Explainable AI for earth observation: A review including societal and regulatory perspectives

Gevaert, C.M.

doi:10.1016/j.jag.2022.102869

Cited by 45 publications

(23 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For stakeholders, users, application and decision making, digital twins need to be transparent and explainable [ 13 , 31 , 32 ]. Addressing this problem, a deeper understanding of the origin of the anomalies is important.…”

Section: Resultsmentioning

confidence: 99%

Towards Digital Twins of the Oceans: The Potential of Machine Learning for Monitoring the Impacts of Offshore Wind Farms on Marine Environments

Schneider

Klüner

2023

Sensors

View full text Add to dashboard Cite

With an increasing number of offshore wind farms, monitoring and evaluating the effects of the wind turbines on the marine environment have become important tasks. Here we conducted a feasibility study with the focus on monitoring these effects by utilizing different machine learning methods. A multi-source dataset for a study site in the North Sea is created by combining satellite data, local in situ data and a hydrodynamic model. The machine learning algorithm DTWkNN, which is based on dynamic time warping and k-nearest neighbor, is used for multivariate time series data imputation. Subsequently, unsupervised anomaly detection is performed to identify possible inferences in the dynamic and interdepending marine environment around the offshore wind farm. The anomaly results are analyzed in terms of location, density and temporal variability, granting access to information and building a basis for explanation. Temporal detection of anomalies with COPOD is found to be a suitable method. Actionable insights are the direction and magnitude of potential effects of the wind farm on the marine environment, depending on the wind direction. This study works towards a digital twin of offshore wind farms and provides a set of methods based on machine learning to monitor and evaluate offshore wind farm effects, supporting stakeholders with information for decision making on future maritime energy infrastructures.

show abstract

Section: Resultsmentioning

confidence: 99%

Towards Digital Twins of the Oceans: The Potential of Machine Learning for Monitoring the Impacts of Offshore Wind Farms on Marine Environments

Schneider

Klüner

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Lezine et al 2021). GANs and other artificial intelligence models used in earth observation also suffer from a lack of explainability (Gevaert 2022), which can make users less likely to evaluate their uncertainties. While the ethical consequences of misplaced Arctic-boreal lakes shown here are innocuous, both type I and type II errors in other applications of SR object detection, such as intelligence gathering (e.g.…”

Section: Ethical Considerations Of Super Resolution Object Detectionmentioning

confidence: 99%

Contemporary and historical detection of small lakes using cross-sensor super resolution Landsat imagery

Kyzivat¹,

Smith²

2024

Preprint

View full text Add to dashboard Cite

Landsat is the longest-running environmental satellite program and has been used for surface water mapping of large water bodies since its launch in 1972. Remote sensing image resolution is increasingly being enhanced through single image super resolution (SR), a machine learning task typically performed by neural networks. Here, we show that a 10x SR model (Enhanced Super Resolution GAN, or ESRGAN) trained entirely with Planet SmallSat imagery (3 m resolution) can be applied to 30 m Landsat imagery to produce 3 m Landsat SR images with preserved radiometric properties. We test the utility of these Landsat SR images for small lake detection by applying a simple water classification to SR and native Landsat imagery and comparing to independent, high-resolution water maps. SR images appear realistic and have fewer missed detections (type II error) compared to LR, but exhibit errors in lake location and shape, and yield increasing false detections (type I error) with decreasing lake size. SR enhancement improves detection of small lakes sized several Landsat pixels or less, with a minimum mapping unit (MMU) of ~ 2/3 of a Landsat pixel. We also apply the SR model to a historical Landsat 5 image and find similar performance gains, using an independent 1985 air photo map of 242 small Alaskan lakes. This demonstration of retroactively generated 3 m imagery dating to 1985 has exciting applications beyond water detection. Yet, much work remains to be done surrounding technical and ethical guidelines for the creation, use, and dissemination of SR satellite imagery.

show abstract

“…In many earth system science applications, the computational efficiency of traditional tools is a significant bottleneck and available training data is voluminous. These models have a major drawback, however: models are a black box and it is therefore often not clear how they are generating their predictions (Gevaert, 2022). Using testing and validation data sets can provide some level of confidence in the models by demonstrating their level of accuracy on data not used for training.…”

Section: Explainable Aimentioning

confidence: 99%

A Machine Learning Augmented Data Assimilation Method for High‐Resolution Observations

Howard,

Subramanian,

Hoteit

2024

J Adv Model Earth Syst

View full text Add to dashboard Cite

The accuracy of initial conditions is an important driver of the forecast skill of numerical weather prediction models. Increases in the quantity of available measurements, particularly high‐resolution remote sensing observational data products from satellites, are valuable inputs for improving those initial condition estimates. However, the traditional data assimilation methods for integrating observations into forecast models are computationally expensive. This makes incorporating dense observations into operational forecast systems challenging, and it is often prohibitively time‐consuming. Additionally, high‐resolution observations often have correlated observation errors which are difficult to estimate and create problems for assimilation systems. As a result, large quantities of data are discarded and not used for state initialization. Using the Lorenz‐96 system for testing, we demonstrate that a simple machine learning method can be trained to assimilate high‐resolution data. Using it to do so improves both initial conditions and forecast accuracy. Compared to using the Ensemble Kalman Filter with high‐resolution observations ignored, our augmented method has an average root‐mean‐squared error reduced by 37%. Ensemble forecasts using initial conditions generated by the augmented method are more accurate and reliable at up to 10 days of forecast lead time.

show abstract

Explainable AI for earth observation: A review including societal and regulatory perspectives

Cited by 45 publications

References 63 publications

Towards Digital Twins of the Oceans: The Potential of Machine Learning for Monitoring the Impacts of Offshore Wind Farms on Marine Environments

Towards Digital Twins of the Oceans: The Potential of Machine Learning for Monitoring the Impacts of Offshore Wind Farms on Marine Environments

Contemporary and historical detection of small lakes using cross-sensor super resolution Landsat imagery

A Machine Learning Augmented Data Assimilation Method for High‐Resolution Observations

Contact Info

Product

Resources

About