Vipin Kumar

Rapid advancement in inverse modeling methods have brought into light their susceptibility to imperfect data. This has made it imperative to obtain more explainable and trustworthy estimates from these models. In hydrology, basin characteristics can be noisy or missing, impacting streamflow prediction. We propose a probabilistic inverse model framework that can reconstruct robust hydrology basin characteristics from dynamic input weather driver and streamflow response data. We address two aspects of building more explainable inverse models, uncertainty estimation (uncertainty due to imperfect data and imperfect model) and robustness. This can help improve the trust of water managers, handling of noisy data and reduce costs. We also propose an uncertainty based loss regularization that offers removal of 17% of temporal artifacts in reconstructions, 36% reduction in uncertainty and 4% higher coverage rate …

Agriculture contributes nearly a quarter of global greenhouse gas (GHG) emissions, which is motivating interest in adopting certain farming practices that have the potential to reduce GHG emissions or sequester carbon in soil. The related GHG emission (including N2O and CH4) and changes in soil carbon stock are defined here as “agricultural carbon outcomes”. Accurate quantification of agricultural carbon outcomes is the basis for achieving emission reductions for agriculture, but existing approaches for measuring carbon outcomes (including direct measurements, emission factors, and process-based modeling) fall short of achieving the required accuracy and scalability necessary to support credible, verifiable, and cost-effective measurement and improvement of these carbon outcomes. Here we propose a foundational and scalable framework to quantify field-level carbon outcomes for farmland, which is based …

Creating separable representations via representation learning and clustering is critical in analyzing large unstructured datasets with only a few labels. Separable representations can lead to supervised models with better classification capabilities and additionally aid in generating new labeled samples. Most unsupervised and semisupervised methods to analyze large datasets do not leverage the existing small amounts of labels to get better representations. In this paper, we propose a spatiotemporal clustering paradigm that uses spatial and temporal features combined with a constrained loss to produce separable representations. We show the working of this method on the newly published dataset ReaLSAT, a dataset of surface water dynamics for over 680,000 lakes across the world, making it an essential dataset in terms of ecology and sustainability. Using this large un- labelled dataset, we first show how a …

A method includes receiving a satellite image of an area and classifying each pixel in the satellite image as representing water, land or unknown using a model. For each of a plurality of possible water levels, a cost associated with the water level is determined, wherein determining the cost associated with a water level includes determining a number of pixels for which the model classification must change to be consistent with the water level and determining a difference between the water level and a water level determined for the area at a previous time point. The lowest cost water level is selected and used to reclassify at least one pixel.

In many environmental applications, recurrent neural networks (RNNs) are often used to model physical variables with long temporal dependencies. However, due to minibatch training, temporal relationships between training segments within the batch (intra-batch) as well as between batches (inter-batch) are not considered, which can lead to limited performance. Stateful RNNs aim to address this issue by passing hidden states between batches. Since Stateful RNNs ignore intra-batch temporal dependency, there exists a trade-off between training stability and capturing temporal dependency. In this paper, we provide a quantitative comparison of different Stateful RNN modeling strategies, and propose two strategies to enforce both intra- and inter-batch temporal dependency. First, we extend Stateful RNNs by defining a batch as a temporally ordered set of training segments, which enables intra-batch sharing of …

Deep learning (DL) models are increasingly used to make accurate hindcasts of management‐relevant variables, but they are less commonly used in forecasting applications. Data assimilation (DA) can be used for forecasts to leverage real‐time observations, where the difference between model predictions and observations today is used to adjust the model to make better predictions tomorrow. In this use case, we developed a process‐guided DL and DA approach to make 7‐day probabilistic forecasts of daily maximum water temperature in the Delaware River Basin in support of water management decisions. Our modeling system produced forecasts of daily maximum water temperature with an average root mean squared error (RMSE) from 1.1 to 1.4°C for 1‐day‐ahead and 1.4 to 1.9°C for 7‐day‐ahead forecasts across all sites. The DA algorithm marginally improved forecast performance when compared with …

Personalized prediction of responses for individual entities caused by external drivers is vital across many disciplines. Recent machine learning (ML) advances have led to new state-of-the-art response prediction models. Models built at a population level often lead to sub-optimal performance in many personalized prediction settings due to heterogeneity in data across entities (tasks). In personalized prediction, the goal is to incorporate inherent characteristics of different entities to improve prediction performance. In this survey, we focus on the recent developments in the ML community for such entity-aware modeling approaches. ML algorithms often modulate the network using these entity characteristics when they are readily available. However, these entity characteristics are not readily available in many real-world scenarios, and different ML methods have been proposed to infer these characteristics from the data. In this survey, we have organized the current literature on entity-aware modeling based on the availability of these characteristics as well as the amount of training data. We highlight how recent innovations in other disciplines, such as uncertainty quantification, fairness, and knowledge-guided machine learning, can improve entity-aware modeling.

Timely and accurate monitoring of tree crop extent and productivities are necessary for informing policy-making and investments. However, except for a very few tree species (e.g., oil palms) with obvious canopy and extensive planting, most small-crown tree crops are understudied in the remote sensing domain. To conduct large-scale small-crown tree mapping, several key questions remain to be answered, such as the choice of satellite imagery with different spatial and temporal resolution and model generalizability. In this study, we use olive trees in Morocco as an example to explore the two abovementioned questions in mapping small-crown orchard trees using 0.5 m DigitalGlobe (DG) and 3 m Planet imagery and deep learning (DL) techniques. Results show that compared to DG imagery whose mean overall accuracy (OA) can reach 0.94 and 0.92 in two climatic regions, Planet imagery has limited capacity to detect olive orchards even with multi-temporal information. The temporal information of Planet only helps when enough spatial features can be captured, e.g., when olives are with large crown sizes (e.g., >3 m) and small tree spacings (e.g., <3 m). Regarding model generalizability, experiments with DG imagery show a decrease in F1 score up to 5% and OA to 4% when transferring models to new regions with distribution shift in the feature space. Findings from this study can serve as a practical reference for many other similar mapping tasks (e.g., nuts and citrus) around the world.