Browse

You are looking at 111 - 120 of 145 items for :

  • Artificial Intelligence for the Earth Systems x
  • Refine by Access: All Content x
Clear All
Andrew Geiss
and
Joseph C. Hardin

Abstract

Recently, deep convolutional neural networks (CNNs) have revolutionized image “super resolution” (SR), dramatically outperforming past methods for enhancing image resolution. They could be a boon for the many scientific fields that involve imaging or any regularly gridded datasets: satellite remote sensing, radar meteorology, medical imaging, numerical modeling, and so on. Unfortunately, while SR-CNNs produce visually compelling results, they do not necessarily conserve physical quantities between their low-resolution inputs and high-resolution outputs when applied to scientific datasets. Here, a method for “downsampling enforcement” in SR-CNNs is proposed. A differentiable operator is derived that, when applied as the final transfer function of a CNN, ensures the high-resolution outputs exactly reproduce the low-resolution inputs under 2D-average downsampling while improving performance of the SR schemes. The method is demonstrated across seven modern CNN-based SR schemes on several benchmark image datasets, and applications to weather radar, satellite imager, and climate model data are shown. The approach improves training time and performance while ensuring physical consistency between the super-resolved and low-resolution data.

Significance Statement

Recent advancements in using deep learning to increase the resolution of images have substantial potential across the many scientific fields that use images and image-like data. Most image super-resolution research has focused on the visual quality of outputs, however, and is not necessarily well suited for use with scientific data where known physics constraints may need to be enforced. Here, we introduce a method to modify existing deep neural network architectures so that they strictly conserve physical quantities in the input field when “super resolving” scientific data and find that the method can improve performance across a wide range of datasets and neural networks. Integration of known physics and adherence to established physical constraints into deep neural networks will be a critical step before their potential can be fully realized in the physical sciences.

Open access
Yongquan Qu
and
Xiaoming Shi

Abstract

The development of machine learning (ML) techniques enables data-driven parameterizations, which have been investigated in many recent studies. Some investigations suggest that a priori-trained ML models exhibit satisfying accuracy during training but poor performance when coupled to dynamical cores and tested. Here we use the evolution of the barotropic vorticity equation (BVE) with periodically reinforced shear instability as a prototype problem to develop and evaluate a model-consistent training strategy, which employs a numerical solver supporting automatic differentiation and includes the solver in the loss function for training ML-based subgrid-scale (SGS) turbulence models. This approach enables the interaction between the dynamical core and the ML-based parameterization during the model training phase. The BVE model was run at low, high, and ultrahigh (truth) resolutions. Our training dataset contains only a short period of coarsened high-resolution simulations. However, given initial conditions long after the training dataset time, the trained SGS model can still significantly increase the effective lead time of the BVE model running at the low resolution by up to 50% relative to the BVE simulation without an SGS model. We also tested using a covariance matrix to normalize the loss function and found it can notably boost the performance of the ML parameterization. The SGS model’s performance is further improved by conducting transfer learning using a limited number of discontinuous observations, increasing the forecast lead-time improvement to 73%. This study demonstrates a potential pathway to using machine learning to enhance the prediction skills of our climate and weather models.

Significance Statement

Numerical weather prediction is performed at limited resolution for computational feasibility, and the schemes to estimate unresolved processes are called parameterization. We propose a strategy to develop better deep learning–based parameterization in which an automatic differentiable numerical solver is employed as the dynamic core and interacts with the parameterization scheme during its training. Such a numerical solver enables consistent deep learning, because the parameterization is trained with a direct target of making the numerical model (dynamic core and parameterization together) forecast match observations as much as possible. We demonstrate the feasibility and effectiveness of such a strategy using a surrogate model and advocate that such machine learning–enabled numerical models provide a promising pathway to developing next-generation weather forecast and climate models.

Open access
Fatemeh Farokhmanesh
,
Kevin Höhlein
, and
Rüdiger Westermann

Abstract

Numerical simulations in Earth-system sciences consider a multitude of physical parameters in space and time, leading to severe input/output (I/O) bandwidth requirements and challenges in subsequent data analysis tasks. Deep learning–based identification of redundant parameters and prediction of those from other parameters, that is, variable-to-variable (V2V) transfer, has been proposed as an approach to lessening the bandwidth requirements and streamlining subsequent data analysis. In this paper, we examine the applicability of V2V to meteorological reanalysis data. We find that redundancies within pairs of parameter fields are limited, which hinders application of the original V2V algorithm. Therefore, we assess the predictive strength of reanalysis parameters by analyzing the learning behavior of V2V reconstruction networks in an ablation study. We demonstrate that efficient V2V transfer becomes possible when considering groups of parameter fields for transfer and propose an algorithm to implement this. We investigate further whether the neural networks trained in the V2V process can yield insightful representations of recurring patterns in the data. The interpretability of these representations is assessed via layerwise relevance propagation that highlights field areas and parameters of high importance for the reconstruction model. Applied to reanalysis data, this allows for uncovering mutual relationships between landscape orography and different regional weather situations. We see our approach as an effective means to reduce bandwidth requirements in numerical weather simulations, which can be used on top of conventional data compression schemes. The proposed identification of multiparameter features can spawn further research on the importance of regional weather situations for parameter prediction and also in other kinds of simulation data.

Open access
Yanan Duan
,
Sathish Akula
,
Sanjiv Kumar
,
Wonjun Lee
, and
Sepideh Khajehei

Abstract

The National Oceanic and Atmospheric Administration has developed a very high-resolution streamflow forecast using National Water Model (NWM) for 2.7 million stream locations in the United States. However, considerable challenges exist for quantifying uncertainty at ungauged locations and forecast reliability. A data science approach is presented to address the challenge. The long-range daily streamflow forecasts are analyzed from December 2018 to August 2021 for Alabama and Georgia. The forecast is evaluated at 389 observed USGS stream gauging locations using standard deterministic metrics. Next, the forecast errors are grouped using watersheds’ biophysical characteristics, including drainage area, land use, soil type, and topographic index. The NWM forecasts are more skillful for larger and forested watersheds than smaller and urban watersheds. The NWM forecast considerably overestimates the streamflow in the urban watersheds. The classification and regression tree analysis confirm the dependency of the forecast errors on the biophysical characteristics. A densely connected neural network model consisting of six layers [deep learning (DL)] is developed using biophysical characteristics, NWM forecast as inputs, and the forecast errors as outputs. The DL model successfully learns location invariant transferrable knowledge from the domain trained in the gauged locations and applies the learned model to estimate forecast errors at the ungauged locations. A temporal and spatial split of the gauged data shows that the probability of capturing the observations in the forecast range improved significantly in the hybrid NWM-DL model (82% ± 3%) than in the NWM-only forecast (21% ± 1%). A trade-off between overly constrained NWM forecast and increased forecast uncertainty range in the DL model is noted.

Significance Statement

A hybrid biophysical–artificial intelligence (physics–AI) model is developed from the first principle to estimate streamflow forecast errors at ungauged locations, improving the forecast’s reliability. The first principle refers to identifying the need for the hybrid physics–AI model, determining physically interpretable and machine identifiable model inputs, followed by the deep learning (DL) model development and its evaluations, and finally, a biophysical interpretation of the hybrid model. A very high-resolution National Water Model (NWM) forecast, developed by the National Oceanic and Atmospheric Administration, serves as the biophysical component of the hybrid model. Out of 2.7 million daily forecasts, less than 1% of the forecasts can be verified using the traditional hydrological method of comparing the forecast with the observations, motivating the need for the AI technique to improve forecast reliability at millions of ungauged locations. An exploratory analysis followed by the classification and regression tree analysis successfully determines the dependency of the forecast errors on the biophysical attributes, which along with the NWM forecast, are used for the DL model development. The hybrid model is evaluated in a subtropical humid climate of Alabama and Georgia in the United States. Long-term streamflow forecasts from zero-day lead to 30-day lead forecasts are archived and analyzed for 979 days (December 2018–August 2021) and 389 USGS gauging stations. The forecast reliability is assessed as the probability of capturing the observations in its ensemble range. As a result, the forecast reliability increased from 21% (±1%) in the NWM only forecasts to 82% (±3%) in the hybrid physics–AI model.

Open access
Antonios Mamalakis
,
Elizabeth A. Barnes
, and
Imme Ebert-Uphoff

Abstract

Methods of explainable artificial intelligence (XAI) are used in geoscientific applications to gain insights into the decision-making strategy of neural networks (NNs), highlighting which features in the input contribute the most to a NN prediction. Here, we discuss our “lesson learned” that the task of attributing a prediction to the input does not have a single solution. Instead, the attribution results depend greatly on the considered baseline that the XAI method utilizes—a fact that has been overlooked in the geoscientific literature. The baseline is a reference point to which the prediction is compared so that the prediction can be understood. This baseline can be chosen by the user or is set by construction in the method’s algorithm—often without the user being aware of that choice. We highlight that different baselines can lead to different insights for different science questions and, thus, should be chosen accordingly. To illustrate the impact of the baseline, we use a large ensemble of historical and future climate simulations forced with the shared socioeconomic pathway 3-7.0 (SSP3-7.0) scenario and train a fully connected NN to predict the ensemble- and global-mean temperature (i.e., the forced global warming signal) given an annual temperature map from an individual ensemble member. We then use various XAI methods and different baselines to attribute the network predictions to the input. We show that attributions differ substantially when considering different baselines, because they correspond to answering different science questions. We conclude by discussing important implications and considerations about the use of baselines in XAI research.

Significance Statement

In recent years, methods of explainable artificial intelligence (XAI) have found great application in geoscientific applications, because they can be used to attribute the predictions of neural networks (NNs) to the input and interpret them physically. Here, we highlight that the attributions—and the physical interpretation—depend greatly on the choice of the baseline—a fact that has been overlooked in the geoscientific literature. We illustrate this dependence for a specific climate task, in which a NN is trained to predict the ensemble- and global-mean temperature (i.e., the forced global warming signal) given an annual temperature map from an individual ensemble member. We show that attributions differ substantially when considering different baselines, because they correspond to answering different science questions.

Open access
Christian Sigg
,
Flavia Cavallaro
,
Tobias Günther
, and
Martin R. Oswald

Abstract

Outdoor webcam images jointly visualize many aspects of the past and present weather, and since they are also easy to interpret, they are consulted by meteorologists and the general public alike. Weather forecasts, in contrast, are communicated as text, pictograms, or charts, each focusing on separate aspects of the future weather. We therefore introduce a method that uses photographic images to also visualize weather forecasts. This is a challenging task because photographic visualizations of weather forecasts should look real and match the predicted weather conditions, the transition from observation to forecast should be seamless, and there should be visual continuity between images for consecutive lead times. We use conditional generative adversarial networks to synthesize such visualizations. The generator network, conditioned on the analysis and the forecasting state of the numerical weather prediction (NWP) model, transforms the present camera image into the future. The discriminator network judges whether a given image is the real image of the future, or whether it has been synthesized. Training the two networks against each other results in a visualization method that scores well on all four evaluation criteria. We present results for three camera sites across Switzerland that differ in climatology and terrain. We show that even experts struggle to distinguish real from generated images, achieving only a 59% accuracy. The generated images match the atmospheric, ground, and illumination conditions visible in the true future images in 67% up to 99% of cases. Nowcasting sequences of generated images achieve a seamless transition from observation to forecast and attain good visual continuity.

Open access
Free access
Free access
Caren Marzban
,
Jueyi Liu
, and
Philippe Tissot

Abstract

Resampling methods such as cross validation or bootstrap are often employed to estimate the uncertainty in a loss function due to sampling variability, usually for the purpose of model selection. In models that require nonlinear optimization, however, the existence of local minima in the loss function landscape introduces an additional source of variability that is confounded with sampling variability. In other words, some portion of the variability in the loss function across different resamples is due to local minima. Given that statistically sound model selection is based on an examination of variance, it is important to disentangle these two sources of variability. To that end, a methodology is developed for estimating each, specifically in the context of K-fold cross validation, and neural networks (NN) whose training leads to different local minima. Random effects models are used to estimate the two variance components—that due to sampling and that due to local minima. The results are examined as a function of the number of hidden nodes, and the variance of the initial weights, with the latter controlling the “depth” of local minima. The main goal of the methodology is to increase statistical power in model selection and/or model comparison. Using both simulated and realistic data, it is shown that the two sources of variability can be comparable, casting doubt on model selection methods that ignore the variability due to local minima. Furthermore, the methodology is sufficiently flexible so as to allow assessment of the effect of other/any NN parameters on variability.

Free access
Utkarsh Mital
,
Dipankar Dwivedi
,
Ilhan Özgen-Xian
,
James B. Brown
, and
Carl I. Steefel

Abstract

An accurate characterization of the water content of snowpack, or snow water equivalent (SWE), is necessary to quantify water availability and constrain hydrologic and land surface models. Recently, airborne observations (e.g., lidar) have emerged as a promising method to accurately quantify SWE at high resolutions (scales of ∼100 m and finer). However, the frequency of these observations is very low, typically once or twice per season in the Rocky Mountains of Colorado. Here, we present a machine learning framework that is based on random forests to model temporally sparse lidar-derived SWE, enabling estimation of SWE at unmapped time points. We approximated the physical processes governing snow accumulation and melt as well as snow characteristics by obtaining 15 different variables from gridded estimates of precipitation, temperature, surface reflectance, elevation, and canopy. Results showed that, in the Rocky Mountains of Colorado, our framework is capable of modeling SWE with a higher accuracy when compared with estimates generated by the Snow Data Assimilation System (SNODAS). The mean value of the coefficient of determination R 2 using our approach was 0.57, and the root-mean-square error (RMSE) was 13 cm, which was a significant improvement over SNODAS (mean R 2 = 0.13; RMSE = 20 cm). We explored the relative importance of the input variables and observed that, at the spatial resolution of 800 m, meteorological variables are more important drivers of predictive accuracy than surface variables that characterize the properties of snow on the ground. This research provides a framework to expand the applicability of lidar-derived SWE to unmapped time points.

Significance Statement

Snowpack is the main source of freshwater for close to 2 billion people globally and needs to be estimated accurately. Mountainous snowpack is highly variable and is challenging to quantify. Recently, lidar technology has been employed to observe snow in great detail, but it is costly and can only be used sparingly. To counter that, we use machine learning to estimate snowpack when lidar data are not available. We approximate the processes that govern snowpack by incorporating meteorological and satellite data. We found that variables associated with precipitation and temperature have more predictive power than variables that characterize snowpack properties. Our work helps to improve snowpack estimation, which is critical for sustainable management of water resources.

Free access