The USGS Monthly Water Balance Model Futures Portal

Climate projections suggest future temperature and precipitation may differ markedly from the past, changing the water cycle — for example, when snowmelt peaks. Resource managers need estimates of the water cycle's parts, with their uncertainty and the limits of the climate data behind them.

The USGS Monthly Water Balance Model Futures Portal (my.usgs.gov/mows) was built for that. It summarizes monthly historical and simulated future conditions across the conterminous United States for seven variables:

The model

The Monthly Water Balance Model (MWBM; McCabe and Markstrom, 2007) is a modular system that turns monthly precipitation and temperature into monthly estimates of the rest of the water cycle. It was calibrated across the conterminous United States and run with every climate dataset.

How the model moves water. Inputs are in blue, outputs in red. Modified from McCabe and Markstrom (2007). USGS.

222 climate datasets

• Scenarios: CMIP3 emission scenarios B1, A1B and A2; CMIP5 representative concentration pathways (RCP) 4.5, 6 and 8.5.
• A gap on purpose: CMIP data for 2006 through 2019 are left out, to steer use toward long-term evaluation.
• Downscaling: the Bureau of Reclamation statistically downscaled the general circulation models (GCMs) — global climate models at about 150 km resolution — to 1/8°, about 12 km, using bias-corrected spatial disaggregation.
• Open data: results are in the Monthly Water Balance Model Futures database, with open, machine-independent access.

How the portal works

Three parts:

1. Places: variables are indexed to the Geospatial Fabric for National Hydrologic Modeling (Viger and Bock, 2014) — hydrologic response units (HRUs, areas with similar slope, soils or topography that feed runoff to streams), stream segments, and summary nodes at segment outlets, including streamgages.
2. A Web Feature Service to find and query those features anywhere in the conterminous United States.
3. On-the-fly graphs and reports for the seven variables at any HRU or node.

The user chooses the plot type, location, climate datasets, period, variable and filters. The portal pulls the output from the database, writes a CSV file, sums it to the right time step, and offers both the plot (PNG) and the CSV for download.

Plot types

• Mean monthly, historical: each dataset's mean monthly values, by water year (October 1 to September 30). One version adds measured streamflow at selected USGS streamgages.
• Envelope: year-to-year change in a variable from a historical baseline to the simulated future.
• Box plots: monthly or seasonal spread of simulated future values.

Mean monthly streamflow at a USGS streamgage: measured, gridded station data and CMIP5 simulations. USGS.

Envelope plot of temperature change for one HRU, historical (1955–2005) to simulated future (2020–2099), by CMIP5 pathway (RCP 4.5 and 8.5). USGS.

Box plot of future monthly runoff for one HRU under CMIP3 scenario A2, for 2030, 2060 and 2090. USGS.

Keeping only models that match the past

Earlier work suggests a downscaled GCM should at least reproduce the past before it is trusted for the future (Wood and others, 2004). The portal can filter datasets with a two-sample Kolmogorov–Smirnov (KS) test (Conover, 1971), a nonparametric test of whether two samples come from the same population.

• What it compares: each dataset's historical distribution of a variable against the GSD "observed" dataset at the chosen place.
• Three levels: p-values of 0.01, 0.05 and 0.10. 0.01 is the least stringent; 0.10 the most, rejecting more datasets and leaving fewer in the plot.
• What happens to failures: a dataset whose p-value falls below the chosen level is dropped from the plot and marked in the CSV header — its column name starts with "0-" instead of "1-".
• Or none: "No Subset" turns the filter off.

Sources

Based on The U.S. Geological Survey Monthly Water Balance Model Futures Portal, U.S. Geological Survey Fact Sheet 2017–3002 (publication page), prepared with the Department of the Interior South Central Climate Science Center and the U.S. Environmental Protection Agency; data in Bock and others (2016), Monthly Water Balance Model Futures. Figures and the dataset table are restored from the fact sheet's PDF; its map and portal screenshot, drawn on a licensed commercial basemap, are left out. A work of the United States government in the public domain.