Evaluating Riparian Restoration Outcomes Using Climate Engine

This article accompanies and supports the video demonstration of Suzie Creek, Nevada, Evaluating Riparian Restoration Outcomes Using Climate Engine, embedded below. This procedure presented here is supported by the publication Assessing the effectiveness of riparian restoration projects using Landsat and precipitation data from the cloud-computing application ClimateEngine.org.


Riparian vegetation along streams provides a variety of ecosystem services such as habitat value for a variety of species, regulation of water temperatures, and regulation of water flow rates. For this reason, when these systems become degraded, land management agencies and organizations prioritize restoring vegetation and hydrologic function. Monitoring the effectiveness of restoration projects is essential, but collecting pre- and post-restoration data is typically expensive and can often be impossible for restoration projects that were completed in the past.

To help to overcome these challenges, satellite datasets provide a compelling option for monitoring vegetation outcomes after restoration. In particular, the USGS/NASA Landsat program has collected satellite imagery in a consistent manner since 1984, including indicators of vegetation health and productivity. To support the monitoring of riparian resources using satellites, Mark Hausner and colleagues at the Desert Research Institute and the University of California-Merced developed statistical tests that can now be applied to any location using the Climate Engine tool. 

About the datasets

See the video discussion of the datasets used in this analysis here.

The two datasets used in this procedure are Landsat and GridMET. Landsat is a satellite mission operated by the US Geological Survey and NASA that has been collecting imagery of the Earth’s surface in a consistent methodology since 1984 — this makes it one of the longest-running environmental datasets available. Specifically, we use the Landsat data to assess the normalized difference vegetation index (NDVI), which is a measure of vegetation health and productivity. Learn more about Landsat here.

GridMET is a climate reanalysis dataset that provides daily maps of meteorological variables such as precipitation, temperature, wind, aridity, and more. Here, we use GridMET to be able to control for the effects of precipitation in our analysis of vegetation productivity — that way, we can be confident that changes to vegetation production are not related to a change in precipitation.

Mapping Riparian Vegetation Condition using Landsat NDVI

See the video demonstration for this portion of the analysis here.

In this section, we simply map the current condition of NDVI for Suzie Creek during the growing season. To produce this map, follow the directions in the video to produce a map of NDVI from Landsat during the months of June, July, and August. If you are running this analysis for a riparian system that has a different growing season (such as in a different part of the world), use the months when vegetation is most active in your region instead of June, July, and August.

Mapping Riparian Vegetation Trends using Landsat NDVI

See the video demonstration for this portion of the analysis here.

To take our analysis a step further, we can use Climate Engine to map trends in NDVI through time, as an indicator of whether vegetation is becoming more or less productive in this area. In addition to simply mapping the trends in the data, Climate Engine offers the ability to apply a statistical test of significance to mask areas of the map where the trends don’t meet a user-defined confidence level. This test can be a good way of highlighting locations where there is a statistically significant trend in vegetation productivity, whether positive or negative.

Charting Riparian Vegetation Health Using Landsat NDVI

See the video demonstration for this portion of the analysis here.

To this point, our analysis has focused on producing maps of vegetation health, but in resource management contexts we are often interested in changes through time for a specific location. In this section of our analysis, we define a geography for our analysis and then Climate Engine does the work of producing a chart of how NDVI has changed through time for this stretch of Suzie Creek. In the chart, we can see that this stretch of Suzie Creek has had a notable increase in NDVI in the years since 2008 or so.

NOTE: Climate Engine now supports shapefile uploads as well, if you would like to analyze a feature in a shapefile.

Charting Riparian Vegetation Health and Precipitation Using Landsat and GridMET

See the video demonstration for this portion of the analysis here.

Changes to vegetation communities and productivity are dynamic and, for any given location, can be related to a variety of factors. One important factor in vegetation productivity is precipitation, especially in many parts of the arid western United States. For this reason, it is important to be able to control for precipitation to ensure that changes in vegetation productivity that can be seen in the chart are not related to an increase in precipitation. For this location, we do not see any trend in precipitation during the growing season, which can give us confidence that the increase in NDVI is likely related to the change in management.

Charting Pre- and Post-Restoration Riparian Vegetation Health Using Landsat and GridMET

See the video demonstration for this portion of the analysis here.

In addition to the time-series charting options available, Climate Engine also allows users to plot the relationship between two variables directly using a scatterplot. To complete our analysis, we will plot the relationship between precipitation and NDVI and then will plot the pre- and post- restoration periods separately. Initially, the scatterplot appears quite noisy, without a clear relationship between the two variables. However, after separating out the pre- and post-restoration periods, we can see that precipitation does influence vegetation productivity, but that this stretch of stream is simply much more productive during the period following the restoration actions. The final statistical test we can do is to plot the 95% prediction interval for the pre-restoration period, which shows that every year since 2010 has been more productive than could have been expected during the pre-restoration period.

Still need help? Contact Us Contact Us