اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدAchieving Conservation of Energy in Neural Network Emulators for Climate Modeling
92
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Artificial neural-networks have the potential to emulate cloud processes with higher accuracy than the semi-empirical emulators currently used in climate models. However, neural-network models do not intrinsically conserve energy and mass, which is an obstacle to using them for long-term climate predictions. Here, we propose two methods to enforce linear conservation laws in neural-network emulators of physical models: Constraining (1) the loss function or (2) the architecture of the network itself. Applied to the emulation of explicitly-resolved cloud processes in a prototype multi-scale climate model, we show that architecture constraints can enforce conservation laws to satisfactory numerical precision, while all constraints help the neural-network better generalize to conditions outside of its training set, such as global warming.
قيم البحث
اقرأ أيضاً
Different definitions of links in climate networks may lead to considerably different network topologies. We construct a network from climate records of surface level atmospheric temperature in different geographical sites around the globe using two
commonly used definitions of links. Utilizing detrended fluctuation analysis, shuffled surrogates and separation analysis of maritime and continental records, we find that one of the major influences on the structure of climate networks is due to the auto-correlation in the records, that may introduce spurious links. This may explain why different methods could lead to different climate network topologies.
We apply two independent data analysis methodologies to locate stable climate states in an intermediate complexity climate model and analyze their interplay. First, drawing from the theory of quasipotentials, and viewing the state space as an energy
landscape with valleys and mountain ridges, we infer the relative likelihood of the identified multistable climate states, and investigate the most likely transition trajectories as well as the expected transition times between them. Second, harnessing techniques from data science, specifically manifold learning, we characterize the data landscape of the simulation output to find climate states and basin boundaries within a fully agnostic and unsupervised framework. Both approaches show remarkable agreement, and reveal, apart from the well known warm and snowball earth states, a third intermediate stable state in one of the two climate models we consider. The combination of our approaches allows to identify how the negative feedback of ocean heat transport and entropy production via the hydrological cycle drastically change the topography of the dynamical landscape of Earths climate.
The connectivity pattern of networks, which are based on a correlation between ground level temperature time series, shows a dominant dense stripe of links in the southern ocean. We show that statistical categorization of these links yields a clear a
ssociation with the pattern of an atmospheric Rossby wave, one of the major mechanisms associated with the weather system and with planetary scale energy transport. It is shown that alternating densities of negative and positive links (correlations) are arranged in half Rossby wave distances around 3,500 km, 7,000 km and 10,000 km and are aligned with the expected direction of energy flow, distribution of time delays and the seasonality of these waves. It is also shown that long distance links (i.e., of distances larger than 2,000 km) that are associated with Rossby waves are the most dominant in the climate network. Climate networks may thus be used as an efficient new way to detect and analyze Rossby waves, based on reliable and available ground level measurements, in addition to the frequently used 300 hPa reanalysis meridional wind data.
This study investigates the trend in Rainfall Onset Dates (ROD), Rainfall Cessation Dates (RCD), Length of Growing Seasons (LGS) and Rainfall Amount at Onset of Rainfall (RAO) using linear regression, Mann-Kendall, Sen Slope and Hurst Exponent for fo
ur locations in tropical Nigeria and the development of a Fourier based model for ROD and RCD. Daily data was obtained from the Nigerian Meteorological Agency for thirty-four (34) years (1979 - 2013). ROD and RCD were computed using the method of cumulative percentage mean rainfall values. Maiduguri, Gusau and Ikom showed positive trends in ROD and RCD while Ibadan exhibited negative trends in the two parameters. Anti-persistence was observed in ROD, RCD and LGS for three locations (Maiduguri, Gusau and Ibadan). A Fourier based model with seven (7) coefficients was developed to model ROD and RCD for all the locations. The model developed performed very well in all locations with the best performance obtained in Gusau and Ibadan for ROD and RCD respectively. The effects of climate change on agricultural output for the four (4) locations under consideration were highlighted and adaption techniques suggested for mitigating the impact on agricultural output and livelihood of citizens in the areas.
One hidden yet important issue for developing neural network potentials (NNPs) is the choice of training algorithm. Here we compare the performance of two popular training algorithms, the adaptive moment estimation algorithm (Adam) and the extended K
alman filter algorithm (EKF), using the Behler-Parrinello neural network (BPNN) and two publicly accessible datasets of liquid water. It is found that NNPs trained with EKF are more transferable and less sensitive to the value of the learning rate, as compared to Adam. In both cases, error metrics of the test set do not always serve as a good indicator for the actual performance of NNPs. Instead, we show that their performance correlates well with a Fisher information based similarity measure.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
