BG9.4 | Large-scale mapping of environmental variables by combining ground observations, remote sensing, and machine learning
EDI
Large-scale mapping of environmental variables by combining ground observations, remote sensing, and machine learning
Co-organized by ESSI4
Convener: Hanna Meyer | Co-conveners: Benjamin Dechant, Alvaro Moreno, Jacob Nelson, Madlene NussbaumECSECS
Orals
| Thu, 18 Apr, 08:30–12:30 (CEST)
 
Room 2.23
Posters on site
| Attendance Thu, 18 Apr, 16:15–18:00 (CEST) | Display Thu, 18 Apr, 14:00–18:00
 
Hall X1
Posters virtual
| Attendance Thu, 18 Apr, 14:00–15:45 (CEST) | Display Thu, 18 Apr, 08:30–18:00
 
vHall X1
Orals |
Thu, 08:30
Thu, 16:15
Thu, 14:00
Environmental data from large measurement campaigns and automated measurement networks are increasingly available and provide relevant information of the Earth System. However, such data are usually only available as point observations and only represent a small part of the Earth´s surface. Upscaling strategies are hence needed to provide continuous and comprehensive information as a baseline to gain insights on large-scale spatio-temporal dynamics.
In the upscaling, machine learning algorithms that can account for complex and nonlinear relationships are increasingly used to link remote sensing datasets to reference measurements. The resulting models are then applied to provide spatially explicit predictions of the target variable, often even on a global scale.
Due to easy access to user-friendly software, model training and spatial prediction using machine learning algorithms is nowadays straightforward at first sight. However, considerable challenges remain: dealing with reference data that are not independent and identically distributed, accounting for spatial heterogeneity when scaling reference measurements to the grid cell scale, appropriately evaluating the resulting maps and quantifying their uncertainties, generating robust maps that do not suffer from extrapolation artifacts as well as the strategies for model interpretation and understanding. This session invites contributions on the methodology and application of large-scale mapping strategies in different disciplines, including vegetation characteristics such as foliar or canopy traits and photosynthesis, soil characteristics such as soil organic carbon, or atmospheric parameters such as pollutant concentration. Methodological contributions can focus on individual aspects of the upscaling approach, such as the design of measurement campaigns or networks to increase representativeness, novel algorithms or validation strategies as well as uncertainty assessment.

Orals: Thu, 18 Apr | Room 2.23

Chairpersons: Hanna Meyer, Jacob Nelson
08:30–08:35
08:35–08:45
|
EGU24-6174
|
On-site presentation
Jens Kattge, Gerhard Boenisch, Benjamin Dechant, Álvaro Moreno-Martínez, Teja Kattenborn, Sandra Díaz, Sandra Lavorel, Iain Colin Prentice, Paul Leadley, Christian Wirth, and the TRY Consortium

The TRY initiative (https://www.try-db.org) was established in 2007 on request from IGBP and DIVERSITAS to develop a joint database of in-situ measured plant traits supporting vegetation modelling and biodiversity research. Based on the mandate from the two initiatives, TRY has received significant contributions of original and integrated datasets from the global research community and achieved unprecedented coverage. The TRY database is regularly updated, and since 2019, data have been publicly available under a CC BY license. 

Trait data from the TRY database are frequently used to map plant traits at global scales. We here provide an overview of the TRY database, briefly summarise trait mapping approaches highlighting caveats, and provide an outlook on upcoming developments in the context of the TRY database.

How to cite: Kattge, J., Boenisch, G., Dechant, B., Moreno-Martínez, Á., Kattenborn, T., Díaz, S., Lavorel, S., Prentice, I. C., Leadley, P., Wirth, C., and Consortium, T. T.: TRY - Plant Trait Database, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6174, https://doi.org/10.5194/egusphere-egu24-6174, 2024.

08:45–08:55
|
EGU24-8898
|
ECS
|
On-site presentation
Valerie Zermatten, Javiera Castillo-Navarro, Diego Marcos, and Devis Tuia

Natural ecosystem maps are a fundamental tool for describing natural habitats. They are used when analysing ecological networks, for studying ecological connectivity, for conservation planning or for the management of ecosystem services. While remote sensing information is commonly employed for mapping land cover, accurate ecosystem maps require going beyond the classification of the physical surface of the land and attempting to distinguish different communities of living beings. New research directions for ecosystem classification leverage species observations created by citizen scientists on social media or crowd-sourcing platforms, profiting from their extensive spatial and temporal coverage and their low acquisition cost.
In this work, we make complimentary use of crowd-sourced data with remote sensing imagery to produce ecosystem maps in alpine areas. Our study area covers the state of Valais, a territory of about 5, 000km2 in southwestern Switzerland. We retrieve nearly 3 million species observations from the Global Biodiversity Information Facility (GBIF) database that includes governmental, crowd-sourced and scientific observations. We combine the species data with high-resolution aerial remote sensing imagery provided by the Swiss Federal Office of Topography swisstopo. As reference ecosystem maps, we follow the European Nature Information System (EUNIS) at a 100m scale. To solve the task of classifying ecosystems, we propose a multi-modal data fusion approach based on a multi-modal transformer architecture. Such a model can handle redundant and complementary information coming from different data sources and provide an explicit and interpretable decision through the visualisation of its attention scores. Through our preliminary experiments, we observed that the unequal distribution of samples between the classes and also the sampling biases negatively impacted the performance of our approach. We are working towards a more informative inclusion of species observation and a more balanced learning of each ecosystem type.

How to cite: Zermatten, V., Castillo-Navarro, J., Marcos, D., and Tuia, D.: Ecosystem mapping with remote sensing images and ground observations, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8898, https://doi.org/10.5194/egusphere-egu24-8898, 2024.

08:55–09:05
|
EGU24-10647
|
ECS
|
Virtual presentation
Achala Shakya

With the advent of a large number of available satellite imagery for classification, several classification techniques have been developed to classify these images over the years. Each classification method has its own set of challenges that restrict its use. Hence, a novel research method is proposed using hybrid techniques (Machine Learning and Deep Learning) and geospatial techniques for enhancing land use and land cover mapping using satellite images. Furthermore, this study will provide insights into integrated agricultural management to achieve the UN’s Sustainable Development goals. This study uses freely available satellite images, i.e., Sentinel 1 and Sentinel 2, for classifying land use/ land cover over an agricultural area in Hissar, India. The major crops grown in this area include paddy, maize, cotton, and pulses during Kharif (summer) and wheat, sugarcane, mustard, gram, and peas during Rabi (winter) seasons. The datasets for the study area were pre-processed using SNAP and ArcGIS software. After pre-processing, a comprehensive feature set is identified consisting of Polarimetric features such as elements of covariance matrix, Entropy/scattering angle (alpha), and traditional geometric features such as shape, size, and area. After this phase, dimensionality reduction techniques (such as PCA) were applied to reduce the no. of features to the most important ones. Utilizing the feature set, a hybrid machine learning model is constructed using ground truth images by fine-tuning the model. For the best split, the test and train ratio is divided into 70:30. Optical (Sentinel 1) and Microwave (Sentinel 2) datasets are fused and the quality of the fused image was evaluated using several fusion metrics, including Erreur Relative Globale Adimensionnelle de Synthese (ERGAS), Spectral Angle Mapper (SAM), Relative Average Spectral Error (RASE), Universal Image Quality Index (UIQI), Structural Similarity Index (SSIM), Peak Signal-to-Noise Ratio (PSNR), and Correlation Coefficient (CC). After obtaining the resultant fused image hybrid technique is used for the final classification. The classification accuracy is represented using overall classification accuracy and kappa value. A comparison of classification results indicates a better performance by the hybrid technique with an overall accuracy of 92%.

How to cite: Shakya, A.: Enhancement of Land Use and Land Cover Mapping using Satellite Imagery through Machine Learning Techniques , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10647, https://doi.org/10.5194/egusphere-egu24-10647, 2024.

09:05–09:15
|
EGU24-20365
|
Virtual presentation
Talie Musavi, Jon Skøien, Bálint Czúcz, Andrea Hagyo, Fernando Sedano, Laura Martinez-sanchez, Renate Koeble, Marijn van der velde, Jean-Michel Terres, and Raphaël d’Andrimont D'andrimont

This study conducts a comprehensive comparison of landscape feature data derived from different sources — Small Woody Features (SWF) product from Copernicus Land Monitoring system, LUCAS Landscape Feature (LF) Module, and LUCAS Transect Module — in EU agricultural landscapes. Additionally, we consider the European Monitoring of Biodiversity in Agricultural Landscapes (EMBAL) project's approach of in-situ data collection for land cover, landscape elements, and biodiversity. This approach offers a promising avenue for integrating detailed field survey data with broad-scale remote sensing observations. Furthermore the EMBAL project has the potential to enhance the previously mentioned datasets by incorporating additional information, such as the nature value of all surveyed land units, habitat types or pollination potential among others. This inclusion could contribute to having better insights  into the ecological significance and monitoring of agricultural landscapes. Our analysis further explores the potential of incorporating cutting-edge datasets for enhanced monitoring of landscape features. Specifically, we consider the high-resolution (3-meter) dataset from Liu et al. (2023), which presents a detailed canopy height map and quantifies tree cover and woody biomass across Europe. 

We critically assess the strengths and limitations of each source: SWF's remote sensing foundation provides broad coverage but focuses only on woody features, the LUCAS LF Module combines photo-interpretation with field survey for a more detailed typology, and the LUCAS Transect, though discontinued, offered valuable field data for linear features. Strategies for monitoring landscape features have been considered for a long time, but are continually updated with the latest available data and methods. A comprehensive comparison and evaluation of the new data sources has not yet been carried out. Our study aims to identify complementarities between the different datasets to  improve both quantitative and qualitative monitoring of landscape features informing sustainable agricultural practices and biodiversity conservation strategies.

How to cite: Musavi, T., Skøien, J., Czúcz, B., Hagyo, A., Sedano, F., Martinez-sanchez, L., Koeble, R., van der velde, M., Terres, J.-M., and D'andrimont, R. D.: Bridging Field Surveys and Remote Sensing for Enhanced Landscape Feature Analysis in EU Agriculture, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20365, https://doi.org/10.5194/egusphere-egu24-20365, 2024.

09:15–09:25
|
EGU24-312
|
ECS
|
On-site presentation
Mapping the multidecadal trends of terrestrial plant nitrogen stable isotope ratios globally
(withdrawn)
Jinyan Yang, Haiyang Zhang, Yiqing Guo, Randall Donohue, Tim McVicar, Simon Ferrier, Warren Müller, Xiaotao Lü, Yunting Fang, Xiaoguang Wang, Peter Reich, Xingguo Han, and Karel Mokany
09:25–09:35
|
EGU24-17559
|
ECS
|
On-site presentation
Martí Perpinyà-Vallès, Claudia Huertas, Maria José Escorihuela, Aitor Ameztegui, and Laia Romero

In recent years, multiple remote sensing technologies have been used to quantify aboveground biomass (AGB) at different scales, from regionally trained models to global maps. The former are capable of providing higher resolution and accurate estimations albeit for smaller regions. Tailoring model weights and inputs to a specific biome or region have proved to be effective to obtain better results. Global maps, on the other hand, provide an attempt to standardizing AGB mapping at slightly to moderately lower resolutions and tend to have differences between them. A similar standardization with the benefits of regional mapping, applied across biomes, has yet to be available. For that, the first step would be comprehensively comparing state-of-the-art regional and local studies. However, existing inconsistencies in the different models and inputs used, which are often region-specific, make it impracticable.

This study addresses the need for a comparison of a single methodology consistent across different biomes in order to understand the nuances that drive the estimation of AGB. We present a data fusion approach to mapping aboveground biomass at a 20m resolution using regionally trained, regression-enhanced Random Forests in 4 different biomes: semi-arid savannas in the Sahel region, dense tropical forests in Brazil, Mediterranean forests in coastal and pre-coastal areas of Catalonia, and temperate/boreal forests in Minnesota. GEDI L4A AGB data (25-m discrete footprints) is used to train a regression model in each study area. We derived predictors from Sentinel-1 SAR, Sentinel-2 multi-spectral and Digital Elevation Model (DEM) datasets, which are common for all locations. Additionally, auxiliary data such as proximity to coastlines, human-made structures or bio-climatic variables are used to enhance predictions in saturation-prone areas. Additional to the goodness of adjustment to the training data from GEDI, we carried out a thorough validation of the results using in-situ data from Forest Inventory plots gathered in all 4 study regions. This enables a comprehensive comparison of the capabilities of Machine Learning modelling to adapt to the particular characteristics of each ecosystem. An in-depth analysis is carried out to find the most important predictor variables in each biome, as well as to assess the accuracy that can be expected across a wide range of AGB values.

How to cite: Perpinyà-Vallès, M., Huertas, C., Escorihuela, M. J., Ameztegui, A., and Romero, L.: Regionally trained models for mapping aboveground biomass from Remote Sensing data fusion: a comparison of the capabilities of Machine Learning in 4 different biomes., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17559, https://doi.org/10.5194/egusphere-egu24-17559, 2024.

09:35–09:45
|
EGU24-4977
|
On-site presentation
Development of the fine resolution fractional vegetation cover product and its application in urban area
(withdrawn)
Dan-Xia Song, Dantong Zhong, Tao He, Shuhang Gao, and Ziyi Chen
09:45–09:55
|
EGU24-15542
|
ECS
|
On-site presentation
Lucas Kugler, Christine Wessollek, and Matthias Forkel

Estimating forest canopy height is a crucial aspect in quantifying forest biomass, carbon stocks, monitoring forest degradation, and succession or restoration initiatives. Operational forest structure monitoring on a large scale involves various satellite sensors and datasets as well as large geographical variations. Challenges remain in obtaining uniformly representative ground truth data across diverse areas, phenology phases and forest types. Recently deep learning models are more frequently used to map forest canopy height at large scales by e.g., using sample observations from space-borne LiDAR sensors as training data. Training deep learning models rely on large amounts of data but are often trained on limited source domain data that is confined to cover the above-mentioned aspects. However, during testing, models may encounter out-of-distribution samples, leading to unexpected model behaviour and predictions. This vulnerability reduces the reliability of deterministic Deep Learning architectures, and finally, reliance on predictions without confidence indicators can result in misleading scientific conclusions or potentially under-informed policy decisions. Due to the varying ways of data processing, differences in forest canopy height products emerge, and hence product inter-comparison is often difficult and debatable. Existing products often do not provide any information about the confidence of their predictions.
To address this lack of confidence, we employ evidential deep Learning, adapting non-Bayesian architectures to estimate forest height and associated evidence. This approach aims to capture both aleatoric and epistemic uncertainties in areas with different forest structures investigating study areas , by incorporating evidential priors into the Gaussian likelihood function. Our method involves training a myriad of deep learning architectures including a basic CNN and Residual neuronal nets for regression to infer the hyperparameters of the evidential distribution.

Alongside other current studies, Sentinel-2 and Sentinel-1 satellite imagery serve as predictors for forest canopy height, with reference data obtained from the Global Ecosystem Dynamics Investigation (GEDI) LiDAR instrument on the International Space Station. The research explores the effects of distributional data shifts on canopy height predictions and identifies footprint samples where prediction uncertainty increases, representing different forest structures.
The application of evidential deep learning could extend far beyond this study, potentially benefiting various tasks in estimating biophysical parameters from remote sensing.

How to cite: Kugler, L., Wessollek, C., and Forkel, M.: Examining the Reliability Gap: Insights into Forest Canopy Height Using Evidential Deep Learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15542, https://doi.org/10.5194/egusphere-egu24-15542, 2024.

09:55–10:05
|
EGU24-17268
|
On-site presentation
Simon Tanner, Madlene Nussbaum, Stefan Oechslin, and Stéphane Burgos

In Switzerland, detailed soil information is missing for many regions – although urgently needed. Soon, authorities will need to legally delineate areas for the high-quality arable land inventory based on high-resolution surveys that require sampling of large numbers of new locations within the next decade. Consequently, cost and time efficient surveying strategies are required which fulfil high quality demands for legally binding decisions. The soil functional evaluation used in these decisions requires soil attributes which cannot only be derived by objective laboratory measurement but are partly based on pedological field description by experts.

In our study area of about 1’000 hectares, we established first a feature space coverage sampling design to sample 1’500 locations based on elevation and land use data, geological information, and expert knowledge of soil scientists. 170 locations were determined by a stratified random sampling design and used for independent validation of mapping results.

We predicted a large range of soil attributes (clay, silt, humus, pH, moisture regime, rootable soil depth) in multiple depth (0-20 cm, 0-30 cm, 20-30 cm, 30-50 cm, 50-100 cm), using the first 1’200 samples, random forest, and a wide range of environmental covariates. The predicted spatial information showed low to medium accuracy and maps exhibited further deficiencies, particularly poor performance for values at the tail of the soil attribute distributions. By visual inspection of prediction interval maps we found high model uncertainties in some specific areas like geological transition zones and anthropogenic altered zones through drainage and covering layers. To improve the quality of the maps we increased the total number of sampling locations up to 2’200 by two in-fill sampling design strategies in two zones:

  • To complement the feature space coverage sampling design potentially based on incomplete environmental factors, experienced surveyors directly added additional sampling locations based on their expert knowledge. Those covered landscape features not contained in the primary sampling design such as local extrema or transition zones.
  • In the second zone a two-level infill sampling design was created. A first general level complemented the feature space further as spanned in the initial sampling design. The second level consisted of additional 250 sampling points within zones of high model uncertainties.

Subsequently generated maps showed increased accuracy with increasing sampling density for most attributes, e.g., an increase of 0.1 for the clay content in the topsoil at a sampling density of 1.7 observations per ha compared to a sampling density of 1.1 Our results further displayed increasing accuracy of 0.05 with higher-weighted data collected by experts and simultaneous lowering the sampling density to 1.5 per ha by ignoring data with the lowest quality, collected before the internal calibration and synchronisation of the field survey.
To upscale the soil mapping in such high resolution and with expert-based parameters it is crucial to have synchronized data in high and stable quality across different soil formation regions.

How to cite: Tanner, S., Nussbaum, M., Oechslin, S., and Burgos, S.: Multi-stage soil surveying complementing statistical sampling designs to provide high-resolution soil maps for policy, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17268, https://doi.org/10.5194/egusphere-egu24-17268, 2024.

10:05–10:15
|
EGU24-17704
|
On-site presentation
Tomislav Hengl, Robert Minarik, Leandro Parente, and Xuemeng Tian

There is an increasing need for dynamic soil information, especially focused on monitoring soil health indicators such as soil organic carbon, soil chemical properties, soil pollution / soil degradation and status of soil micro-, meso- macro-fauna. To detect change over time, AI4SoilHealth project (https://cordis.europa.eu/project/id/101086179) is developing a spatiotemporal Machine Learning framework based on large EO data cubes (Witjes et al., 2023) to model soil health indicators e.g. to map them at high spatial resolution (30-m) with annual increments (2000–2022+) and for standard depth intervals (e.g. 0–30 cm, 30–60 cm, 60–100 cm). Produced time-series of predictions are then analyzed for trends and slope and similar coefficients are derived (per pixel) showing positive and negative changes in soil health indicators over longer periods of time (25+ years) (for the time-series method see: Hackländer et al. 2024). Areas where the trends are especially negative (e.g. significant decrease in soil carbon, significant salinization, significant loss of land cover / FAPAR etc) are flagged as requiring further soil sampling and detection of drivers of soil degradation, which should be ideally done jointly with national soil monitoring system in Europe.

The difference between spatiotemporal vs purely 2D / 3D mapping is in the three main aspects: (1) points and covariate layers are matched in spacetime (usually month-year period or at least year), (2) covariate layers are based on time-series data and include also accumulative indices (e.g. cumulative rainfall, cumulative snow cover, cumulative cropping fraction and similar) and derivatives, (3) during model training and validation, points are subset in both spacetime to avoid overfitting and bias in predictions. The rationale for using spatiotemporal machine learning is fitness of data for reliable time-series analysis: the predictions for anywhere in the spacetime cube need to be unbiased, with objectively quantified prediction errors (uncertainty), so that hence changes can be derived without a risk for serious over-/under-estimation. This framework has been tested on local and regional data sets (e.g. LUCAS soil samples covering 2009, 2012, 2015, 2018 for Europe) and can be now potentially applied using global compilations of soil points (https://opengeohub.github.io/SoilSamples/). Spatiotemporal machine learning could also potentially be used for predicting future states of soil, e.g. by extrapolating models to future climate scenarios and future land use systems (Bonannella et al., 2023). We are currently building a Soil Health Data Cube for Europe that will include some 15–20 biophysical indices (annual tillage index, bare surface cover, bimonthly FAPAR, NDWI, SAVI), climatic, terrain and parent material covariates and including the time-series of predictions of the key soil properties. This data will be made available under open data license through https://EcoDataCube.eu.

Cited references:

  • Bonannella, C., et al. (2023). Biomes of the world under climate change scenarios: increasing aridity and higher temperatures lead to significant shifts in natural vegetation. PeerJ, 11, e15593. https://doi.org/10.7717/peerj.15593 
  • Hackländer, J., et al. (2023). Land potential assessment and trend-analysis using 2000–2021 FAPAR monthly time-series at 250 m spatial resolution. PeerJ, in review. https://doi.org/10.21203/rs.3.rs-3415685/v1
  • Witjes, M., et al. (2023). Ecodatacube. eu: Analysis-ready open environmental data cube for Europe. PeerJ, 11, e15478. https://doi.org/10.7717/peerj.15478 

How to cite: Hengl, T., Minarik, R., Parente, L., and Tian, X.: Mapping dynamic soil properties at high spatial resolution using spatio-temporal Machine Learning: towards a consistent framework for monitoring soil health across borders, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17704, https://doi.org/10.5194/egusphere-egu24-17704, 2024.

Coffee break
Chairpersons: Benjamin Dechant, Alvaro Moreno, Madlene Nussbaum
10:45–10:50
10:50–11:00
|
EGU24-20044
|
ECS
|
Virtual presentation
Soil property, carbon stock and peat extent mapping at 10m resolution in Scotland using digital soil mapping techniques
(withdrawn)
Ciaran Robb, Matt Aitkenhead, Fraser MacFaralane, and Keith Matthews
11:00–11:10
|
EGU24-10504
|
ECS
|
On-site presentation
Jeran Poehls, Lazaro Alonso, Sujan Koirala, Nuno Carvalhais, and Markus Reichstein

Soil moisture is a key factor that influences the productivity and energy balance of ecosystems and biomes. Global soil moisture measurements have coarse native resolutions of 36km and infrequent revisits of around three days. However, these limitations are not present for many variables connected to soil moisture such as land surface temperature and evapotranspiration. For this reason many previous studies have aimed to discern the relationships between these higher resolution variables and soil moisture to produce downscaled soil moisture products.

In this study, we test four ensembles of simple machine learning models for this downscaling task. These ensembles use a dataset of over 1,000 sites across the US to predict soil moisture at sub-km scales. We find that all ensembles, particularly one with a very simple structure, can outperform SMAP on  a cross-fold analysis of the 1,000+ sites. This ensemble has an average ubRMSE of 0.058 vs SMAPs 0.065 and an average R of 0.639 vs SMAPs 0.562. However, not all ensembles are beneficial, with some architectures performing better with different training weights than with ensemble averaging. Additionally, we find that although general improvements over SMAP are observed, there appears to be difficulty in consistently doing so in cropland regions with high clay and low sand content.

Key Points:

  • Ensembles of simple ML architectures can downscale SWC predictions to sub 1km resolutions
  • Simpler architectures can outperform these ensembles and may be further enhanced with an improved weighting scheme during training
  • Training the models on temporally padded data provides more benefits than drawbacks in terms of overall performance.
  • The top performing ensemble is unreliable on croplands with higher than average clay and lower than average sand content.

How to cite: Poehls, J., Alonso, L., Koirala, S., Carvalhais, N., and Reichstein, M.: Downscaling Soil Moisture with Simple Machine Learning Ensembles, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10504, https://doi.org/10.5194/egusphere-egu24-10504, 2024.

11:10–11:20
|
EGU24-19789
|
ECS
|
On-site presentation
Linda Toca, Alessandro Gimona, Gillian Donaldson-Selby, Catherine Smart, Konstantinos Sideris, Jonathan Ball, and Rebekka Artz

High water table depths (WTD) and water-saturated soil are important elements for peatlands to remain healthy and are crucial targets in peatland restoration projects. Recent studies have suggested that Earth Observation data might be applicable for this task, but proposed models often lack either the spatial extension or the temporal dimension. This study has been developing a spatio-temporal model of peatland water table depth by combining time series of satellite data, namely Sentinel-1, Sentinel-2; aerial data, namely Getmapping Digital Surface Model (DSM); and field collected water table depth measurements to provide the ability to evaluate WTD changes both spatially and over time. 

Water table depth measurements were collected from 59 loggers between February and September of 2018, with loggers covering peatlands in various condition clustered around four research sites in the North of Scotland. NDVI, NDWI and OPTRAM indices were derived from reconstructed cloud-free imagery of Sentinel-2 at 5-day intervals. Similarly, VV, VH, and incidence angle values were obtained from Sentinel-1 imagery at the same time interval. Finally, a Topographic Wetness Index (TWI) was calculated using GetMapping DSM data. A Generalised Additive Model (GAM) was then fitted using all above mentioned inputs with 70% training and 30% testing split method. The model showed a good overall fit (R2=0.59 for training data; R2=0.49 for testing data), with optical covariates outperforming the radar covariates. The model was then applied spatially using the R terra package, providing raster imagery of predicted WTD for 24 unique dates with clear distinction in wetness both over different seasons, and spatially in the landscape.

Following this successful application, work is in progress to test the model on additional sites across Scotland and on European level to further test the applicability of the model to a wider range of northern peatlands.

How to cite: Toca, L., Gimona, A., Donaldson-Selby, G., Smart, C., Sideris, K., Ball, J., and Artz, R.: The bigger peat picture: Spatio-temporal modelling of peatland water table depth using Sentinel-1, Sentinel-2, DSM, and field collected data., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19789, https://doi.org/10.5194/egusphere-egu24-19789, 2024.

11:20–11:30
|
EGU24-19664
|
ECS
|
Virtual presentation
Fraser Macfarlane, Ciaran Robb, Margaret McKeen, Matt Aitkenhead, and Keith Matthews

Peat makes up roughly 28% of Scotland’s soil and is critical in many areas, including biodiversity and habitat support, water management, and carbon sequestration. The latter is only possible in healthy, undisturbed peatland habitats where the water table is sufficiently high, otherwise this potential carbon sink becomes a carbon source, that if left untreated will disappear forever. Drainage and erosion features are crucial indicators of peatland condition and are key for estimating national greenhouse gas emissions.
    
Previous work on mapping peat depth and condition in Scotland has provided maps with reasonable accuracy at 100 metre resolution, allowing land managers and policymakers to both plan and manage these soils and to work towards identifying priority peat sites for restoration. However, the spatial variability of the surface condition is much finer than this scale, limiting the ability to inventory greenhouse gas emissions or develop site-specific restoration and management plans.
    
This work involves an updated set of mapping using high-resolution (25 cm) aerial imagery which provides the ability to identify and segment individual drainage channels and erosion features. Combining this imagery with a classical deep learning-based segmentation model, enables high spatial resolution, national scale mapping to be carried out allowing for a deeper understanding of Scotland’s peatland resource and which will enable various future analyses using this data.

How to cite: Macfarlane, F., Robb, C., McKeen, M., Aitkenhead, M., and Matthews, K.: Using Deep Learning and High-Resolution Imagery to Map the Condition of Scotland’s Peatland Resource., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19664, https://doi.org/10.5194/egusphere-egu24-19664, 2024.

11:30–11:40
|
EGU24-5488
|
ECS
|
On-site presentation
Qianqian Han, Yijian Zeng, Yunfei Wang, Fakhereh (Sarah) Alidoost, Francesco Nattino, Yang Liu, and Bob Su

In this study, we generate a global, long-term, hourly terrestrial water-energy-carbon fluxes, using model simulations, in-situ measurements, and physics-informed machine learning. The soil-plant model, STEMMUS-SCOPE was deployed to simulate land surface fluxes over 170 Fluxnet sites (STEMMUS – Simultaneous Transfer of Energy, Mass, and Momentum in Unsaturated Soil; SCOPE - Soil Canopy Observation, Photochemistry and Energy fluxes radiative transfer model). The model input and output data were then used as training data-pairs to develop the STEMMUS-SCOPE emulator using multivariate random forests regression algorithm. Here, physics-informed machine learning refers to the fact that the emulator was trained and constrained by the physical consistency represented by the soil-plant model. We compared the physics-informed emulator (RF_S-S) with the one trained using only Fluxnet in-situ measurements (RF_in-situ), and found that the land surface fluxes predicted by RF_S-S are less scattered than that by RF_in-situ.

We estimate six variables simultaneously: net radiation, latent heat flux, sensible heat flux, gross primary productivity, solar-induced fluorescence in 685 nm and 740 nm. Results show that RF_S-S can estimate fluxes with Pearson Correlation Coefficient score (r-score) higher than 0.97 for these six variables. The testing result using independent stations (not included for developing emulators) show a r-score higher than 0.94. The feature importance shows that incoming shortwave radiation, surface soil moisture, and leaf area index are top predictor variables that determine the prediction performance. We further explore the performance of RF_S-S in predicting soil heat flux, root zone soil moisture, and leaf water potential, which assist the understanding of ecosystems’ drought responses to climate change.

How to cite: Han, Q., Zeng, Y., Wang, Y., Alidoost, F. (., Nattino, F., Liu, Y., and Su, B.: Global long-term hourly 9 km terrestrial water-energy-carbon fluxes with physics-informed machine learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5488, https://doi.org/10.5194/egusphere-egu24-5488, 2024.

11:40–11:50
|
EGU24-4743
|
On-site presentation
|
Paul Stoy, Sadegh Ranjbar, Sophie Hoffman, and Danielle Losos

The terrestrial carbon and water cycles can change rapidly in response to extreme events, human management, meteorological drivers, and more. Many established remote sensing-based estimates of terrestrial ecosystem function are not designed to observe such rapid changes. To address this shortcoming, we developed a machine learning (ML) approach trained on eddy covariance measurements to estimate terrestrial carbon dioxide and water vapor fluxes at five-minute intervals using geostationary (“weather”) satellite observations from the Advanced Baseline Imager on the GOES-R satellite series. Our approach, which we call ALIVE (Advanced Baseline Imager Live Imaging of Vegetated Ecosystems), creates a ‘hypertemporal’ view of terrestrial ecosystem functioning over the diurnal cycle and in near-real time.

We will describe the training and testing procedure used to develop ALIVE with a focus on ML model skill and uncertainty estimation. We then compare ALIVE against the historical MODIS eight-day gross primary productivity (GPP) product to demonstrate how ALIVE can estimate the carbon cycle consequences of land management and extreme events to lead to an improved understanding of the carbon and water cycles. Comparisons during seasonal transitions and in response to extreme events including continental-scale wildfire smoke across North America in 2023 highlight the importance of rapid observations of land surface function.

ALIVE is currently limited to the Conterminous United States (CONUS) and surrounding areas. We discuss the steps necessary to expand it, namely the availability of eddy covariance observations across the Americas and opportunities to apply the ALIVE framework to similar geostationary satellites worldwide like the new Meteosat-12 (formerly MTG-I1) which observes Africa, Europe, and parts of western Asia and is currently scheduled to become operational in Spring 2024. We also discuss opportunities to combine ALIVE estimates with other carbon and water cycle products in a remote sensing data fusion framework to help observe terrestrial ecosystems ‘everywhere, all the time’. By quantifying the carbon and water cycle across all of the time scales over which they vary, we hope to improve the ability of satellite remote sensing to understand our changing planet. 

How to cite: Stoy, P., Ranjbar, S., Hoffman, S., and Losos, D.: Estimating terrestrial carbon dioxide and water vapor fluxes from geostationary satellites in near-real time: the ALIVE framework, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4743, https://doi.org/10.5194/egusphere-egu24-4743, 2024.

11:50–12:00
|
EGU24-4823
|
ECS
|
On-site presentation
Chad Burton, Luigi Renzullo, Sami Rifai, and Albert Van Dijk

We discuss the development of high-resolution (1 km) estimates of terrestrial carbon and water fluxes over the Australian continent by empirical upscaling of a regional network of flux tower measurements (“AusEFlux” v1.1 https://zenodo.org/records/7947265). We detail our ensemble learning approach for estimating the per-pixel epistemic uncertainty in flux predictions.  Our investigations demonstrate that regional or continental upscaling has several advantages over global upscaling, including: the ability to use regionally derived covariable datasets tailored to the regional environmental context; reduced computational constraints allowing for higher-resolution predictions, thus reducing the impacts of sub-cell landscape heterogeneity; ameliorating spatial biases present in global datasets that often have a strong northern hemisphere bias; simpler interpretation of results due the reduced requirement to generalise across vastly different climates, ecosystem types, and plant functional traits; and increased relevance to local stakeholders.  We compare AusEFlux with estimates from nine other products that cover the three broad categories that define current methods for estimating the terrestrial carbon cycle. We argue that consiliences between datasets derived using different methodologies offer alternative value for assessing the quality of an upscaling product than any given cross-validation technique, especially where training datasets have spatial or temporal biases that are difficult to mitigate. Lastly, we discuss the benefits of regularly updating our upscaling product to arrive at a systematic monitoring of terrestrial carbon and water fluxes.

How to cite: Burton, C., Renzullo, L., Rifai, S., and Van Dijk, A.: Empirical upscaling of OzFlux eddy covariance for accurate, high-resolution monitoring of terrestrial carbon and water fluxes in Australia., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4823, https://doi.org/10.5194/egusphere-egu24-4823, 2024.

12:00–12:10
|
EGU24-6067
|
ECS
|
On-site presentation
Laurent Bataille, Ronald Hutjes, Bart Kruijt, Laura van der Poel, Wietse Franssen, Jan Biermann, Wilma Jans, Ruchita Ingle, Anne Rietman, Alexander Buzacott, Quint van Giersbergen, and Reinder Nouta

Peat soil degradation in rural areas contributes to 3% of the annual GHG emissions in the Netherlands. In the 2019 climate agreement, the Dutch government set a goal to cut these emissions by 25% by 2030 and initiated a research consortium to achieve this, the Dutch National Research Programme on Greenhouse Gases in Peatlands (NOBV). The NOBV established a GHG monitoring network to map emissions based on the diversity of peat, edaphic conditions, grassland management, and water table management.

Eddy-Covariance plays an essential role in this monitoring network. More than 20 sites are part of it, including permanent and mobile EC towers, while an intensive airborne measurement campaign co-occurs above the studied areas. The first offers a high-temporal resolution monitoring of flux, covering a limited spatial landscape diversity; the latter provides a comparative map embracing the whole heterogeneity of the landscape during short timeslots.

A data-driven bottom-up model will be implemented. This model will focus on considering site-specific factors and characterizing the heterogeneities of the surroundings through footprint analysis. This approach is a progression from previous efforts that compared annual carbon budgets across various locations based on external data sources, which combined remote sensing and hydrological models and implemented a machine-learning framework to gap-fill time series.

Validation of the model includes a comparison with the airborne datasets. Beyond evaluating the quality of the model, the objective is to assess the strengths and limitations of this bottom-up approach when compared to real, independent datasets describing accurately spatial fluctuation of fluxes in heterogeneous landscapes.

The insights are essential for future developments, including models that leverage ground network and airborne measurements in the training process for more robust and accurate CO2 Fluxes modelling.

How to cite: Bataille, L., Hutjes, R., Kruijt, B., van der Poel, L., Franssen, W., Biermann, J., Jans, W., Ingle, R., Rietman, A., Buzacott, A., van Giersbergen, Q., and Nouta, R.: Data-Driven Modelling and Comparative Analysis of CO2 Exchanges in Dutch Peatlands via Eddy-Covariance: Ground-calibrated bottom-up model vs airborne flux measurements, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6067, https://doi.org/10.5194/egusphere-egu24-6067, 2024.

12:10–12:20
|
EGU24-19181
|
ECS
|
On-site presentation
Egor Prikaziuk, Christiaan Van der Tol, and Enrico Tomelleri

The Eddy Covariance (EC) method is a state-of-the-art ecosystem flux measurement technique. EC data is used to calibrate and validate biogeochemical models for downstream Earth Observation (EO) products. Despite being a point measurement, the EC method samples an area around the measurement tower, called a flux footprint (FFP). The extent and direction of an FFP depend on the wind speed and direction and the state of the atmosphere, thus it changes continuously. Although most of the EC towers are placed in homogeneous areas, where the footprint direction should not play a role, at the modern 10-m spatial resolution of EO satellites such as Sentinel-2, the area under FFP may be seen as heterogeneous because of phenological differences or temporal changes in vegetation cover. The land cover heterogeneity under the FFP may cause challenges in interpreting the EC data and discrepancies in the calibration and validation of EO downstream products.

In this study, we investigated the FFP heterogeneity of 72 European EC sites from the ICOS Warm Winter 2020 dataset (https://doi.org/10.18160/2G60-ZHAK). Firstly, the footprint size was statistically estimated according to the dominant plant functional type. Secondly, the heterogeneity was assessed for a circular buffer and twelve sub-sectors of the buffer on Sentinel-2-derived normalised difference vegetation index (NDVI) with several statistical criteria (e.g. Tukey’s test, Z-score). This approach was demonstrated to be an alternative to precise FFP models, as the latter might require parameters not available to the end users, such as standard deviation of the v-wind component or atmospheric boundary layer height. The tool was developed as a stand-alone application and can be used for spatial heterogeneity assessment beyond the tested EC footprints. 

How to cite: Prikaziuk, E., Van der Tol, C., and Tomelleri, E.: A rapid approach for integrating high-resolution remote sensing and eddy flux observations, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19181, https://doi.org/10.5194/egusphere-egu24-19181, 2024.

12:20–12:30
|
EGU24-13485
|
ECS
|
On-site presentation
Wei He, Chengcheng Huang, Jinxiu Liu, Ngoc Tu Nguyen, Hua Yang, and Mengyao Zhao

Estimation of regional-scale carbon budget still faces considerable uncertainties, whereby reconciling bottom-up estimates and top-down inversions are essential steps towards reducing such uncertainties. Here,  we make full use of available flux tower observations and novel satellite land surface observations (e.g., soil moisture and solar-induced chlorophyll fluorescence) to constrain the spatiotemporal regimes of net ecosystem carbon exchange (NEE) for North America based on the advanced Long Short-term Memory (LSTM) networks. The LSTM-based model is capable of incorporating the memory effects of climate and environments on NEE variations, thus more accurately simulating the interannual variations (IAVs) and long-term trends of NEE. We produced gridded NEE at a spatial resolution of 0.1° × 0.1°and monthly time-step over 2001–2021 for North America.

The annual total NEE during 2001–2021 is -1.74 ± 0.10 Pg C yr-1, which is much closer to the top-down inversions (-0.73 and -0.63 Pg C yr-1 for CarbonTracker2022 over the same period and the Orbiting Carbon Observatory-2 (OCO-2) v10 MIP ensemble mean over 2015–2020 respectively) than the global flux upscaling products (-3.04 and -2.75 Pg C yr-1 by FLUXCOM2020 RS+CRUJRA1.1 and RS+ERA5 respectively and -3.30 Pg C yr-1 by NIES2020). Moreover, the spatial patterns matched with the OCO-2 v10 MIP ensemble mean reasonably well, which indicated the largest carbon uptake in the Midwest Corn-Belt area during peak growing seasons and in the Southeast on an annual basis. The estimated IAV of NEE is highly correlated with that by CarbonTracker2022. The LSTM estimate captured the NEE anomalies caused by the droughts in 2011, 2012, 2017, 2020–2021, and the 2019 Midwest floods. The performances are clearly better than existing global flux upscaling products. In addition, the estimated annual NEE exhibited a significant decline at the rate of -0.008 Pg C yr-2 (p < 0.05), indicating an overall enhanced carbon sink during the recent two decades, which is in line with contemporary estimates.

These results suggest that the LSTM-based NEE upscaling provides an improved estimation of North American NEE for both spatial and temporal characteristics, narrowing down the gap between bottom-up estimates and top-down inversions, which is an important step towards robust regional carbon budget estimations.

How to cite: He, W., Huang, C., Liu, J., Nguyen, N. T., Yang, H., and Zhao, M.: Improved estimation of net ecosystem CO2 exchange for North America in eddy flux upscaling with memory-based deep learning , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13485, https://doi.org/10.5194/egusphere-egu24-13485, 2024.

Posters on site: Thu, 18 Apr, 16:15–18:00 | Hall X1

Display time: Thu, 18 Apr 14:00–Thu, 18 Apr 18:00
Chairpersons: Madlene Nussbaum, Hanna Meyer, Benjamin Dechant
X1.103
|
EGU24-4638
|
ECS
Diagnosis of Nitrogen Nutrition in Summer Maize based on Simulated Multispectral Data
(withdrawn)
lu liu
X1.104
|
EGU24-5476
|
ECS
Benedikt Hiebl, Gianmaria Bonari, Nicola Alessi, Alessandro Bricca, Giacomo Calvia, Georg Wohlfahrt, Clemens Geitner, and Martin Rutzinger

Climate change-induced shifts, such as prolonged growing seasons and milder winters, coupled with land-use alterations like forest management abandonment, are reshaping species composition across European forests. A significant spread of evergreen broad-leaved species (EVEs) was observed in southern European forests, driven by global change dynamics. However, large-scale spatio-temporal analysis of these changes are lacking and emphasizes the necessity of mapping these dynamics. The TRACEVE projects (“Tracing the evergreen broad-leaved species and their spread”) primary goal is therefore tracking  EVEs' cover, spread and diversity on a national-scale in Italy. As part of the project, this study focuses on satellite remote sensing, Deep Learning, and forest mapping, aiming at creating seamless maps quantifying the current degree of EVEs cover within forests in Italy.

Challenges arise in the transitional zones between evergreen and deciduous forests, where EVEs initially spread in the understory of a deciduous canopy. Tracking EVEs at the edge of their range, where abundance is rare and in mixed forests is therefore difficult. Leveraging Sentinel-2 remote sensing time series covering the full annual phenological cycle addresses this challenge, utilizing leaf-on and leaf-off canopy conditions. Values of species cover derived from ad-hoc forest plot observations in Italian protected areas across a latitudinal gradient serve as initial training data, although the small sampling size (~1000 plots) poses challenges for the generalizability of time series extrinsic regression models, particularly when employing state-of-the-art Deep Learning architectures.

The main aim of the study is therefore the development of a robust, Sentinel-2 based mapping procedure to track EVEs within Italian forests on a national-scale. A remote sensing time series extrinsic regression model based on a Deep Learning architecture for cover degree mapping will be developed. Sentinel-2 annual time series, along with derived indices serve as predictors for the models, while target cover will be derived from the plot observations. The study is built around exploring selected strategies to address the issue of large-scale model generalizability, given a small training sample size, that is recorded within small representative areas scattered across Italy. This entails assessing the efficacy of self-supervised pretraining methodologies applied to remote sensing time series within forested regions, pretraining on a more extensive forest database, and evaluating the viability of training data augmentation techniques.

To validate and evaluate results on a national scale, an independent forest vegetation database containing around 17,000 forest plots sampled across Italy is employed. This extensive dataset enhances the understanding of EVEs' distribution within Italy's diverse forest ecosystems and can further enhance understanding of complex model results.

In conclusion, the study combines advanced satellite remote sensing technologies, Deep Learning methodologies coupled with vegetation plot datasets to map current distribution of EVEs in Italian forests. The findings contribute to the TRACEVE project's future objectives but also offer insights into the challenges and opportunities of Deep Learning models in large-scale forest mapping applications.

Acknowledgements

This research has been conducted within the project “TRACEVE - Tracing the evergreen broad-leaved species and their spread” (I 6452-B) funded by the Austrian Science Fund (FWF).

How to cite: Hiebl, B., Bonari, G., Alessi, N., Bricca, A., Calvia, G., Wohlfahrt, G., Geitner, C., and Rutzinger, M.: Mapping evergreen broad-leaved species spatial cover in Italian forests from Sentinel-2 time series using Deep Learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5476, https://doi.org/10.5194/egusphere-egu24-5476, 2024.

X1.105
|
EGU24-7304
Tomoko Akitsu, Hiroshi Murakami, Atsushi Kume, Hideki Kobayashi, and Roxanne Lai

For long-term vegetation monitoring, satellite-derived vegetation indices (VI) are important in environmental science and, thus, have been generally used. Toward the contribution of long-term monitoring, the Global Change Observation Mission-Climate (GCOM-C) satellite, launched in 2017, provides the VI of a 250 m spatial resolution. This study aims to inform the characteristics of VI derived from the second-generation global imager (SGLI) on GCOM-C from the viewpoint of leaf optical properties and to propose a new VI sensitive to large chlorophyll content. To obtain a leaf-level VI, we measured the leaf reflectance of 25 species (including major plant functional types (PFT): evergreen needle and broad leaves; deciduous needle and broad leaves) using an integrating sphere and a spectral radiometer in summer and autumn. Each ‘leaf-level satellite VI’ was calculated using leaf reflectance and relative spectral response curves of satellite bands.

As for the normalized difference vegetation index (NDVI), NDVI derived from SGLI and Moderate Resolution Imaging Spectroradiometer (MODIS) was similar to each other but has a discrepancy in soil moisture dependency and absolute value[1]. Therefore, we obtained the leaf-level correction factor between SGLI and MODIS, supposing their concatenated use in long-term research. The leaf-level correlation between SGLI NDVI and MODIS NDVI had no dependence on PFT and seasons.

As for the photochemical reflectance index (PRI), we calculated the following ‘leaf-level satellite PRI’: SGLI PRI (using bands 5 and 6), MODIS PRI (bands 11 and 12), MODIS PRI (bands 11 and 1). Each was compared with the ‘leaf-level definition PRI’ calculated from the leaf reflectance at 531 nm and 570 nm wavelengths. As a result, the SGLI PRI (bands 5 and 6) showed the largest R2 with the definition PRI in summer and throughout seasons (R2=0.94 and R2=0.97, respectively). The MODIS PRI (bands 11 and 1) was better correlated with the definition PRI than MODIS PRI (bands 11 and 12) throughout seasons (R2=0.81 and R2=0.43, respectively) but depended on PFT with no correlation in summer (R2=0.0049).

As for the chlorophyll index (CI), the CI using red edges is popular. However, the satellite-based time series of CI using SGLI bands 8 and 9 adjacent to the red edge (673.5 nm and 763 nm, respectively) saturated in small chlorophyll content in early summer. Thus, we investigated which SGLI bands are sensitive to large chlorophyll content using the SCOPE2.0 model[2][3], which combines radiative transfer in plant leaves, canopies, and soil with photosynthesis. In this study, we propose a new SGLI CI sensitive to large chlorophyll content.

 

References

[1] Bayarsaikhaan U. et al. (2022) Early validation study of the photochemical reflectance index (PRI) and the normalized difference vegetation index (NDVI) derived from the GCOM-C satellite in Mongolian grasslands, Int. J. Remote Sens., 43:14, 5145-5172.

[2] Yang P. et al. (2021) SCOPE 2.0: a model to simulate vegetated land surface fluxes and satellite signals, Geosci. Model Dev., 14, 4697–4712.

[3] Van der Tol, C. et al. (2009) An Integrated Model of Soil-Canopy Spectral Radiances, Photosynthesis, Fluorescence, Temperature and Energy Balance, Biogeosciences, 6 (12): 3109–29.

How to cite: Akitsu, T., Murakami, H., Kume, A., Kobayashi, H., and Lai, R.: The GCOM-C/SGLI vegetation indices (NDVI, PRI, Chlorophyll Index) from the viewpoint of leaf optical property, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7304, https://doi.org/10.5194/egusphere-egu24-7304, 2024.

X1.106
|
EGU24-8869
|
ECS
Sélène Ledain, Felix Stumpf, and Helge Aasen

In land surface monitoring, parameters such as the Leaf Area Index (LAI) have been widely studied to describe the canopy structure, foliage cover, crop yield and growth. Accurate and up to date mapping of such biophysical variables at global scale is thus essential for decision makers and agricultural management, as it can help monitor growth conditions and adapt practices. The retrieval of LAI from remotely sensed data is commonly done through the inversion of a combined leaf and canopy radiative transfer model (RTM). These models relate leaf biochemistry and canopy structure to spectral variation. However, the inversion process is computationally expensive and the ill-posed nature of the problem does not ensure a unique LAI retrieval solution. 

The aim of this research is to accelerate LAI retrieval by exploiting machine learning regression algorithms and thus allow more widespread crop monitoring from remote sensing data. We propose a more operational method for large-scale retrieval by emulating the inversion problem with a machine learning algorithm. These models effectively capture non-linearities, proving particularly pertinent in the context of land surface modeling. Our work focuses on the agricultural land in Switzerland, monitored through Sentinel-2 imagery between 2017 and 2023. We use the PROSAIL RTM to simulate the spectral response of Sentinel-2 to various combinations of biophysical and canopy variables, including LAI. A look-up table (LUT) relating leaf and canopy parameters to reflectance spectra is thus generated and used to train the machine learning algorithm. Methods among Random Forest, Neural Network and Gaussian Process Regression are tested and the best model is selected according to the root mean squared error (RMSE) on in-situ validation measurements of LAI of several fields in Switzerland.

To further improve LAI retrieval, we include phenological a priori knowledge to constrain the underdetermined problem. Specifically, we limit LAI values knowing the general relation of crop development with temperature. Accumulated Growing Degree Days (GDD) represent the cumulative temperature above the base temperature of a plant, where no growth occurs. We thus use GDDs to determine phenological macro-stages and constrain LAI values using physically sound assumptions. With the incorporation of phenological constraints, we additionally enable the customization of the model to suit specific crops or seasons, offering a simple solution to increase model performance according to stakeholder needs.

How to cite: Ledain, S., Stumpf, F., and Aasen, H.: Improval of radiative transfer model-based LAI retrieval from Sentinel-2 data through machine learning and phenological constraints, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8869, https://doi.org/10.5194/egusphere-egu24-8869, 2024.

X1.107
|
EGU24-10962
|
ECS
Shaopeng Li and Stefan Wunderle

Terrestrial vegetation is a key component of the biosphere, which regulates global carbon, water and energy cycles. Meanwhile, global environmental change is rapidly altering the dynamics of terrestrial vegetation by functioning the Earth system and providing ecosystem services. Therefore, systematically monitoring global vegetation dynamics is critical to understand the basic biogeochemical processes, and their possible feedbacks to the climate system. Besides, characterizing spatial heterogeneity and temporal variation of surface albedo is also great importance to monitor land cover change and to determine energy exchange between ground and atmosphere. 
Remote sensing offers the only effective method for measuring and monitoring the vegetation dynamics, heterogeneity of albedo and its directional signature on regional and global scales. In this study, 4 km × 4 km GAC AVHRR data from extensive satellite data archive collected by the Remote Sensing Research Group at the University of Bern (RSGB) from 1981 to 2022 was applied to analyze albedo coupling with vegetation dynamics over global. The Simplified Method for Atmospheric Correction (SMAC) radiative transfer code was employed to do the atmospheric correction, and Ross-Thick/Li-Sparse-Reciprocal (RTLSR) kernel-driven model was applied using all contemporaneous NOAA and MetOp satellites data for removing anisotropic effects by means of the Bidirectional Reflectance Distribution Function (BRDF). Therefore, it is expected that the new generation of global Normalized Difference Vegetation Index (NDVI) and Albedo products could be produced from 1981 until 2022, which could potentially provide a long-term dataset for global climate change studies. Some preliminary evaluations showed that the new generation of NDVI product follows a generally similar trend to other products such as SPOT VGT, AVH13C1 NDVI, and Terra MODIS NDVI on selected validated sites. 

How to cite: Li, S. and Wunderle, S.: Global vegetation dynamics using multiple satellite observations throughout the past four decades, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10962, https://doi.org/10.5194/egusphere-egu24-10962, 2024.

X1.108
|
EGU24-19656
|
Joshua Castro, Catriona L. Fyffe, Vinisha Varghese, Evan Miles, Martin Hoelzle, and Francesca Pellicciotti

The highland ecosystems in the southern Peruvian Andes represent an important resource for people who depend on them for water regulation, agriculture, and energy generation. These ecosystems change over time due to natural causes, climate variations, or anthropic intervention. Remote sensing studies have quantified land cover changes in this domain annually, but have neglected the seasonal ecosystem variations, which are conditioned by seasonal weather patterns, species phenology, or applied management. One high-elevation headwater in this region, the Vilcanota-Urubamba Basin (VUB) in Cusco-Peru is characterized by glaciers surrounded by barren areas, flooded areas such as lakes and wetlands, and vegetation ranging from sparse grasslands to forests in the northwest of the catchment.

In this study, we used a land cover classification model at a multi-seasonal scale over 10 years to observe the seasonal dynamics and links between the land cover classes in VUB. We applied the Random Forest classification model on Landsat 7 and 8 imagery (30 m/pixel), trained with selecting 10 points for seven land cover classes, to discriminate land cover every 3 months from 2013 to 2022 over an extended area of the VUB system. We ran the model in the Google Earth Engine platform, using as inputs six spectral indices and three topographic indices obtained from an SRTM DEM.

Our overall results indicate that the VUB area (11,047 km2) is mainly occupied by Agriculture and Pasture (~52.5%) distributed in the upper-middle areas above 3000 masl, followed by Barren (~18.9%). Shrub (~8.4%) and Forest (~8.25%) areas which are more concentrated in the northwest region, and Wetland (~3.94%), Water (~2%), and Snow and Ice (~1.88%) areas, which are mostly located in the Southeast region. We apply two general accuracy assessments based on a set of collected validation points (0.74) and random points (0.72). The results show limited availability of high-quality images during January to March of almost every year, related to the high cloudiness, but much better image availability in the other months.

We find strong seasonal variations in the Snow/Ice, Water, and Wetland classes related to the precipitation regime of the region; but Barren, Shrub, and Forest areas do not vary much seasonally. We determined the correlation between land cover classes for each season and considering all seasons together to identify relevant interrelationships between classes. We find that Snow/Ice changes are correlated (p<0.05) to the Wetland areas (r=0.35) and Water bodies (r=0.71) which are also related to each other (r=0.49). The Agriculture and Pasture areas change with Barren areas (r=-0.65) but have a slight inverse correlation to Wetlands (r=-0.27) highlighting the importance of the seasonal climate. We can interpret these results to infer the dependence between highland vegetated ecosystems and the seasonal hydrological response to glaciers and weather patterns. Overall this work provides important insights into the seasonal landcover change dynamics in this region and the important interrelationships between components of Peruvian highland ecosystems.

How to cite: Castro, J., Fyffe, C. L., Varghese, V., Miles, E., Hoelzle, M., and Pellicciotti, F.: Multi-seasonal land cover changes of South Peruvian Highland Andean ecosystems, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19656, https://doi.org/10.5194/egusphere-egu24-19656, 2024.

X1.109
|
EGU24-20836
|
ECS
|
Jorge Rodríguez, Kasper Johansen, Hua Cheng, Areej Alwahas, Victor Angulo-Morales, Samer Almashharawi, and Matthew F. McCabe

As global biodiversity faces increasing threats from climate change, habitat loss, and human activities, effective methods for assessing and monitoring biodiversity are crucial. Drylands are particularly vulnerable to the impacts of climate change, and integrating remote sensing technology with ecological research can help protect these environments. Identification of individual trees and shrubs is fundamental to assess biodiversity and can also improve carbon stocks estimates. However, identifying individual trees using medium resolution satellite images is often not feasible. The use of advanced technologies, such as machine learning and satellite imagery in environmental management plays a key role in biodiversity conservation and can potentially fill this gap. The use of readily available Maxar satellite imagery makes conservation approaches accessible and cost-effective, which is crucial for widespread adoption, especially in resource-limited settings or for large-scale studies. This study aims to improve the identification of vegetation in dryland ecosystems by integrating deep learning methods to remote sensing. The primary objective was to distinguish vegetation from rocks or shadows in these areas, which is often challenging due to the dark appearance of vegetation and the similar visual features of the landscape, such as landforms textures, water bodies or man-made objects.. To address this challenge, a Vision Transformer (ViT) deep learning model was developed to estimate near infrared (NIR) spectral bands from high resolution Maxar Satellite images. By enhancing the spectral richness, the model aids in the differentiation of vegetation. Maxar satellite imagery was primarily used because of its accessibility through Google services, making it ideal for planning initial surveys to identify areas of interest for more detailed study. The accuracy of the model was validated against high-resolution SkySat NIR imagery, and achieved an R2 of 0.92. The obtained NIR band helped to clearly distinguish vegetation from non-vegetative surfaces such as soil, rocks, and water, which were not as discernible in RGB imagery alone. The use of a deep learning method to estimate a synthetic NIR band proved to be cost-effective and efficient for large-scale identification of individual trees and shrubs in drylands, overcoming the limitations of medium-resolution satellite imagery. The findings of the study are crucial to the conservation of biodiversity and offer a practical approach for environmentalists and researchers. Future work includes expanding the dataset to include various dryland environments, and integrating additional data sources such as soil data and topographic features for a more comprehensive analysis.

How to cite: Rodríguez, J., Johansen, K., Cheng, H., Alwahas, A., Angulo-Morales, V., Almashharawi, S., and McCabe, M. F.: Biodiversity Assessment in Drylands Using Augmented Satellite Imagery Through Deep Learning Models, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20836, https://doi.org/10.5194/egusphere-egu24-20836, 2024.

X1.110
|
EGU24-6928
|
ECS
Carlos Alberto Silva, Caio Hamamura, and Carine Klauberg

The Southern U.S. hosts some of the most productive forests globally, playing a crucial role in the U.S. terrestrial carbon sink and contributing significantly to timber production (over 60% in commercial terms). Despite their importance, these forests face frequent hurricanes, leading to substantial harm to the forest structure and related ecosystem functions. This damage extends to the loss of timber supply, heightened wildfire risk, and diminished recreational opportunities. With an increasing frequency of hurricanes in the region, accurately gauging the harm inflicted on these forests becomes imperative for formulating effective protective measures and comprehending the dynamics of forest recovery.

To address this issue, the study aimed to create a data fusion framework utilizing NASA's ICESat-2 and Landsat 8 OLI for mapping large-scale canopy height. This mapping would then be employed to assess the severity of post-hurricane disturbance damage and monitor forest recovery. The research utilized ICESat-2-derived canopy height estimates to calibrate a Random Forest model, predicting and mapping canopy height both before and after Hurricane Michael. The findings revealed that a combination of spectral bands and vegetation indices from Landsat 8 explained a significant portion of the canopy height variation. In areas heavily affected by Hurricane Michael in 2018, the average canopy height dropped from approximately 18 meters to 12 meters in 2019, experienced a slight increase to around 12.5 meters in 2021, and reached about 13 meters in 2022. Despite three years post-Hurricane Michael, the canopy height did not fully recover to pre-disturbance conditions.

This research introduces an innovative approach to enhance forest structure mapping by integrating ICESat-2 and Landsat 8 data streams. The advancement in data fusion methodology provides an opportunity for more precise and detailed assessments of the impacts of natural disasters, such as hurricanes, on forest ecosystems in the Southern U.S

How to cite: Silva, C. A., Hamamura, C., and Klauberg, C.: Machine Learning-driven Fusion of NASA’s ICESat-2 and Landsat 8 OLI Data for Assessing Forest Recovery Following Hurricane Disturbance in Southern Forests, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6928, https://doi.org/10.5194/egusphere-egu24-6928, 2024.

X1.111
|
EGU24-13734
|
ECS
Jie Li and Kun Jia

Involving the effect of atmospheric CO2 fertilization is effective for improving gross primary production (GPP) estimation accuracy using light use efficiency (LUE) model. However, the widely used LUE model, the remote sensing-driven Carnegie-Ames-Stanford Approach (CASA) model, is scarcely considering the effects of atmospheric CO2 fertilization, which cause GPP estimation uncertainties. Therefore, this study proposed an improved GPP estimation method based on CASA model integrating the atmospheric CO2 concentration and generated a long time series GPP dataset with high precision for the Tibetan Plateau. The CASA model was improved by considering the atmospheric CO2 effect on vegetation productivity and distinguishing the CO2 gradients differences within the canopy and leaves brought by the influence of leaf stomatal conductance and leaf saprophyte activity. A 500m monthly GPP dataset for the Tibetan Plateau from 2003 to 2020 were generated. The results showed that the improved GPP estimation model achieved better performances on estimating GPP (R2 = 0.68, RMSE = 406 g C/m2/year) than the CASA model (R2 = 0.67, RMSE = 499.32 g C/m2/year), and MODIS GPP products. The GPP on Qinghai-Tibet Plateau increased significantly with the increase of atmospheric CO2 concentration and the gradual accumulation of dry matter. The improved GPP estimation method can also be used for other regions and the generated GPP dataset is valuable for further understanding the ecosystem carbon cycles on Qinghai-Tibet Plateau.

How to cite: Li, J. and Jia, K.: Improving gross primary production estimation accuracy on the Qinghai-Tibet Plateau considering the effect of atmospheric CO2 fertilization, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13734, https://doi.org/10.5194/egusphere-egu24-13734, 2024.

X1.112
|
EGU24-14253
Sheng Wang, Boqin Yuan, Kaiyu Guan, Jørgen Eivind Olesen, Rui Zhou, and René Gislum

Conservation tillage practices with crop residues covering soils can improve soil health to increase agronomic and environmental benefits for croplands. Accurate information of field-level crop residue cover is highly important for evaluating the implementation of government conservation programs and voluntary ecosystem service markets, as well as supporting agroecosystem modeling to quantify cropland biogeochemical processes. Remote sensing has been demonstrated to detect crop residue cover cost-effectively, yet existing regional-scale studies in Europe are rare. To fill this data gap, our study developed an explainable machine learning algorithm to integrate multi-source satellite data (Sentinel-2, Sentinel-1, and SMAP soil moisture) to quantify crop residue presence for the EU croplands. Specifically, we utilized satellite time series data of optical spectral tillage index, soil background reflectance, soil moisture, and SAR backscattering information to detect field-level crop residue cover. With 41,325 ground records of 10 major crops, we developed highly robust and explainable machine learning models with unbalanced label correction approaches to predict residue presence. Results show that models achieved high accuracy of 0.78 and F1-score of 0.70 to detect crop residue presence. We also aggregated field-level estimates to the regional level, which shows high match with regional census data. Among crop types, wheat and barley got higher prediction performance than other crop types. Our work highlights the feasibility of integrating multi-source satellite data with machine learning for detecting field-level crop residue cover at continental scale across the EU. These crop residue datasets can support analyzing the spatiotemporal variability of tillage practices across the EU and their potential impact on agroecosystem productivity and sustainability. 

How to cite: Wang, S., Yuan, B., Guan, K., Olesen, J. E., Zhou, R., and Gislum, R.: Detecting Field-level Crop Residue Cover across the EU Using Multi-source Satellite Data and Explainable Machine Learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14253, https://doi.org/10.5194/egusphere-egu24-14253, 2024.

X1.113
|
EGU24-19458
Kun Shang, Chenchao Xiao, and Hongzhao Tang

Soil organic matter (SOM) and particle size are key indicators for evaluating cultivated soil quality. Conventional soil quality surveys based on field sampling are resource-intensive and can only obtain the data at sampling points, making it difficult to meet the needs of plot-level cultivated land management. In recent years, a series of hyperspectral satellite sensors have provided an important data source for the estimation of cultivated soil parameters. In this study, we took all 1.26 million km2 of cultivated land (including paddy fields, dry land, and irrigated fields) in Northeast China as the study area. In April 2021, we conducted a synchronized sample collection experiment utilizing ZY1-02D satellite hyperspectral data, gathering a total of 171 soil samples. More than 1,400 hyperspectral images of satellites including GF5, ZY1-02D and ZY1-02E covering the entire study area were collected and preprocessed. Firstly, we developed a bare soil identification method by combining cultivated soil spectral library, spatial-spectral filtering, and spectral angle mapping. The average accuracy of bare soil identification results varies from 90% to 95%. Secondly, we analyzed the correlation between soil parameters and dual-band spectral indices using multi-platform observed data, as well as the radiation quality of massive satellite images. Combining the results of spectral band radiation quality analysis, the optimal spectral indices of SOM, sand, silt, and clay were constructed based on the collaborative observation of multi-source data. Then, we developed a soil parameter prediction model that combines topography and spectral information. In this research, a new feature selection technique called VIP-CARS-Frog based on multi-index evaluation, which combines three algorithms including variable importance plots, competitive self-organizing selection, and random frog, was proposed to select high-quality and stable features. These technologies have been successfully applied to hyperspectral satellite data such as GF5, ZY1-02D, and ZY1-02E to map the spatial distribution of SOM and particle size in cultivated land in Northeast China. The inversion results of SOM, sand, silt, and clay have R2 of 0.84, 0.9, 0.79 and 0.76, RMSE of 5.16g/kg, 7.16%, 7.25% and 4.7%, and RPD of 2.32, 2.87, 1.72, and 1.73, respectively. From the results, it can be seen that in Songnen Plain and Sanjiang Plain, where black soil and chernozem are concentrated, the SOM content is higher and the sand content is lower. On the contrary, in the southwestern region, the sand content is higher and the SOM content is lower. The results indicate that hyperspectral satellite images can be used to estimate SOM and particle size content at a regional scale, showing its great potential in cultivated land quality surveys and agricultural precision management.

How to cite: Shang, K., Xiao, C., and Tang, H.: Regional mapping of soil organic matter and particle size in Northeast China using hyperspectral satellite images, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19458, https://doi.org/10.5194/egusphere-egu24-19458, 2024.

X1.114
|
EGU24-16810
Stéphane Jacquemoud, Dung Tri Nguyen, Antoine Lucas, Sylvain Douté, Cécile Ferrari, Sophie Coustance, Sébastien Marcq, and Aymé Meygret

Numerous research projects have successfully exploited remote sensing data to analyze Earth and planetary surfaces. The use of radiative transfer models simulating the interaction between electromagnetic radiation and bare soils, such as the Hapke model, is becoming increasingly widespread. However, the in- version of these models is relatively uncommon due to a number of difficulties. To address these issues, our team has collected relevant field and satellite data from the Asal-Ghoubbet rift (Republic of Djibouti) and developed a comprehensive framework for analyzing these data. This site was chosen for the diversity of its terrains, characterized by varied albedo and surface roughness, which are well preserved due to the desert climate. It also has the advantage of being easily accessible (Labarre et al., 2019).

To carry out our study, we used images from the Pleiades-1B satellite captured in video mode over the Asal-Ghoubbet rift on January 26, 2013, during the in-flight commissioning of the satellite. This unique four-minute flyby produced 21 images at viewing angles ranging from from -56.7° to +52.6°. The images were corrected for atmospheric effects, which modify the photometric response of surfaces.  To achieve this, experts from CNES applied a variant of the MACCS ATCOR Joint Algorithm (MAJA), using auxiliary data to take into account the water vapor content and aerosol optical thickness (Hagolle et al., 2015). In addition, to validate the results of the Hapke model inversion, a field experiment was conducted in February 2016 in the Asal-Ghoubbet rift to collect soil samples and acquire data (ground truth).

To meet the challenge of limited geometric observation configurations, essential for constraining model parameters, our global approach tackled the known coupling effect between parameters. We also had to take into account the prohibitive computation time required to process millions of pixels over multiple spectral bands, which was a major obstacle to generating the Hapke parameter map. We applied the fast Bayesian inversion method developed by Kugler et al. (2022), which offers an efficient solution to overcome this problem. In parallel, a geometrical correction was applied using a previously constructed digital elevation model (DEM) of the rift that used the same data set. In the end, with each spectral band, we obtained four maps of Hapke model parameters corresponding to the single scattering albedo w, the photometric roughness θ, the asymmetry b and backscattering c parameters of the phase function. The areas of low reconstruction error (less than 1.5%) represents 70% of the entire region. The remainder can be attributed to areas with extremely steep slopes and heterogeneous terrains along slopes such as mass wasting deposits, or areas hindered by clouds and their associated shadows on the ground. The correlation between the parameters and the geological map, the analysis of the soil samples of each terrain units will be presented and discussed.

How to cite: Jacquemoud, S., Nguyen, D. T., Lucas, A., Douté, S., Ferrari, C., Coustance, S., Marcq, S., and Meygret, A.: Photometric characterization of the Asal-Ghoubbet rift (Republic of Djibouti) by massive inversion of the Hapke model, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16810, https://doi.org/10.5194/egusphere-egu24-16810, 2024.

Posters virtual: Thu, 18 Apr, 14:00–15:45 | vHall X1

Display time: Thu, 18 Apr 08:30–Thu, 18 Apr 18:00
Chairpersons: Alvaro Moreno, Jacob Nelson, Hanna Meyer
vX1.11
|
EGU24-9374
|
ECS
Upscaling terrestrial Evapotranspiration: A framework based on a spatial heterogeneitymodel and machine learning algorithms
(withdrawn)
Yixiao Zhang and Tao He
vX1.12
|
EGU24-16460
|
ECS
Mukund Narayanan, Ankit Sharma, and Idhayachandhiran Ilampooranan

Traditional approaches to mapping rice areas depend on detailed field surveys to gather label data. This data gathering, while thorough, is often time-consuming, and costly. Further, for long-term rice area mapping, especially at large scales, ground validation can only be made in recent years, which may lead to uncertainties in rice areas of the past. To solve this limitation, this study introduces a novel method for classifying rice cultivation areas without the need for field data by employing a self-supervised learning framework within Google Earth Engine. Using MODIS data to calculate indices such as the Normalized Difference Vegetation Index (NDVI), the Enhance Vegetation Index (EVI), and the Land and Surface Water Index (LSWI), a pseudo-label boundary was delineated. The delineation of the boundary involved marking flooded regions, where LSWI + 0.05 ≥ NDVI EVI. Within these flooded regions, instances when EVI in the first 40 days was at least half of the peak EVI during the growing season were assigned as rice. In contrast, regions that did not meet these criteria were considered for non-rice. As proof of concept, the delineated pseudo-label boundaries of rice and non-rice were randomly sampled and trained on several machine learning models like random forests, support vector machines, gradient-boosted trees, and decision trees, to classify rice areas in Punjab, India, from 2003 to 2022. The random forest model demonstrated superior performance, achieving an Area Under the Curve of receiver-operating characteristics (AUC) of 0.71, compared to other models (AUC of ~0.55). Furthermore, comparing the self-supervised models against the same machine learning models, which were traditionally trained on field survey data (228 ground points: 164 rice, 64 non-rice was collected), the self-supervised models showed ~10% higher performance than their traditionally supervised counterparts. Therefore, this study demonstrates that using this self-supervised modeling framework reduces the need for field-based annotations, while still providing reasonably accurate rice area maps.

How to cite: Narayanan, M., Sharma, A., and Ilampooranan, I.: Self-supervised learning for mapping rice areas reduces the need for field surveys, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16460, https://doi.org/10.5194/egusphere-egu24-16460, 2024.