Runoff Risk Models

Sacramento Soil Moisture Accounting Model (SAC-SMA)

This map is based on outputs from the National Weather Service's Sacramento Soil Moisture Accounting (SAC-SMA) Model. This model takes into account snow accumulation and melt, soil moisture content, and forecast precipitation and temperatures to simulate runoff for moderately sized watersheds (i.e., hundreds of square miles) in Wisconsin. The simulated runoff falls into two main categories: baseflow and faster responding runoff. The faster responding runoff can be broken down further into "impervious runoff", "direct runoff", "interflow", and "surface runoff". These terms are in quotes because these are the names of the runoff mechanisms within the SAC-SMA model and do not necessarily reflect the physical reality of what is happening throughout the watershed.

We determined interflow was the best predictor for field-level runoff risk. This was due to the fact that it is simulated for nearly all significant runoff events in a given basin. Furthermore, virtually all basins generate interflow during runoff events. This is something that isn't necessarily true for the other components. The impervious and direct runoff components represent runoff from precipitation that falls on or near the river channel, so they are not be appropriate indicators for farm field runoff. When surface runoff is simulated this means that 100% of the precipitation falling on the ground is becoming runoff at a watershed scale. This is a fairly rare occurrence over much of Wisconsin and if we used it as a predictor we would miss most field scale runoff events.

Risk Calculations

From Model Values to Spreading Risk

Using the SAC-SMA model's soil moisture estimates, we can make a prediction about field-level runoff. Factors to consider when using such predictions:

  • Models are inevitably much simpler than the physical reality
  • Since a crop field is of such a smaller scale than the watersheds considered in SAC-SMA, local effects can significantly affect the model's applicability
  • Your knowledge of and experience with the landscape is not available to the model!
What this means is that while we hope you will find this model useful, you should always apply your own knowledge of the landscape when assessing spreading risk.


Based on a statistical study of the model results, and comparing them with actual field-runoff data, we came up with threshold values for each watershed basin; when the modeled interflow exceeded this value, we could expect runoff to occur on an "average" crop field in this watershed.


Single-Day Risk

For each of the next 5 days, the risk calculator uses the model results to classify the predicted interflow as runoff risk:

  • Low Risk: The model predicts no interflow for that day, hence no runoff
  • Moderate Risk: The model predicts some interflow, but below the threshold value
  • High Risk: The model predicts interflow in excess of the threshold value
You can view the single-day risk forecasts for any watershed basin by clicking it on the map.


Three-Day Risk Forecast

Next, the calculator assigns a composite risk score for the first, second, and third days of the forecast, looking 72 hours into the future for each one and taking the highest risk seen. This is the color displayed on the map. If a basin shows red on Monday, it means that the model identified Monday, Tuesday, or Wednesday as high-risk days. Since the 590 standard includes a 72-hour incorporation time, the team felt this to be adequately conservative.

Precipitation Forecast

The NWS supplies us with precipitation forecasts for each watershed basin. These are factored into the runoff calculations, but may well be useful guides as you assess risk. You can see the precip forecast for a watershed by clicking on the map.


The SAC-SMA runoff forecast model includes a snowmelt predictor, and reports if any of the predicted runoff has a snowmelt component. We use this in two ways.

  1. 10-day snowmelt alert: If any snowmelt is predicted in the basin for the next 10 days, we alert you by rendering the basin with a red crosshatch pattern.
  2. Daily runoff risk: We report if the runoff risk for any given day has any snowmelt component. This allows you to modulate your risk assessment based on what's in your fields. If all the forecast runoff risk is due to snowmelt and there is no snowpack on or near your fields, your risk assessment would be different from someplace elsewhere in the watershed with significant snow cover.

Note that a 10-day snowmelt alert's border and crosshatch are independent of the overall basin color! Recall that the basin's color (yellow, orange, or red) is driven by the 3-day risk outlook, whether from rain, snowmelt, stored soil moisture, or some combination of factors.

For example, a basin will show yellow (which normally indicates "low risk") if there are no runoff events predicted in the next 3 days. But if snowmelt is forecast on Day 8, the red crosshatch will appear.

This lets you apply your own knowledge of the situation; if no snow is on the ground on or around your fields, the snowmelt alert by itself is not a factor (although you'll certainly want to consider the additional risk of spreading atop frozen soil).

You can view the next 5 day-by-day snowmelt predictions by clicking on the map.

Example 1: Yellow Basin with Snowmelt on Day 4


The Manitowoc River basin is colored yellow, since there is no runoff risk predicted within three days. However, since snowmelt is predicted on Day 4, the basin is crosshatched on the map, and the popup indicates a 10-day snowmelt alert.

Example 2: Yellow Basin with Snowmelt after Day 5


The Bad River basin shows no runoff events forecast within the next 5 days. However, the basin is crosshatched because snowmelt is predicted sometime in the next 10 days.

Forecast Update and Valid Times

The "forecast valid time" is the start point for the model measurements. The forecast is run for valid times of 00Z (6 PM CST) and 12Z (6 AM CST). However, due to model run times and adjustments made as forecasters acquire new data, the forecasts are made available after their "valid times" start. Usually the 00Z forecast is uploaded around midnight CST; the 12Z forecast is uploaded around 9 AM CST, and is revised around 11 AM CST.