13.1 Remote Sensing Modes for Vegetation Monitoring

Remote sensing includes any data that are collected without directly observing or coming in contact with the object or area of study. Remote sensing ranges from field-based methods (e.g., on site downward-looking photographs, ground-based multispectral vegetation sensors), to aerial (e.g., aerial photography, LiDAR, hyperspectral) and satellite (e.g., hyper-spatial/spectral/temporal) sensors.



Monitoring using remotely sensed information provides the ability to estimate ground conditions as well as characterize patterns across broad spatial and temporal scales. These spatial patterns may be direct estimates of indicators that would be measured in the field or may serve as proxies for various structural or biophysical properties of the landscape which influence many processes that are both ecologically important and provide societal benefits (i.e., ecosystem services). The application of remote sensing technologies within a monitoring program may significantly increase monitoring cost-effectiveness and significantly improve the ability to detect ecosystem change. But remote sensing also carries costs and has limitations that must be weighed in deciding whether there is sufficient benefit in applying the technology for monitoring.


Remote sensing prediction of cheatgrass cover in the Great Basin and Columbia Basin. From Boyte et al. (2016). https://doi.org/10.1016/j.rama.2016.03.002


The goal of these modules is to introduce some remote sensing concepts and provide examples of how remote sensing can be used in natural resource monitoring. There are entire classes dedicated to different remote sensing concepts, so it is beyond the scope of this course to try to dig deep into the mechanics of specific techniques. Remote sensing technologies and their application to natural resource monitoring also continue to develop rapidly. So rather than providing a comprehensive overview of remote sensing applications to natural resource monitoring, this set of online modules provides a general outline of how remote sensing can be incorporated into a natural resource monitoring program.


While the information in these modules does not require a lot of technical background, a working knowledge of remote sensing concepts is helpful. If you have not previously had an introduction to remote sensing, I would suggest that you look for one online. Check out the Additional Learning Resources module for some suggestions.


Learning Guide

Modes of Remote Sensing for Rangeland Monitoring


Remote sensing can be used in many different steps of designing and implementing a monitoring program including helping with sampling design and stratification, evaluating monitornig points, and tracking short-term changes. For this module, however, we will focus on using remote sensing to directly or indirectly measure monitoring indicators.


There are three main ways (modes) in which remote sensing can be used in rangeland monitoring.


1. Remote Sensing as a Replacement for Field Measurements


The first mode is to use remote-sensing technologies and products to replace field measurements.


Many research studies have shown that interpretation or classification of high-resolution imagery can match or outperform many field-based measurements of indicators like cover, frequency, or density.

Very-high resolution digital aerial images can be used to estimate vegetation cover (left), canopy size or density (center), or length and frequency (right). Image from Booth and Cox (2011), https://doi.org/10.2111/1551-501X-33.4.27


Land cover classifications or predictions of vegetation attributes (e.g., cover) have been used for landscape-scale rangeland assessment and monitoring

Malmstrom et al. (2009) estimated forage biomass over time from Landsat imagery to look at the effectiveness of management actions. Image from http://dx.doi.org/10.1111/j.1526-100X.2008.00411.x


Regression models or geostatisical techniques are used to predict rangeland indicators over landscapes from a set of field samples.

Regression kriging is a procedure that associates field data with satellite imagery to estimate ground conditions – shrub cover in this example. Example using methods from Karl (2010) http://dx.doi.org/10.2111/REM-D-09-00074.1

Homer et al. (2012) used a regression tree method to predict vegetation cover and bare ground over the state of Wyoming using Landsat imagery. Image from http://dx.doi.org/10.1016/j.jag.2011.09.012


Historic analysis of rangeland condition is also possible using archives of aerial photography or satellite imagery.

Rango et al. (2005) used historic aerial photographs to document the reestablishment of shrubs following a shrub-removal experiment in the 1930’s in southern New Mexico. From: https://doi.org/10.1016/j.jaridenv.2004.11.001.

2. Remote Sensing to Augment Field Measurements

The second mode of remote-sensing application is to supplement or augment field-based activities. For example, high-resolution aerial photographs ay be acquired at field sample locations and at nearby, similar sites for use in a double-sampling approach.


Another example of this mode of remote-sensing application is model-assisted inference which uses a statistical model developed between field measurements and remote-sensing products (e.g., NDVI) to improve indicator estimates from field data for larger areas. For example, Opsomer et al. (2007) found that estimates of forest amount and biomass from field surveys were better if satellite NDVI was used as a covariate in the estimates.


Estimates of forest attributes from field data were improved by using NDVI as a covariate in the analysis. Data from Opsomer et al. (2007), https://www.jstor.org/stable/27639872
Variable Estimated mean from field data alone Estimated standard error from field data alone Estimated mean with NDVI assist Estimated standard error with NDVI assist
Proportion of area that is forested 0.54 0.02 0.54 0.01
Forest biomass (tons/ac) 13.51 0.69 13.60 0.49


This mode of remote sensing can also be employed to make improved spatial predictions of indicators such as vegetation cover. Gu et al. (2013) used biomass models developed from MODIS NDVI to remove artifacts from rangeland productivity estimates due to administrative boundaries (i.e., state and county lines).

Gu et al. (2013) used regional maps of biomass from MODIS satellite data to correct/smooth out inconsistencies in NRCS forage production estimates. Image from: https://doi.org/10.1016/j.ecolind.2012.05.024


3. Use of Remote-Sensing-Specific Indicators

Finally, remote-sensing techniques can be used to generate new or synthetic indicators of rangelands that are difficult or impossible to characterize through field techniques. Examples of this mode are:


The differenced normalized burn ratio (dNBR) was developed from Landsat satellite imagery as a proxy for burn severity and has proven to be strongly correlated to ground measurements of burn severity.

Example of burn severity mapping. Image from Eidenshink et al. (2007), https://doi.org/10.4996/fireecology.0301003


Wylie et al. (2012) used NDVI from time series satellite imagery to define average vegetation performance for rangelands and identify regions that were “under-performing” or “over-performing” for a given year, indicating changes in rangeland condition that were due to disturbances or changes in management, and not annual precipitation variations.

Wylie et al.’s map of performance anomalies as a result of the Murphy Complex fire in Idaho in 2008. Image from https://doi.org/10.2111/REM-D-11-00058.1


Okin et al. (2009) looked at changes in vegetation patch size and distribution as an early-warning indicator of desertification in arid rangelands.

Okin et al’s (2009) conceptual diagram and example of how changes in vegetation patch size and distribution cause desertification in arid rangelands. Image from https://doi.org/10.1525/bio.2009.59.3.8


While the diversity of new indicators available from remote-sensing techniques is increasing, a challenge with this remote-sensing mode for rangeland monitoring, however, is translating the remote-sensing-derived indicators into statements of rangeland quality or health.