Representation learning has proven to be a powerful methodology in a wide variety of machine-learning applications. For atmospheric dynamics, however, it has so far not been considered, arguably due to the lack of large-scale, labeled datasets that could be used for training. In this work, we show how to sidestep the difficulty and introduce a self-supervised learning task that is applicable to a wide variety of unlabeled atmospheric datasets. Specifically, we train a neural network on the simple yet intricate task of predicting the temporal distance between atmospheric fields from distinct but nearby times. We demonstrate that training with this task on the ERA5 reanalysis dataset leads to internal representations that capture intrinsic aspects of atmospheric dynamics. For example, when employed as a loss function in other machine-learning applications, the derived AtmoDist distance leads to improved results compared to the $ {\mathrm{\ell}}_2 $-loss. For downscaling one obtains higher resolution fields that match the true statistics more closely than previous approaches and for the interpolation of missing or occluded data the AtmoDist distance leads to results that contain more realistic fine-scale features. Since it is obtained from observational data, AtmoDist also provides a novel perspective on atmospheric predictability.

A common task in Earth Sciences is to infer climate information at local and regional scales from global climate models. Dynamical downscaling requires running expensive numerical models at high resolution, which can be prohibitive due to long model runtimes. On the other hand, statistical downscaling techniques present an alternative approach for learning links between the large- and local-scale climate in a more efficient way. A large number of deep neural network-based approaches for statistical downscaling have been proposed in recent years, mostly based on convolutional architectures developed for computer vision and super-resolution tasks. This paper presents deep learning for empirical downscaling (DL4DS), a python library that implements a wide variety of state-of-the-art and novel algorithms for downscaling gridded Earth Science data with deep neural networks. DL4DS has been designed with the goal of providing a general framework for training convolutional neural networks with configurable architectures and learning strategies to facilitate the conduction of comparative and ablation studies in a robust way. We showcase the capabilities of DL4DS on air quality Copernicus Atmosphere Monitoring Service (CAMS) data over the western Mediterranean area. The DL4DS library can be found in this repository: https://github.com/carlos-gg/dl4ds

Climate change has an impact on soybean production in the USA, necessitating thorough impact studies across broad geographic areas and extended periods to develop appropriate coping strategies. This study investigates the simulated response of soybean in ten major soybean-growing states of the USA under Climate Model Intercomparison Project Phase 5 based on multiple global climate models, two representative concentration pathways [RCP8.5 and RCP4.5] under rainfed and irrigated conditions for 2013–2039, 2043–2069, 2063–2099. The future climate series was developed using Agricultural Model Intercomparison and Improvement Project protocol by applying mean and variability, and CROPGRO-soybean model was explored for soybean simulation under 400 ppm CO2 level and a set of management. Under future climate, intense changes in temperature, precipitation amount and variability are anticipated under RCP8.5 than RCP4.5. As a result, a shorter life cycle, more evapotranspiration, lower grain production, higher water consumption and water productivity were expected under RCP8.5 than RCP4.5 scenarios. A higher reduction in grain yield and water productivity is expected under rainfed than irrigated conditions and intensity increases with advancement towards end of the century. Irrigation tends to decrease adverse climate change effects; however, the marginal economy for irrigation water must be assessed. Since the northern states under study are likely to experience increased grain yields or lower negative impacts, these areas could be the major production zones for soybean production in the future if only climate change is taken into account. Before reaching a convincing conclusion, different adaptation strategies must be thoroughly investigated.

Policy-critical, micro-level statistical data are often unavailable at the desired level of disaggregation. We present a Bayesian methodology for “downscaling” aggregated count data to the micro level, using an outside statistical sample. Our procedure combines numerical simulation with exact calculation of combinatorial probabilities. We motivate our approach with an application estimating the number of farms in a region, using count totals at higher levels of aggregation. In a simulation analysis over varying population sizes, we demonstrate both robustness to sampling variability and outperformance relative to maximum likelihood. Spatial considerations, implementation of “informative” priors, non-spatial classification problems, and best practices are discussed.

The publicly available Digital Soil Map of Sweden (DSMS) contains topsoil clay content information in a 50 m×50 m grid, and can be used as decision support in precision agriculture. However, it is also common that farmers have undertaken their own soil sampling (one sample per hectare with texture analysed in every third sample). In the present study, such soil samples from 403 farms were used to validate topsoil clay content information derived from 1) DSMS, 2) DSMS locally adapted by residual kriging, 3) DSMS locally adapted by regression kriging and 4) inverse distance weighting interpolation of the soil sample data without using DSMS. The latter has been common practice until now. The best method differed depending on the local accuracy of DSMS, the quality of the soil sampling and the spatial variation structure of the topsoil texture. The ‘Best method’ strategy, which meant to apply all the above methods and choose the one that performed best at each farm, significantly reduced the mean absolute error. We recommend using this strategy to locally adapt regional digital soil maps to derive accurate decision support for use in precision agriculture.

We study the numerical performance of a continuous data assimilation (downscaling) algorithm, based on ideas from feedback control theory, in the context of the two-dimensional incompressible Navier-Stokes equations. Our model problem is to recover an unknown reference solution, asymptotically in time, by using continuous-in-time coarse-mesh nodal-point observational measurements of the velocity field of this reference solution (subsampling), as might be measured by an array of weather vane anemometers. Our calculations show that the required nodal observation density is remarkably less than what is suggested by the analytical study; and is in fact comparable to the number of numerically determining Fourier modes, which was reported in an earlier computational study by the authors. Thus, this method is computationally efficient and performs far better than the analytical estimates suggest.