Urban Green Cooling Services Using AlphaEarth Foundations

    by Hanyu Li, Alby Duarte Rocha, Christine Wallis

    Under the combined influence of global warming and rapid urbanization, extreme heat has become a major challenge for urban resilience and public health. Urban green infrastructure provides important cooling benefits through evapotranspiration and shading, yet spatially explicit assessment of these services remains challenging. Existing approaches rely heavily on computationally expensive physical models and dense input data, which limits their applicability beyond well-studied regions. In this study, we present a scalable approach for mapping urban green cooling services by combining Earth observation foundation models with insights from process-based modeling. We use the Green Cooling Services Index (GCoS) as the core metric, which is derived from simulations of the Soil-Canopy-Observation of Photosynthesis and Energy Fluxes (SCOPE) model. To enable large-scale applications, we build a surrogate model that maps annual multimodal satellite embedding vectors from AlphaEarth Foundations to GCoS reference data. These embeddings integrate multisource Earth observation information across the full year, capturing key vegetation phenology and climate dynamics. The analysis covers 14 Functional Urban Areas across Europe and surrounding regions. Model performance is evaluated through three complementary experiments: a continent-scale assessment, a leave-one-city-out test, and stratified error analyses in representative cities. Results show that the surrogate approach can reproduce vegetation cooling effects with high accuracy while requiring substantially fewer data and computational resources than conventional physical models. Importantly, the model maintains stable performance when applied to cities not included in training. This framework addresses a key scalability gap in urban heat assessments and enables consistent mapping of green cooling services in data limited regions.

  • Open Access

    When is evapotranspiration at an urban tree site energy-limited versus water-limited? Evidence from eddy-covariance and soil moisture measurements in Berlin

    by Basem Aljoumani, Alby Durate Rocha, Stenka Vulova, Fred Meier, Benjamin Dechant, Salwa Saidi, Christine Wallis

    Urban vegetation helps reduce heat stress through evapotranspiration (ET) and shading. However, we still do not fully understand how atmospheric energy demand and soil water availability influence ET in an urban environment. Specifically, it is unclear when urban ET shifts from being limited by energy to being limited by water. In this study, we examine this transition at an urban tree-dominated site in Berlin, Germany. We use eddy-covariance (EC) flux measurements combined with soil moisture observations at various depths.We analyze two years (2019-2020) of EC data, which includes latent heat flux, net radiation, and weather variables, as well as soil moisture observations at six depths ranging from 5 cm to 1 m. After applying quality control, turbulence filtering, and selecting condensation-free daytime data, we aggreage the ET data to daily daytime averages. We then relate ET to net radiation, vapor pressure deficit (VPD), and soil moisture. We represent near‑surface soil moisture as the average of the 5-20 cm layers, which reflects the most dynamically active portion of the root zone under the constrained soil conditions typical of this urban environment.We hypothesize that ET is mainly energy-limited when the soil is wet. However, as the topsoil dries, it becomes water-limited, even when the atmospheric demand is high. Our hypothesis is supported by several exploratory analyses. Scatter plots showing ET against net radiation, classified by soil moisture levels (wet, medium, dry), reveal three consistent trends: (A) in wet conditions, ET rises sharply with radiation, showing an energy-limited state; (B) in dry conditions, the ET-radiation relationship weakens and levels off, indicating water limitation despite high radiation; and (C) in intermediate soil moisture conditions, we see a transitional response. Linear regression models demonstrate that the slope of the ET-radiation relationship significantly declines from wet to dry soil states. Adding VPD enhances the performance of the linear regression model (R² ≈ 0.75), highlighting the influence of atmospheric demand. Meanwhile, the interaction terms between soil moisture and radiation remain significantly important. A linear mixed-effects model, which includes year as a random factor, produces similar results, indicating that these patterns hold steady across different years. Segmented regression of ET against topsoil moisture identified a statistically significant breakpoint at approximately 8-9% volumetric soil moisture, marking a transition from water-limited conditions at low soil moisture to weak ET sensitivity at higher soil moisture. Below this threshold, ET responds strongly to changes in soil moisture, indicating a water-limited regime. Above the threshold, ET shows little additional sensitivity to soil moisture and is predominantly controlled by energy availability. In conclusion, our results provide clear quantitative evidence that urban evapotranspiration alternates between energy-limited and water-limited regimes, with shallow soil moisture exerting a dominant control during dry periods. These findings highlight the vulnerability of urban vegetation to soil drying and have important implications for urban climate adaptation, green infrastructure management, and land-atmosphere modeling under increasing drought frequency.

  • Open Access

    Identifying valuable forest habitats for conservation in north-western Germany using AI and citizen science

    by Katharina Horn, Daniele Silvestro, Christine Wallis, Pedro J. Leitao, Ender Daldaban, Annette Rudolph

    Around the globe we experience a significant biodiversity loss, mainly driven by direct anthropogenic exploitation, land use changes, and climate change. The most effective strategy to limit biodiversity loss is the designation and management of protected areas. Consequently, the European Union has adopted the EU Biodiversity Strategy for 2030, aiming to protect 30% of aquatic and terrestrial ecosystems by 2030. However, a consistent framework to designate protected areas across all EU member states is lacking. Additionally, the monitoring of biodiversity is challenged by the dynamic nature of the biological system, exacerbated by ongoing climate change, putting additional pressure on the member states in the identification of suitable areas for conservation. In contrast, the increasing amount of detailed geospatial and climatic data contains valuable information that can be used to optimise protected area designation. Recent developments in artificial intelligence and machine learning now provide us with powerful tools to best utilise these vast amounts of data. In this study, we develop a transparent and reproducible framework to prioritise protected areas in forests. Here we apply the CAPTAIN framework based on reinforcement learning (RL) to identify valuable forest habitats for conservation in the federal state of North Rhine-Westphalia (NRW), Germany. First, we model habitats of ten forest bird indicator species across the period of 2016-2024. Second, we use the changing habitat patterns to train a RL model that identifies 30% of the most valuable forest sites in the federal state. Finally, we model valuable forest sites under different policies (e.g., including or excluding opportunity costs for nature conservation) to illustrate how potential limitations of nature conservation management can be addressed. Our results indicate that forest sites in the south-east of NRW are most suitable for conservation. Furthermore, we find that including opportunity costs for nature conservation in the model predictions produces similarly strong outcomes for safeguarding the most endangered bird species. The framework makes use of open-source data and can be applied to any other region or country to support strategic nature conservation management.

  • Open Access

    Integrating Open-Source Data and Advanced Machine Learning for Forest Conservation Prioritisation in Germany

    by Katharina Horn, Daniele Silvestro, Christine Wallis, Annette Rudolph

    Biodiversity is declining worldwide, driven by anthropogenic pressures such as land sealing, air pollution, agricultural practices, forest clearing, and climate-change impacts. These stressors interact in complex ways, altering habitats, disrupting ecosystem functions, and challenging the long-term survival of many species. In response, policy frameworks such as the European Union’s Biodiversity Strategy for 2030 aim to protect 30% of terrestrial and aquatic ecosystems by the end of the decade. Meeting these goals requires transparent and scalable approaches for identifying ecologically valuable areas. However, protected-area designation across Europe remains inconsistent, with national frameworks differing and often lacking transferability. At the same time, the growth of geospatial datasets, citizen science observations, and environmental time-series data offers new opportunities for data-driven conservation planning. Advances in machine learning lead to new tools to analyse heterogeneous datasets and support decision-making in complex ecological contexts.In this study, we use open-source environmental data together with citizen-science species observations collected between 2016 and 2024. These time-series data allow us to assess how environmental conditions, land use, and climate variability influence species occurrences over time. Using these inputs, we apply species distribution modelling to estimate habitat suitability and to map how suitable environments shift in response to external pressures. To support conservation planning, we apply the CAPTAIN reinforcement learning framework. Developed by Silvestro et al., this tool (2022) enables the optimisation of different conservation targets under ecological, spatial, and socio-economic constraints. It is based on the interaction between an agent and its environment, where the agent learns to take decisions that maximise a defined reward. The approach evaluates trade-offs, such as balancing biodiversity outcomes with economic considerations, and produces spatially explicit prioritisation maps that identify forest areas of consistently high ecological value, even under changing environmental conditions. The resulting framework is designed to be transferable and scalable. It supports nature-conservation planning in Germany and can be adapted to other regions, ecosystems, or policy objectives. By combining open data, citizen-science observations, and a machine learning-based optimisation, the approach contributes to national biodiversity strategies and aligns with broader international efforts such as the EU Biodiversity Strategy for 2030.

  • Open Access

    Exploring the spectral variation hypothesis for α- and β-diversity: a comparison of open vegetation and forests

    by Christine I B Wallis, Shan Kothari, Johanna R. Jantzen, Anna L. Crofts, Sabine St-Jean, Deep Inamdar, J. Pablo Arroyo-Mora, Margaret Kalacska, Anne Bruneau, Nicholas C Coops, et al.

    Airborne hyperspectral imaging holds great promise for estimating plant diversity and composition, given its unprecedented combination of aerial coverage, spatial resolution, and spectral detail. Recently, there has been renewed attention toward the spectral variation hypothesis (SVH), which predicts that higher spectral variation is correlated with greater plant diversity. While several studies have highlighted methodological challenges involved with the SVH, there is little consensus about when it yields strong predictions of taxonomic, functional, and phylogenetic diversity. In part, this may be because prior studies have not explicitly considered how underlying environmental gradients drive changes in spectral and species composition. In this study, we tested the SVH separately in open vegetation (i.e. grasses and shrubs) and in forests at five sites across Canada. Generalized additive models revealed that spectral diversity was a better predictor of functional α-diversity than of taxonomic or phylogenetic α-diversity in both vegetation types. Mantel tests and Procrustes analyses revealed weak to moderate associations between spectral and plant β-diversity and composition in open vegetation, and moderate associations in forests. The better fit in forests appeared to be influenced by the presence of an elevational gradient and associated species turnover (from deciduous to coniferous trees); we observed weaker relationships when examining only a subset of this gradient. We suggest that the high variability in the strength of associations between plant and spectral diversity reported to date might be affected by the presence of environmental gradients. Finally, we found that different wavelength bands contributed to spectral α-diversity in open vegetation vs. forests, suggesting different spectral features are important for different vegetation types. In conclusion, spectral diversity is a potentially powerful tool for biodiversity assessment, but it requires a context-specific approach.