Research news on time series analysis

Time series analysis comprises a set of statistical and computational methods for modeling, decomposing, and forecasting data indexed in time order, explicitly accounting for temporal dependence, nonstationarity, and serial correlation. Core approaches include autoregressive (AR), moving average (MA), and ARIMA-family models, state-space and Kalman filter methods, spectral and wavelet analysis, and modern machine learning–based sequence models. Typical objectives are identification of underlying stochastic processes, estimation of parameters, detection of structural breaks, and prediction of future values, while addressing issues such as seasonality, trends, heteroskedasticity, and autocorrelated residuals under rigorously defined probabilistic frameworks.

Filling the gaps in global rainfall maps

Global rainfall maps are built from observations collected by multiple satellites, but they have an important limitation. The satellites that provide these measurements cannot see the whole Earth at once. As a result, some ...

Can satellites detect when forests lose resilience?

For years, scientists have been using a method called "temporal autocorrelation (TAC)" to measure forest resilience using satellite data. However, scientists have not used TAC uniformly, and its use as a resilience indicator ...

Seismic activity in California varies with the seasons

Earthquakes occur when the tectonic plates of the Earth's crust shift, jolting past each other in a release of built-up tension. However, other natural forces can also influence seismic activity: Hydrological dynamics, like ...

page 1 from 2