Monitoring Forest Change

Worldwide, forests are being lost at an alarming rate due to natural disasters and human intervention. As important environments that support many ecosystems, the close monitoring and quantification of forest change is critical. With available data, Global Mapper can be used as a tool to explore and analyze forest change with respect to tree growth and loss.

Stands of trees and vegetation are represented in many commonly encountered data types. When considering aerial or satellite imagery, the presence of vegetation is obvious. From a remote perspective, forests and vegetated areas are clearly depicted.

Multi-band Image Analysis

Most often used as a base map for GIS projects, raster imagery can also provide valuable insight into the state and change of land cover. Using multiband satellite data collected worldwide through Landsat and other programs, image analysis methods in Global Mapper can create new layers of data depicting the land cover and vegetation health.

Each band of satellite imagery shows how a specific range of electromagnetic radiation interacts and reflects from the earth’s surface and land cover. A portion of the electromagnetic radiation used in this data collection is visible light, but bands of data extending beyond the visible spectrum are also used. Combining different bands of this data, characteristics of land cover can be visually enhanced.

The display of 4 different bands of data, each represented as a layer in Global Mapper, shows how each wavelength class of electromagnetic radiation provides slightly different information. This data is from August 2020, and from top to bottom, the bands are 2 (blue), 3 (green), 4 (red), and 5 (near-infrared).

A traditional true-color image combines the red, green, and blue bands of data to create an RGB image. This visualization is the most commonly encountered multiband image as it depicts how we see and interpret the surface and land cover. In this true-color image, vegetation appears green as expected, water is blue, and non-vegetated land is shown in brown and tan colors. In this example, data from two different years is shown, August 2013 and August 2020. Comparing only the true color images, it can be seen how the land cover and vegetation have changed over this study period.

The true color visualization shows the land cover in the expected colors.

Combining additional bands of data in different combinations, false-color images are created. The combination of near-infrared, red, and green creates a false-color composite image that shows vegetation in magenta. To clearly contrast the vegetated areas, bare earth, and water are shown in cyan hues.

This false-color visualization created by combining bands of data in a different way highlights vegetated areas in shades of magenta.

Calculating a Vegetation Index

Based on how different wavelengths of electromagnetic radiation interact with different types of land cover, standards for measuring vegetation, among other cover types, have been developed. A widely used measurement for vegetation presence and health is the Normalized Difference Vegetation Index (NDVI). Using the previously noted bands of Landsat data, NDVI is calculated using a built-in formula from Global Mapper’s Raster Calculator. This formula is NDVI = (NIR – Red) / (NIR + Red).The layers calculated with this method are handled similarly to elevation layers in Global Mapper. Instead of a color value attached to each pixel in the raster layer determining the value and display, the value associated with each pixel is the NDVI, and the visualization of these values can be altered by applying different shaders in Global Mapper.

Using the built-in NDVI formula for Landsat 8 data, new layers for 2013 and 2020 are individually calculated.

By default, the generated NDVI layers for 2013 and 2020 are shaded with the built-in NDVI shader. This shader covers the range of possible values in the NDVI, from -1 to +1, with the positive values generally representing vegetated land cover.

With the built-in NDVI shader applied, the darker green areas represent healthier vegetation, while the white and blue areas indicate no vegetation.

While this default shader provides a visual scale for the vegetation presence or greenness of the area, further classification of the NDVI values accentuates differences in the forest cover. A custom shader designed in the Global Mapper Configuration settings segments the NDVI images for 2013 and 2020 into more discrete categories.

This custom shader is applied to the NDVI layers for 2013 and 2020. Working primarily with the positive NDVI values, bare earth values range from 0 to 0.1, low vegetation and shrub coverage from 0.2 to 0.4, established vegetation from 0.4 to 0.5, and densely vegetated areas are represented by NDVI values above 0.5.

Created and edited in Configuration > Shader Options, custom shaders can be designed, allowing the user to select the colors and set the threshold values.

Creating and applying a custom shader helps to visually differentiate between the meaningful NDVI values.

Helping to affirm the threshold values used in this custom shader and subsequent analyses, the shaded NDVI layer for one year can be compared to the false-color composite image that highlights vegetation. By clipping the displayed NDVI values to those representing established vegetation that is likely forest (0.4 and above), it is seen how these values align with the brighter magenta areas indicating vegetation in the false-color image.

Looking at the 2013 data only, the NDVI values are clipped to only show values 0.4 and above. Global Mapper’s image swipe tool then provides a method for visually analyzing the difference between these two raster layers.

Raster Change Analysis

To this point, vegetation index values have been calculated and visually classified with a shader for a particular study area. A simple visual analysis infers that some degree of vegetation loss has occurred, and the general area of forest loss can be identified. To further identify the change in vegetation, these raster NDVI layers can be used in a difference calculation.

Raster Difference Grid

The creation of a difference grid involves subtracting the corresponding pixel values in one NDVI layer from the other to determine the degree of change on a pixel-by-pixel basis. This can be achieved in Global Mapper using the Combine/Compare Terrain tool, or by creating a custom formula in the Raster Calculator.

Using Global Mapper’s Raster Calculation tool, a custom formula, B1-B2, is added where B1 and B2 are variables that will be assigned to the two previously calculated NDVI layers. Setting the 2020 NDVI layer as B1 and the 2013 NDVI layer as B2, the difference calculation is set up as NDVI Difference = 2020 NDVI – 2013 NDVI.

Similar to the NDVI layers, Global Mapper handles the pixel values in the calculated single band difference layer-like elevation allowing the visualization of this layer to be customized. To better show the distribution and extent of NDVI change, a custom shader is created and applied to illustrate positive and negative changes in the area. Positive change, shown in green, indicates vegetation gain over the study period, and negative change, shown in brown, indicates vegetation loss.

The custom formula option in the Raster Calculator allows any custom formula to be used, adding versatility to this tool.

By altering the threshold values and blending setting for the custom shader, the degree of change or explicitly positive or negative change can be shown.

Vector Change Analysis

In order to transition from working with raster, pixel-based data to vector data, the Vectorization tool in Global Mapper Pro can be used to create polygon features outlining the forest areas identified by the calculated NDVI layers.

Vectorization in Global Mapper Pro is a one-step tool that uses the color, or elevation/slope pixel values within a layer to build polygons bounding areas of equal value or values that fall within a specified range. In this case, the NDVI values have been calculated and are managed like elevation values in Global Mapper, so the range of values indicating strong vegetation and likely forest areas can be extracted with this tool to create a layer of polygons. Based on the previously defined classification used to design the custom NDVI shader, values from each NDVI layer greater than 0.4 will be included in the polygon feature creation.

To more accurately retain all forest areas, the generated features are only slightly smoothed and simplified, and even the smallest polygons are kept.

Similar to the shaded NDVI layers, the change in forested areas can be seen by visually comparing the layers of vector features.

Spatial Operations

With vector features now generated representing the forested areas in 2013 and 2020, Global Mapper’s Spatial Operations tool can be used to find discrete areas of forest growth and loss over this period. The Spatial Operations tool contains many operations to analyze the spatial relationships between vector features in Global Mapper. A series of predicates execute spatial searches by selecting or copying features that meet a spatial relationship criterion. Additional operations create new layers of features segmenting existing features based on a specified relationship.Using the difference operation with the two polygon layers, polygons showing areas of forest loss and forest gain are generated. Showing patterns of change that are similar to the raster difference map, it is clear that forest and vegetation loss outweigh growth.

With operations to analyze many relationships between vector features, the difference operation is used to find the forest area gain and loss based on the polygons created with the Vectorize Raster tool.

With each layer styled in the Vector Layer Options, the forest gain and loss areas can be displayed together.

From each of these layers, land area measurements reflecting the total forest gain and forest loss can be calculated. A two-dimensional footprint measurement is calculated by displaying the feature measurement for each layer, but a more meaningful assessment of forest area change is the 3D surface area or ground area covered by the polygons.

To calculate 3D surface area, publicly available elevation data is added to the map display from Global Mapper’s Connect to Online Data tool, and the elevation and slope statistics are computed for each layer. This measurement tool, accessed from the Digitizer menu under the Analysis/Measurement sub-menu, populates new attributes for each feature in the layer with calculated statistics relating to slope, elevation, and area.

Using available terrain data and Global Mapper’s measurement tools, the total surface area of forest loss can be calculated.

Using the attribute statistics option for the calculated 3D_SURFACE_AREA attribute, the total surface area for all features in each layer is determined. Over this study period, 6.36 square miles of forest were gained in this area, and 144.74 square miles of forest were lost. This is a net negative change of 138.38 square miles.

WORK MADE EASY WITH GLOBAL MAPPER

Using Global Mapper to identify vegetation coverage and create additional data layers from available satellite collected data, a comprehensive forest change analysis is conducted. In the end, both raster and vector depictions of change are created with measurements of forest area gain and loss derived from the generated polygon features. This work on forest and change analysis can be taken further in Global Mapper by combining the areas of loss and gain with town and county boundary data, and by adding attributes to enhance the information held in the map. With the Map Layout Editor and Global Mapper Mobile, the new data generated in Global Mapper can be shared publicly or used by teams researching these issues in the field.

Want to try Global Mapper? Sign up for a 14-day free trial. You can also request a demo from one of our experts to see this workflow or other Global Mapper processing abilities.

Learn More

References:

Nugteren, Jacob. Monitoring deforestation & land cover change in the Santa Cruz region of Bolivia using Landsat satellite imagery. - Wur E-Depot Home. edepot.wur.nl/305297.

Taufik, Afirah & Syed Ahmad, Sharifah Sakinah & Ahmad, Asmala. (2016). Classification of Landsat 8 satellite data using NDVI thresholds. 8. 37-40.

Weier, John, & David Herring. “Measuring Vegetation (NDVI & EVI).” NASA, NASA, 30 Aug. 2000, earthobservatory.nasa.gov/features/MeasuringVegetation.