Overcoming Data Scarcity in Earth Science

Abstract

The Data Scarcity problem is repeatedly encountered in environmental research. This may induce an inadequate representation of the response’s complexity in any environmental system to any input/change (natural and human-induced). In such a case, before getting engaged with new expensive studies to gather and analyze additional data, it is reasonable first to understand what enhancement in estimates of system performance would result if all the available data could be well exploited. The purpose of this Special Issue, “Overcoming Data Scarcity in Earth Science” in the Data journal, is to draw attention to the body of knowledge that leads at improving the capacity of exploiting the available data to better represent, understand, predict, and manage the behavior of environmental systems at meaningful space-time scales. This Special Issue contains six publications (three research articles, one review, and two data descriptors) covering a wide range of environmental fields: geophysics, meteorology/climatology, ecology, water quality, and hydrology.

Multi-Year Datasets of Groundwater Level and Surface Water Budget in a Central Mediterranean Site

“…They have been taken using a portable electronic water-table meter (Boviar GST-FR100), with an average frequency of one measure every about three days for both wells (more frequently in conjunction with some rainy events; for more details see see Appendix B). Despite the laboriousness of the field operations, the use a water-table meter allow to overcome problems related to the lack of funding for research and the reliability of the measurements in case of equipment malfunction [31][32][33]. Embedded in well hydrographs there is valuable information on aquifer responses to both natural and human stresses [34].…”

Self-Supervised Learning for Data Scarcity in a Fatigue Damage Prognostic Problem

“…At the end of the procedure, the model was trained on the whole dataset (training and validation sets) with a lower learning rate of 10 −5 until convergence. Calculations were performed using PyTorch's core library in Python on NVIDIA V100 GPUs, hence the batch size was chosen to be as large as possible in order to speed up calculations, here 2 12 = 4096 depending on the available memory of the used GPUs, and not too large in order to avoid numerical instability. The model hyperparameters were optimized using a Grid Search algorithm, listed in Table 4: • Autoencoder model illustrated in Fig.…”

“…), which makes it difficult to apply the latest Machine Learning (ML) methods. Many approaches have been proposed to address data scarcity in these various domains, as recently reviewed by [11][12][13]. As faults are rare and structures can be replaced before reaching failure, data scarcity is becoming one of the most important challenges in PHM [14,15].…”

A mixed methods approach to reconstructing hydrographs of an extreme flood in an ungauged catchment in Ostional, Nicaragua

“…Hydrologic processes in these data-limited regions can be difficult to characterize due to several factors, including insufficient human and financial resources, many ungauged catchments, difficult-toaccess terrain, and sparsely placed, shorter-term monitoring sites (Kundzewicz, 2007;Wohl et al, 2012;Zheng et al, 2018;Gorgoglione et al, 2020;Nigussie et al, 2020). These constraints result in spatial and temporally limited hydrometeorological monitoring networks, and few long-term comprehensive hydrologic analyses (Calderón, 2015).…”