Skip to content

BLM Climate Engine - Riparian & Wetland

Alt text

Training Structure and Slides

The training was recorded and is available on the Climate Engine YouTube Channel. This training recording is embedded throughout this article with timestamped clips for each section. The slides from the training are available below.

Access the slides at this link.

Self-paced Study

This page mirrors the workshop agenda and demos. Each section starts with a timestamped, embedded clip from the full recording.

Sections in this training

Full Recording

Introduction & Session Overview

This thematic training focuses on using Climate Engine to monitor riparian and wetland resources with satellite-derived indicators and supporting climate/hydrology context. The session is designed to move quickly during delivery (because it’s recorded) and to support self-paced follow-up through linked examples and repeatable workflows.

After following this training you will have:

  • Gained an understanding of the core satellite and contextual climate and hydrology datasets that are most applicable for R&W decision support.
  • Learned workflows in the Climate Engine web application to analyze status and trends in R&W resources using satellite indicators.
  • Gained confidence applying satellite indicators to produce maps and figures for decision documents, resource monitoring activities, and other reporting processes.
  • Become able to access recordings and supporting materials for the session for asynchronous study and integration of materials.

Background and Overview of Climate Engine

This section emphasizes the role of this riparian/wetland training within the broader Climate Engine and BLM collaboration and demonstrates opportunities to use Climate Engine for decision-support:

  • Why Climate Engine exists: reduce technical barriers to accessing and analyzing large EO/climate archives in a consistent, scalable way.
  • Why this matters for BLM monitoring: agency policies emphasize measurable outcomes, thresholds, and standardized monitoring, and recognize the value of integrating remote sensing into monitoring strategies.
  • How remote sensing supports riparian/wetland decisions: 1) historical context (multi-decadal archives), 2) consistent methods across space, 3) multi-scale analysis (site → valley → basin), 4) frequent observations (sub-annual dynamics), 5) paired climate/hydrology context for interpretation.

New Functionality in Climate Engine

This section introduces three new features used heavily in the demos:

1) High-resolution basemaps (interpretation + QA/QC)

Used to visually interpret coarser satellite products and understand on-the-ground context.

  • NAIP: true color, false color, NDVI (high-resolution aerial imagery)
  • Historical imagery (where available)
  • USGS 3DEP: 1 m topographic derivatives (e.g., hillshade, slope)

Why it matters: many riparian questions involve narrow corridors and fine geomorphic controls; high resolution basemaps help users to interrogate and interpret the patterns 10–30 m satellite pixels are actually capturing.

2) Categorical raster overlays (context layers)

New “map overlay” layers provide categorical context such as:

  • National Landcover Database (land cover),
  • National Wetland Inventory (wetland inventory),
  • Surface Management Agency (ownership/management),
  • Valley Bottom Extraction Tool (topographic valley bottoms),
  • other thematic layers (e.g., GDE indicators in relevant regions).

3) Categorical masking (focus the analysis where it matters)

This new feature leverages the new categorical raster overlays as data masks, allowing users to subset their maps and time series to focus on:

  • valley bottoms (VBET),
  • wetland classes (NWI),
  • land ownership/management classes,
  • or combinations (e.g., VBET ∪ NWI, or BLM land only).

This makes “riparian-only” and “network-scale” analysis much more tractable and reduces the risk of diluting a signal by averaging across upland pixels.

Dataset Overview for Riparian & Wetland Monitoring

This section provides a practical overview of the core dataset “families” that tend to be most useful in riparian/wetland applications.

A) Extent / mapping layers (where is the riparian/wetland system?)

These layers are used to define analysis footprints and/or create masks.

  • NWI (National Wetlands Inventory)
    Strength: detailed wetland mapping where updates are recent/complete.
    Caveat: update age and mapping completeness vary widely by location; Climate Engine uses a simplified categorical representation of wetland types.

  • VBET (Valley Bottom Extraction Tool)
    Strength: geomorphic delineation of valley bottoms; useful for network-scale corridor analysis and focusing on connected valley-bottom environments.
    Caveats: 1) performance varies with DEM quality and hydrologic inputs and 2) often better in larger valley bottoms than headwater/spring-head contexts.

  • NLCD (National Land Cover Database)
    Strength: broad land cover context; includes wetland-related classes (e.g., woody wetlands, emergent herbaceous wetlands, open water).
    Caveat: 30 m resolution can be limiting for narrow riparian corridors, and class confusion can occur (e.g., wetlands vs pasture/hay in some western settings).

B) Climate and drought context (what are the drivers?)

The training revisits gridded and climate products (introduced in the pre-requisite “Climate Engine Basics” session in June 2025) with riparian-specific considerations:

  • For riparian/wetland questions, you often want to aggregate climate over meaningful windows (season, water year).
  • Standardized indices (e.g., SPI or related metrics) help compare across years and interpret hydrologic drought at longer time scales (often 9–12 months in water-limited systems).

C) Vegetation + water indicators (what is changing?)

In riparian/wetland monitoring, we focus on using spectral indices (e.g. NDVI, NDWI) from satellite imagery rather than upland-optimized derived products (e.g. RAP, RCMAP).

  • Greenness indices (e.g., NDVI)
    Proxy for photosynthetic activity/vigor; helpful for tracking riparian vegetation response.

  • Wetness / water indices (commonly labeled NDWI variants)
    Important for distinguishing vegetation change from surface water dynamics.

Common pitfalls and how to avoid them

  • Water can reduce NDVI: expanding ponds/open water can produce “declines” in NDVI that are not vegetation degradation.
    Remedy: interpret NDVI alongside wetness/water indices and NAIP.

  • Phenology timing differs: riparian systems can shift timing and magnitude year-to-year.
    Remedy: choose consistent windows (often summer/late summer) and verify with time series.

  • Spatial vs temporal trade-offs:

  • Landsat: longer record (back to mid-1980s) but coarser (30 m).
  • Sentinel-2: finer (10 m) but shorter record (generally since ~2017).

  • NAIP: very fine but infrequent (every 2–3 years, and timing varies).

  • Upland-derived products may fail in wetlands: some rangeland products (e.g., fractional cover/production models tuned for uplands) can perform poorly in riparian/wetland pixels. Use them cautiously and validate.

A worked example in this section illustrates how NDVI “declines” in a beaver meadow can reflect pond expansion/contraction rather than straightforward vegetation degradation—reinforcing the need for multi-layer interpretation.

BLM Practitioner Highlight (Wyatt Fereday)

This practitioner segment shows how Climate Engine is used in real BLM workflows to support monitoring, interpretation, and communication—often by pairing NDVI trends/time series with precipitation or groundwater context.

Three example patterns are highlighted:

1) Pre/post exclosure response at a spring
NDVI time series show a clear change after fencing, with precipitation plotted alongside to demonstrate that the observed improvement is not simply an effect of variable weather conditions.

2) Riparian decline plausibly consistent with groundwater drawdown
A riparian area showing strong NDVI decline is compared with nearby groundwater-level trends. The framing is intentionally cautious: the analysis is presented as correlation/screening evidence, not a definitive attribution study.

3) Evaluating potential effects of groundwater/geothermal development
NDVI is related to available discharge/monitoring information to extend context back in time and strengthen interpretation where field records are limited.

Demo 1: Maggie Creek (network-scale status and trend mapping)

This demo focuses on map-based analysis (rather than time series) and demonstrates how to create riparian network-scale products using the new masking tools.

Workflow highlights

  • Build a late-season NDVI map (often Jul 15–Sep 30) to emphasize water-limited conditions and riparian signal.
  • Use VBET to restrict analysis to low-lying and elevated valley bottoms, turning a broad NDVI map into a corridor-focused product.
  • Compute long-term change with slope of trend (example windows span multiple decades).
  • Apply a p-value/confidence mask to keep only statistically supported trends.
  • Use NAIP and 3DEP hillshade/slope to interpret where changes are occurring (channel position, corridor geomorphology, woody expansion, etc.).

Key idea

The VBET mask approach makes it much easier to create decision-ready valley-bottom maps without manually digitizing corridors or diluting the riparian signal with surrounding upland pixels.

Demo 2: Diamond Valley (groundwater impacts + climate context)

This demo shifts from mapping to time series + climate-adjusted inference in a groundwater-impacted wetland complex.

Step 1: Establish spatial context

  • Use historical imagery to see pre-expansion agricultural patterns.
  • Turn on Surface Management Agency to understand the management mosaic (private vs public/BLM).
  • Use National Wetland Inventory to identify wetland footprints to analyze.

Step 2: Identify where change is occurring (trend map)

  • Create a long-term Landsat NDVI trend map over a summer window.
  • Focus attention on wetland features showing coherent decline signals.

Step 3: Verify with high-resolution imagery

  • Use multi-year NAIP to visually confirm whether mapped wetlands show contraction or compositional change (recognizing NAIP timing differences between years).

Step 4: Quantify with masked time series

  • Make Graph → Summary time series
  • Draw a broad AOI polygon (precision is less important because the mask does the selection)
  • Mask by NWI wetland class (e.g., freshwater emergent wetlands)
  • Extract annual late-season NDVI summaries through time

Step 5: Add climate context (avoid over-attribution)

  • Run a two-variable analysis:
  • Variable 1: gridMET precipitation (water year) or an appropriate drought index
  • Variable 2: late-season NDVI
  • Use a two variable scatterplot (baseline vs impacted) to show whether the vegetation signal shifts downward even when precipitation is comparable.

Why this matters

In settings where field records are incomplete or where management/legal decisions require historical context, this workflow provides:

  • a consistent long-term record (satellite),
  • interpretability support (NAIP + overlays),
  • and more defensible inference by accounting for precipitation variability.

Demo 3: Rowland Spring (restoration effectiveness + AIM integration)

This capstone demo shows how Climate Engine can complement AIM field data as part of a multiple-lines-of-evidence workflow.

Site framing (why this is a good case study)

Rowland Spring is presented as a system with layered disturbance and management actions:

  • wildfire history,
  • spring-head exclosure (earlier intervention),
  • expanded fencing and restoration structures (recent intervention),
  • and targeted AIM sampling across multiple positions (inside/near/below exclosures).

What the AIM data contributes

AIM indicators provide detailed, on-the-ground condition information (e.g., cover, wetland-associated species signals), but are often limited to a small number of sampling years.

What Climate Engine adds

1) Historical and spatial context - Use NAIP + 3DEP to interpret channel geometry, access/road impacts, and spatial footprint of greenness.

2) Resolution trade-offs - Compare Sentinel-2 (10 m) to Landsat (30 m) for the same plot footprints to determine whether Landsat is acceptable for long-term context at a small site.

3) Pre/post inference with climate adjustment - Use a climate-adjusted scatterplot approach to assess whether post-intervention points shift outside the baseline relationship (with appropriate caution about causality).

4) Sampling-date comparability - Use dense time series (e.g., harmonized Landsat/Sentinel) to compare phenology and contextualize AIM sampling dates across years. - Download CSV outputs for custom plotting/analysis in R/Python when needed.

Takeaway

This section models a practical workflow for integrating:

  • field field (AIM) with
  • satellite and climate context (long-term + spatial continuity), to support more transparent interpretation and stronger documentation.

Closing, Contacts & Resources

The closing reinforces that the core workflows demonstrated are reproducible without code and are designed to support rapid exploration and decision-ready outputs.

Resources

Contacts

  • Eric Jensen eric.jensen@dri.edu
  • Alex Brooks alex.brooks@dri.edu