Ankush Khandelwal

High spatiotemporal resolution of river planform dynamics from Landsat: The RivMAP toolbox and results from the Ucayali River

Quantifying planform changes of large and actively migrating rivers such as those in the tropical Amazon at multidecadal time scales, over large spatial domains, and with high spatiotemporal frequency is essential for advancing river morphodynamic theory, identifying controls on migration, and understanding the roles of climate and human influences on planform adjustments. This paper addresses the challenges of quantifying river planform changes from annual channel masks derived from Landsat imagery and introduces a set of efficient methods to map and measure changes in channel widths, the locations and rates of migration, accretion and erosion, and the space-time characteristics of cutoff dynamics. The techniques are assembled in a comprehensive MATLAB toolbox called RivMAP (River Morphodynamics from Analysis of Planforms), which is applied to over 1500 km of the actively migrating and predominately meandering Ucayali River in Peru from 1985 to 2015. We find multiscale spatial and temporal variability around multidecadal trends in migration rates, erosion and accretion, and channel widths revealing a river dynamically adjusting to sediment and water fluxes. Confounding factors controlling planform morphodynamics including local inputs of sediment, cutoffs, and climate are parsed through the high temporal analysis.

Global Monitoring of Inland Water Dynamics: State-of-the-Art, Challenges, and Opportunities

Inland water is an important natural resource that is critical for sustaining marine and terrestrial ecosystems as well as supporting a variety of human needs. Monitoring the dynamics of inland water bodies at a global scale is important for: (a) devising effective water management strategies, (b) assessing the impact of human actions on water security, (c) understanding the interplay between the spatio-temporal dynamics of surface water and climate change, and (d) near-real time mitigation and management of disaster events such as floods. Remote sensing datasets provide opportunities for global-scale monitoring of the extent or surface area of inland water bodies over time. We present a survey of existing remote sensing based approaches for monitoring the extent of inland water bodies and discuss their strengths and limitations. We further present an outline of the major challenges that need to be addressed for monitoring the extent and dynamics of water bodies at a global scale. Potential opportunities for overcoming some of these challenges are discussed using illustrative examples, laying the foundations for promising directions of future research in global monitoring of water dynamics.

Predictive Learning in the Presence of Heterogeneity and Limited Training Data.

A large number of real-world domains possess heterogeneity in their data, which implies that different partitions of the data show different relationships between explanatory and response variables. This increases the overall model complexity of predictive learning in the presence of heterogeneity. Additionally, a number of real-world domains lack sufficient training data, making the learning algorithm prone to over-fitting, especially when the model complexity is large. However, there often exists a structure among the data instances and their partitions which can be appropriately leveraged for reducing the model complexity along with addressing heterogeneity. In this paper, we present a framework for learning robust predictive models in real-world heterogeneous datasets which lack sufficient number of training samples. We demonstrate the usefulness of our framework in the domain of remote sensing for forest cover estimation. Through a series of comparative experiments with baseline approaches, we are able to show that our framework: (a) captures meaningful information about heterogeneity in the data, (b) improves prediction performance by addressing data heterogeneity, (c) is robust to over-fitting in the presence of limited training data, and (d) is robust to the choice of the number of partitions used for representing heterogeneity.

Incremental Dual-memory LSTM in Land Cover Prediction

Land cover prediction is essential for monitoring global environmental change. Unfortunately, traditional classification models are plagued by temporal variation and emergence of novel/unseen land cover classes in the prediction process. In this paper, we propose an LSTM-based spatio-temporal learning framework with a dual-memory structure. The dual-memory structure captures both long-term and short-term temporal variation patterns, and is updated incrementally to adapt the model to the ever-changing environment. Moreover, we integrate zero-shot learning to identify unseen classes even without labelled samples. Experiments on both synthetic and real-world datasets demonstrate the superiority of the proposed framework over multiple baselines in land cover prediction.

Post Classification Label Refinement Using Implicit Ordering Constraint Among Data Instances

Classification of instances into different categories in various real world applications suffer from inaccuracies due to lack of representative training data, limitations of classification models, noise and outliers in the input data etc. In this paper we propose a new post classification label refinement method for the scenarios where data instances have an inherent ordering among them that can be leveraged to correct inconsistencies in class labels. We show that by using the ordering constraint, more robust algorithms can be developed than traditional methods. Moreover in most applications where this ordering among instances exists, it is not directly observed. The proposed approach simultaneously estimates the latent ordering among instances and corrects the class labels. We demonstrate the utility of the approach for the application of monitoring the dynamics of lakes and reservoirs. The proposed approach has been evaluated on synthetic datasets with different noise structures and noise levels.

Earth Science Applications of Sensor Data

Advances in earth observation technologies have led to the acquisition of vast volumes of accurate, timely and reliable environmental data which encompass a multitude of information about the land, ocean and atmosphere of the planet. Earth science sensor datasets capture multiple facets of information about natural processes and human activities that shape the physical landscape and environmental quality of our planet, and thus, offer an opportunity to monitor and understand the diverse phenomena affecting earth’s complex system. The monitoring, analysis and understanding of these rich sensor datasets is thus of prime importance for the efficient planning and management of critical resources, since the societal costs of mitigation or adaptation decisions for natural or human-induced adverse events are significant. Hence, a thorough understanding of earth science sensor datasets has a direct impact on a range of societally relevant issues. Moreover, earth science sensor datasets possess unique domain-specific properties that distinguish them from sensor datasets used in other domains, and thus demand the need for novel tools and techniques to be developed for their analysis, adhering to their characteristic issues and challenges.

RAPT: Rare Class Prediction in Absence of True Labels

Many real-world problems involve learning models for rare classes in situations where there are no gold standard labels for training samples but imperfect labels are available for all instances. In this paper, we present RAPT, a three step predictive modeling framework for classifying rare class in such problem settings. The first step of the proposed framework learns a classifier that jointly optimizes precision and recall by only using imperfectly labeled training samples. We also show that, under certain assumptions on the imperfect labels, the quality of this classifier is almost as good as the one constructed using perfect labels. The second and third steps of the framework make use of the fact that imperfect labels are available for all instances to further improve the precision and recall of the rare class. We evaluate the RAPT framework on two real-world applications of mapping forest fires and urban extent from earth observing satellite data. The experimental results indicate that RAPT can be used to identify forest fires and urban areas with high precision and recall by using imperfect labels, even though obtaining expert annotated samples on a global scale is infeasible in these applications.

Sparse Gaussian Markov Random Field Mixtures for Anomaly Detection

We propose a new approach to anomaly detection from multivariate noisy sensor data. We address two major challenges: To provide variable-wise diagnostic information and to automatically handle multiple operational modes. Our task is a practical extension of traditional outlier detection, which is to compute a single scalar for each sample. To consistently define the variable-wise anomaly score, we leverage a predictive conditional distribution. We then introduce a mixture of Gaussian Markov random field and its Bayesian inference, resulting in a sparse mixture of sparse graphical models. Our anomaly detection method is capable of automatically handling multiple operational modes while removing unwanted nuisance variables. We demonstrate the utility of our approach using real equipment data from the oil industry.

Learning large-scale plantation mapping from imperfect annotators

Plantation mapping is important for understanding deforestation and climate change. Most existing plantation products rely heavily on visual interpretation of satellite imagery, which results in both false positives and false negatives. In this paper we aim to design an automatic framework that map plantations in large regions. Conventional classification methods cannot be directly applied due to the lack of ground-truth data. To this end, we propose a novel method that learns from multiple imperfect annotators. Since each annotators labeling accuracy varies across different land covers due to his expertise and reference imagery, we model the annotators reliability level to be associated with different types of locations. On the other hand, the temporal variation of land covers also greatly impacts the performance of conventional learning model. Therefore we utilize the remote sensing data which are available at multiple periods of a year and extend our proposed method by incorporating multi-instance learning. Finally, we show the superiority of the proposed method over multiple baselines in both synthetic dataset and real-world dataset. In addition, through several case studies we demonstrate that our method can achieve a better balance of precision and recall than the existing plantation products.

Unsupervised Method for Water Surface Extent Monitoring Using Remote Sensing Data

Inland surface water availability is a serious global sustainability challenge. Hence, there is a need to monitor surface water availability, in order to better manage it under an increasingly changing planet. So far, a comprehensive effort to understand changes in inland surface water availability and dynamics is lacking. Remote sensing instruments provide an opportunity to monitor surface water availability on a global scale, but they also introduce significant computational challenges. In this chapter, we present an unsupervised method that overcomes several challenges inherent in remote sensing data to effectively monitor changes in surface water bodies. Using an independent validation dataset, we compare the proposed method with two cluster algorithms (K-MEANS and EM) as well as an image segmentation algorithm (normal-cut). We show that our method is more efficient and reliable.