Water Cycle Nonstationarity

The global water cycle is undergoing unprecedented shifts from climate change, intensified by human water and land management practices. These changes are evident in phenomena such as depleted groundwater, earlier snowmelt, and erratic fluctuations in floods and drought occurrences. To better understand these changes in terrestrial water storage, scientists have integrated multiple remote sensing datasets with NASA’s advanced land surface model through data assimilation, creating a global water storage reanalysis dataset. The results capture the complex patterns of global water cycle shifts in response to both climate and human activities. Using this new integrated dataset, scientists use statistical methods (time series analysis) to identify trends (TR), seasonal shifts (SS), and changes in extreme events (EFR), ultimately developing an index, the “Nonstationarity Index,” (NSI) that quantifies the degree of nonstationarity within the global water system.

The following outlines the method we use to estimate nonstationarity for a single grid cell in southeastern Australia. For a comprehensive overview of the changes in Terrestrial Water Storage (TWS) in this region, please visit the Terrestrial Water Storage: Regional Views 2003–2019 webpage.

First, the original TWS time series is decomposed into three components: long-term trend, seasonal variation, and residual (remainder). Next, we estimate three key metrics based on the decomposed time series:

1. Theil-Sen Slope (TR): This measures the overall trend of the water storage time series.

2. Seasonal Peak Shift (SS): This quantifies any shift in the timing of the seasonal peak of terrestrial water storage.

3. Extreme Frequency Ratio (EFR): For this metric, we convert the residual time series into a binary series using standard scores, where 0 represents normal conditions and 1 represents extreme events. We then identify a change point in the binary series, and the extreme frequency ratio is calculated as the ratio of extreme events occurring before and after the change point.

Each of these three metrics is then ranked within its global context and integrated into the nonstationarity index (NSI).

Related references:

Featured story: NASA Scientists Find New Human-Caused Shifts in Global Water Cycle

NASA Land Information System: NASA LIS Framework