Bisrat, E., & Berhanu, B. (2018). Identification of Surface Water Storing Sites Using Topographic Wetness Index (TWI) and Normalized Difference Vegetation Index (NDVI). Journal of Natural Resources and Development, 8, 91–100.

            This study does not deal directly with northern peatlands but the research question posed and subsequent analysis likely have applications in the remote sensing of peatlands themselves and their associated properties. Bisrat et al. attempted to identify three unique water storing types (water bodies, wetlands, and land based) across Ethiopia using the Topographic Wetness Index (TWI) and Normalized Difference Vegetation Index (NDVI) derived from Landsat 8 OLI/TIRS imagery. While it is encouraging to note that the researchers were successful in creating a workflow that produced accurate, low-cost results for their study site, their methodologies for determining wetlands across a large area are of particular interest. Consistent with many boreal peatlands, wetlands in Ethiopia are typically topographic lowlands. Bisrat et al. incorporated a moderate resolution DEM in conjunction with their derived TWI and NDVI values to identify wetlands. Adaption of this technique for boreal peatlands specifically could yield expanded classification and identification of peatlands.

Blackmore, D. S. (2016). Use of Water Indices Derived from Landsat OLI Imagery and GIS to Estimate the Hydrologic Connectivity of Wetlands in the Tualatin River National Wildlife Refuge. Dissertations and Theses, paper 3205,

            This thesis paper presents a comparison of the Normalized Difference Water Index (NDWI) and the Modified Normalized Difference Water Index (MNDWI) in their effectiveness in identifying wetlands and estimating soil wetness. Both NDWI and MNDWI utilize the green light portion of the electromagnetic spectrum and infrared radiation. Differences arise in what portion of the infrared spectrum are used when calculating the index. NDWI utilizes near-infrared radiation while MNDWI utilizes mid-infrared radiation with MNDWI performing better when there are higher levels of ‘built-up’ land in a given study area. Blackmore found that both indices consistently correctly predicted open water features but NDWI seemed more sensitive to sites with increased vegetation while MNDWI showed increased detail when dry conditions were present. Soil wetness levels were validated in the field. Additionally, the author identified that perennial wetland connectivity was capable of being distinguished using Landsat derived water indices. While understanding open water within and surrounding northern peatlands is important when visualizing/mapping groundwater level, among other things, the understanding of wetland connectivity shows promise in management decision making. Northern peatlands, specifically those on the Kenai Peninsula, Alaska, have been shown to provide services nearby salmon bearing streams such as water temperature and flow modulation as well as nutrient subsidies that aid in basal stream productivity. Identification of connection paths between peatlands and nearby waterbodies using the described methodology could provide critical management information.

Campos, J., Silva, A., & Vidal-Torrado, P. (2012). Mapping, Organic Matter Mass and Water Volume of a Peatland in Serra do Espinhaço Meridional. Revista Brasileira de Ciencia do Solo, 36, 723-732.

            Introductory materials present in this study illustrate the importance of peatlands for not only their carbon storage potential but also for the role they play in the hydrologic functions of entire regions. Statistics presented in this research report that for the given watershed, peat covered roughly 12% of the land but stored a staggering 78% of the annual water excess for the watershed. Northern peatlands are expected to provide similar water storing potential. Peatlands in the study site were mapped and sampled on the ground along transects and georeferenced using handheld GPS units. Several key characteristics were described including vegetation type and coverage, soil density, peat depth, mineral material, moisture content, and state of decomposition. Highest levels of water volume and water expelled per sample were seen in the deepest parts of the peatland. While this study did not use a remote sensing approach, characteristics of peat that were measured are critical in understanding the state and quality of a given peatland. It is expected that potential remote sensing projects centered on northern peatlands will require ground truthing. Taking similar measurements as the ones in this study may allow researchers to both ground truth aerial imagery and relate certain characteristics to specific spectral signatures.

Gatis, N., Luscombe, D. J., Carless, D., Parry, L. E., Fyfe, R. M., Harrod, T. R., Brazler, R. E., & Anderson, K. (2018). Mapping Upland Peat Depth Using Airborne Radiometric and LiDAR Survey Data. Geoderma, 335, 78-87. https:/

            Authors of this study approach the quantification of peat volume as it relates to carbon storage and peatland management. The researchers acknowledge that traditional, direct methods for measuring peat depth i.e. probing and/or the use of ground penetrating radar are time and labor intensive. In response, a remote sensing approach is used. It is well documented that several topographic factors strongly influence peat deposition and accrual including elevation, slope, and wetness (in this study the Topographic Wetness Index is used to represent overall wetness). Using LiDAR derived topographic information in conjunction with ‘radiometric dose’ (calculated total air absorbed dose rates of thorium, uranium, and potassium) and ground truthed depths, researchers were able to use linear regression modeling in order to explain significance of different variables in observed peat depths. Radiometric dose was collected via airborne gamma-ray spectrometer and is noted to supply information regarding underlying bedrock as well as the overlaying soils (characteristics derived from this type of sensor include porosity, saturation, and density of soil as well as bedrock type). Results of the analysis showed similar extents of peatlands to previous mapping efforts with increased carbon stock estimates due to greater estimates of peat depth. This methodology is shows promising applications, especially for areas like the Kenai Peninsula where LiDAR data already exists. The cost of radiometric dose data acquisition was not noted in this paper and may be a limiting factor.

Guo, M., Li, J., Sheng, C., Xu, J., & Wu, L. (2017). A Review of Wetland Remote Sensing. Sensors, 17 (777).

            This study provides a comprehensive review of the effectiveness of different sensor types regarding their abilities to detect various components of wetland function. Wetland functions that were identified as viable factors to be sensed in some capacity via remote sensors included: landcover classification, inundation/flood level, biodiversity, biomass/carbon stock assessment, water quality, leaf chemistry, soil identification, disturbance level, elevation, and sea level rise among other more obscure applications. Through the review process, the authors came to differing conclusions regarding best use practices associated with data of different resolutions. Aerial photographs were best used to classify wetlands, identify species, and verify other classifications. Coarse resolution data (examples given were from MODIS and AVHRR) were best used at large scale to identify wetland areas and in combination with higher resolution data for more comprehensive investigations. Medium resolutions data, such as Landsat imagery, were employed the most in wetland remote sensing applications and have value when attempting to classify wetlands, calculate inundation levels, map habitats, and estimate biomass. High resolution imagery was found to be more case specific, lending itself to smaller study sites where high accuracy classification is necessary. Hyperspectral imagery was found to be useful when attempting to identify species based on unique spectral signatures and when studying vegetation chlorophyll and nutrient concentration. The authors note that radar and LiDAR both have applications in areas with dense vegetation with LiDAR also providing detailed topographic data. This review is extremely helpful in the planning stage of any remote sensing project centered on wetlands of any kind.  

Harris, A., Charnock, R., & Lucas, R. M. (2015). Hyperspectral Remote Sensing of Peatland Floristic Gradients. Remote Sensing of Environment, 162, 99-111.

            This study describes a novel approach to mapping floristic gradients in peatlands using hyperspectral imagery and regression modeling. Both species level and plant functional types were mapped in this way. Isometric feature mapping was used to describe plant communities within the peatlands as well as identify the floristic gradients that would be identified in the hyperspectral imagery. Advantages of this methodology center on its’ ability to visualize differing vegetation stands without having to identify unique spectral signatures for each plant species. In the regression analysis, researchers showed that the levels of agreement (as represented by r squared values) were proportional to the floristic variation present in a given vegetation stand. The authors acknowledge that while detailed hyperspectral imagery is becoming more widely available, it is still fairly difficult to acquire and time consuming/expensive to process. With this in mind, the authors advise that any sensor capable of adequately characterizing the red, red-edge, and near infrared portions of the electromagnetic spectrum will most likely work in similar applications. With price drops in unmanned aerial vehicles (UAV) as well as capable sensor payloads, similar methodology may be applicable using UAV on the Kenai Peninsula at smaller scales.

Krankina, O. N., Pflugmacher, D., Friedl, M., Cohen, W. B., Nelson, P., & Baccini, A. (2008). Meeting the Challenge of Mapping Peatlands with Remotely Sensed Data. Biogeosciences, 5, 1809-1820.

            This study attempts to address some of the main issues that arise from remotely sensing peatlands with coarse resolution datasets as well as provide guidance in future attempts to quantify peatland coverage over large areas. Introductory passages in this paper touch on the role of northern peatlands in the global carbon cycle as well as the importance of accurately quantifying this habitat. Central to this research is the usage of hyperspectral, multiresolution MODIS data. Krankina et al. identified that peatlands are largely correctly identified using MODIS data but are consistently underestimated due to the coarseness of the data used. This trend was consistent across differing coarse resolution datasets. Inclusion of field-sourced data and identification of distinct spectral signatures across time were found to greatly increase the accuracy of remotely sensed classification using MODIS data specifically. Techniques and conclusions from this study should translate to higher resolution data sources and result in even greater accuracy in peatland classification and identification.      

Lovitt, J., Rahman, M. M., & McDermid, G. J. (2017). Assessing the Value of UAV Photogrammetry for Characterizing Terrain in Complex Peatlands. Remote Sensing, 9 (715).

            Introductory material presented in this paper reflects on the effectiveness of unmanned aerial vehicles (UAVs) ability to capture detailed microtopographic data across a variety of terrestrial settings. The authors also describe several of the shortcomings associated with UAV photogrammetry including limitations associated with environmental factors such as wind as well as known limitations presented by variations in vegetation canopy cover. This study aims to correct for limitations of UAV photogrammetry by introducing LiDAR data into the workflow. UAV imagery was collected in both the morning and evening to correct for opposing shadow angles. Dense point clouds were derived from the UAV imagery using Agisoft software. A two-way mixed model ANOVA test was conducted on the UAV photogrammetry, LiDAR, and UAV + LiDAR datasets to model accuracy with ground survey sites acting as reference data. Results of the analysis indicate that photogrammetric data acquired via UAV is comparable to LiDAR data when mapping microtopography in all but the most heavily vegetated areas. The researchers also found that the incorporation of both LiDAR and UAV photogrammetric data did not produce significantly better topographic maps than one or the other on their own. This study shows that photogrammetric data acquired via UAV is a viable option in collecting high resolution topographic information for study sites where LiDAR data is not present.  

McPartland, M. Y., Falkowski, M. J., Reinhardt, J. R., Kane, E. S., Kolka, R., Turetsky, M. R., Douglas, T. A., Anderson, J., Edwards, J. D., Palik, B., & Montgomery, R. A. (2019). Characterizing Boreal Peatland Plant Composition and Species Diversity with Hyperspectral Remote Sensing. Remote Sensing, 11 (1685).

            This study set out to answer several questions using hyperspectral imagery and treatment plots within two boreal peatlands, one located in central Alaska and the other located in northern Minnesota. Main research questions posed in this study were centered on the changes to peatland reflectance values under differing potential climatic situations. Between the two study sites, treatments included raised water level, lowered water level, control water level, and differing levels of carbon dioxide and temperature. For each different treatment type, researchers manually collected vegetation coverage data as well as gathered spectral signatures. High resolution airborne hyperspectral imagery was also collected at each site. Analysis consisted of regression analyses for vegetation composition and diversity, principal component analyses for spectral data, and supervised random forest classification of the hyperspectral imagery. A focus between the different analyses was characterization of plant functional types. Results showed several trends. Lowered water tables dramatically increased litter coverage, spectral reflectance was influenced to some degree by all treatments, and random forests supervised classification was highly successful. Conclusions the authors drew from this varied study include the effectiveness of hyperspectral remote sensing in understanding various effects/stressors of peatlands that are associated with climate change and that these techniques can most likely be applied to small and large study areas. The authors recognize that boreal peatlands will disproportionately feel the effects of a changing climate and note that these methodologies may prove invaluable in the tracking and quantification of shifting ecosystems.

Mohanty, B. P., Cosh, M. H., Lakshmi, V., & Montzka, C. (2017). Soil Moisture Remote Sensing: State-of-the-Science. Valdose Zone Journal, 16 (1),

            This study reviews passive microwave remote sensing techniques and applications for mapping and monitoring soil moisture. Datasets used in this review focus on L band, low frequency passive radiometric data. Common sensor capable of detecting these wavelengths include SMOS, AQUARIUS, and SMAP. The data provided by these sensors is capable of monitoring near surface (up to 5 centimeters in depth) soil moisture at 35-kilometer resolution at intervals of 2-3 days. The authors note that the SMAP sensor also produces soil moisture products that extend to 1 meter in depth at a 9-kilometer resolution on a weekly basis using compiled data. While data in these frequencies are often rough, the authors note that they are sufficient when examining phenomenon associated with land-use, land-change, agricultural trends, landscape level hydrology, and meteorology. Applications in remote sensing of northern peatlands include monitoring of drying exacerbated by climate change and hydrologic flux patterns.

Rahman, M. M., McDermid, G. J., Strack, M., & Lovitt, J. (2017). A New Method to Map Groundwater Table in Peatlands Using Unmanned Aerial Vehicles. Remote Sensing, 9 (1057).

            Researchers in this study attempted to map ground water level in a northern peatland using a novel unmanned aerial vehicle (UAV) approach. The authors note that ground water influences a myriad of peatland functions including rates of peat accumulation and net gas exchange. Imagery for the study was collected via UAV and post-processed using Agisoft software. Open water from the imagery was identified using ENVI software. The researchers detail how open water present in a peatland generally reflects water table levels across that specific peatland. Topographic information for the study site was created via dense point clouds derived from the UAV imagery and verified against existing LiDAR data. An interpolated surface representing ground water level was calculated for the study area using identified elevations of open water scattered throughout the reach. This surface was then validated against 31 water well locations distributed through the study area. The resulting ground water level model showed accuracy to within 20 centimeters of the well data. The authors note that increased identification and incorporation of open water sites will only aid in creating more accurate ground water level surfaces. They also note that this methodology has potential to scale over larger areas but is limited by the availability of identifiable open water. The understanding of ground water level across northern peatlands has important implications as climate change increases peatland drying and changes hydrologic regimes.

Senthikumar, M., Gnanasundar, D., & Arumugam, R. (2019). Identifying Groundwater Recharge Zones Using Remote Sensing & GIS Techniques in Amaravathi Aquifer System, Tamil Nadu, South India. Sustainable Environment Research, 29(15),

            While this study examined a groundwater recharge on a landscape scale, there are applications for identification of northern peatlands. Researchers used multiple data layers (geology, slope, geomorphology, land cover, drainage, water bodies, weathered zones, and depth to groundwater) in a GIS environment to identify groundwater recharge zones. The study area was classified into zones identified as suitable for groundwater recharge using an indexed overlay model with values ranging from 1-100. The results were then further classified into four subcategories of likelihood of being a recharge structure (very high, high, moderate, and poor). Several conclusions are significant and translatable to northern peatland studies. First, wetlands were identified as high ranking sites for groundwater recharge. Second, porous soils allow for the most percolation and collection-as-groundwater for precipitation. Northern peatlands often meet both of these criteria. Noting this, there is high potential that peatlands are prime areas for groundwater recharge adding increased ecosystem services to this ecosystem type. Methodologies laid out in this study could be used in identifying which peatlands are more likely to support groundwater recharge.

Shekhawat, N., Pran Dadhich, A., & Goyal, R. (2018). Temporal Analysis of Land Surface Temperature and Land Use/ Land Cover using Remote Sensing. Journal of Civil Engineering and Environmental Technology, 5(2), 71–77.

While this study focused on land use cover change, their methodology and data sources have potential uses in peatland science. Researchers calculated change over time using Landsat TM, ETM+, and Landsat 8 between the years of 2000 and 2016. Indices used in this study included the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Water Index (NDWI), and the Normalized Difference Bareness Index (NDBaI) with NDVI and NDWI having known applications in wetland remote sensing. Methodology, namely the transformation of Landsat data and calculation of indices, is translatable to any area of study that uses this data source or indices. Additionally, conclusions drawn regarding the relationship between land surface temperature and the respective influence it has on the chosen indices may inform how climate change could influence peatland drying in boreal environments.

Walter, M. & Mondal, P. (2019). A Rapidly Assessed Wetland Stress Index (RAWSI) Using           Landsat 8 and Sentinel-1 Radar Data. Remote Sensing, 11 (2549).            https://doi.10.3390/rs11212549

In this study, Walter and Mondal developed a novel stress indicator deemed ‘Rapidly Assessed Wetlands Stress Index’ (RAWSI) using a combination of machine-learning classifications, the normalized difference vegetation index (NDVI), and a proxy for human alterations to local hydrology. While this study does not focus on peatlands specifically, data and methodologies used could potentially transfer to northern peatlands. Land cover classification utilized Landsat 8 multispectral imagery in conjunction with Sentinel-1 synthetic aperture radar (especially useful for its’ ability to penetrate dense foliage and soil) data. The authors used two classification methods, support vector machine and random forest, both of which are based on machine-learning algorithms. The authors additionally calculated NDVI and stream channelization (a proxy for human impact) for the study area. All of these initial analyses were combined to create an overall index for wetland stress on a pixel-by-pixel basis termed RAWSI. Hotspot analysis performed in the RAWSI indexed land cover showed that forested wetlands were disproportionately impacted in the study area. While not all of the factors put forward in this study will impact northern peatlands, similar methodology could be used to create a stress index. A boreal peatland stress index would be extremely useful in management decision making.

Weissert, L. F., Disney, M. (2013). Carbon Storage in Peatlands: A Case Study on the Isle      of Man. Geoderma, 4(16), 111-119.

Background information provided in this study stresses the importance of sphagnum sp. in their role in peat accumulation and peatland function in general. This study attempted to map peat area, depth, and sphagnum coverage. Geospatial data used included multispectral aerial imagery and a digital elevation model. It was hypothesized that higher sphagnum coverage would indicate higher quality peat. Supervised classification was performed on the imagery and ground sites were sampled for peat depth. Peat depth was found to be positively correlated with sphagnum coverage, but slope and curvature calculated from the DEM did not have statistically significant influence on peat depth. It was found that supervised classification of peatlands correctly identified peatlands at a high accuracy (99.59%) but levels of omission were somewhat high (30.83%). Information from this study will be useful in creating and executing a remotely sensed estimation of peatland coverage in unmapped portions of the Kenai Peninsula. It is also important to understand expected accuracies when performing this type of analysis.

Print Friendly, PDF & Email