Remote Sensing of Above-Ground Biomass

Accurate measurement and mapping of biomass is a critical component of carbon stock quantification, climate change impact assessment, suitability and location of bio-energy processing plants, assessing fuel for forest fires, and assessing merchandisable timber. While above-ground biomass includes both live and dead plant material, most of the recent research effort on biomass estimation has focussed on the ‘live’ component (live trees) due to the prominence of this component. Accurate estimates of biomass is a prerequisite for better understanding of the impacts of deforestation and environmental degradation on climate change. The Intergovernmental Panel on Climate Change (IPCC) [1] has listed five terrestrial ecosystem carbon pools involving biomass: above-ground biomass, below-ground biomass, litter, woody debris and soil organic matter. Of these five, above-ground biomass is the most visible, dominant, dynamic and important pool of the terrestrial ecosystem, constituting around 30% of the total terrestrial ecosystem carbon pool. Above-ground biomass estimation, and especially forest biomass, has received considerable attention over the last few decades because of increased awareness of climate warming and the role forest biomass plays in carbon sequestration and release of greenhouse gases due to deforestation. Above-ground biomass estimates are the central basis for carbon inventories and most international negotiations in carbon trading schemes. Carbon trading markets require long-term information on carbon stocks, particularly on the above-ground ‘live’ biomass component as this is the most dynamic, changing and manipulable component of all the biomass pools. This is the ‘merchantable’ component of biomass. Above-ground forest biomass accounts for between 70% to 90% of total forest biomass [2]. While soil organic matter holds two to three times more carbon than biomass on a global scale, much of the soil carbon is more protected and not easily oxidised [3]. On the other hand, above-ground biomass is in a continuous state of flux due to fire, logging, storms, landuse changes, etc., and thus contributes to atmospheric carbon fluxes to a much greater extent and so is of much greater interest. Due to this dynamism of above-ground biomass, it is necessary to monitor it continuously and not measure once and forget. While detailed estimations of biomass is necessary for accurate carbon accounting, reliable estimation methods are few. Accurate estimates of stored carbon (biomass as dry weight is 50% carbon [4] and understanding sources and sinks can improve the accuracy of carbon flux models and thus lead to better projections of climate change and impacts. Initiatives such as Reducing Emissions from Deforestation and Forest Degradation (REDD) and REDD+ also rely heavily on above-ground biomass estimates [5,6]. REDD+ includes financing schemes and incentives with the aim of mitigating climate change by reducing deforestation and forest degradation through sustainable forest management and conservation, and enhancement of carbon stocks [7,8]. The countries that


Importance of Above-Ground Biomass
Accurate measurement and mapping of biomass is a critical component of carbon stock quantification, climate change impact assessment, suitability and location of bio-energy processing plants, assessing fuel for forest fires, and assessing merchandisable timber.While above-ground biomass includes both live and dead plant material, most of the recent research effort on biomass estimation has focussed on the 'live' component (live trees) due to the prominence of this component.Accurate estimates of biomass is a prerequisite for better understanding of the impacts of deforestation and environmental degradation on climate change.
The Intergovernmental Panel on Climate Change (IPCC) [1] has listed five terrestrial ecosystem carbon pools involving biomass: above-ground biomass, below-ground biomass, litter, woody debris and soil organic matter.Of these five, above-ground biomass is the most visible, dominant, dynamic and important pool of the terrestrial ecosystem, constituting around 30% of the total terrestrial ecosystem carbon pool.Above-ground biomass estimation, and especially forest biomass, has received considerable attention over the last few decades because of increased awareness of climate warming and the role forest biomass plays in carbon sequestration and release of greenhouse gases due to deforestation.
Above-ground biomass estimates are the central basis for carbon inventories and most international negotiations in carbon trading schemes.Carbon trading markets require long-term information on carbon stocks, particularly on the above-ground 'live' biomass component as this is the most dynamic, changing and manipulable component of all the biomass pools.This is the 'merchantable' component of biomass.
Above-ground forest biomass accounts for between 70% to 90% of total forest biomass [2].While soil organic matter holds two to three times more carbon than biomass on a global scale, much of the soil carbon is more protected and not easily oxidised [3].On the other hand, above-ground biomass is in a continuous state of flux due to fire, logging, storms, landuse changes, etc., and thus contributes to atmospheric carbon fluxes to a much greater extent and so is of much greater interest.Due to this dynamism of above-ground biomass, it is necessary to monitor it continuously and not measure once and forget.
While detailed estimations of biomass is necessary for accurate carbon accounting, reliable estimation methods are few.Accurate estimates of stored carbon (biomass as dry weight is 50% carbon [4] and understanding sources and sinks can improve the accuracy of carbon flux models and thus lead to better projections of climate change and impacts.Initiatives such as Reducing Emissions from Deforestation and Forest Degradation (REDD) and REDD+ also rely heavily on above-ground biomass estimates [5,6].REDD+ includes financing schemes and incentives with the aim of mitigating climate change by reducing deforestation and forest degradation through sustainable forest management and conservation, and enhancement of carbon stocks [7,8].The countries that participate in the REDD+ scheme have a requirement to produce accurate estimates of their forest carbon stocks and changes.
Wildfire and fuel management are becoming an increasingly important part of forest management.Forest biomass, and especially crown biomass and the dry litter component, are important factors in any fire model.Traditionally crown biomass was a neglected component as there was much greater emphasis on the commercial component of the trees but with fire playing a more important role in environmental planning, this biomass component has gained prominence.
Biomass is also a highly abundant source of energy that is widely used around the world.Its attractiveness is that it is a renewable fuel.However, biomass resources are distributed over wide geographical regions and their qualities for energy production varies over space and time.Also, very often the resources are located far away from the centres where energy generation is required.Because of this link between distribution over space and time and centres of requirements, it is important to have accurate and consistent means of measurement methods for biomass to evaluate feasibility of biomass for energy generation.

Methods of Assessing Above-Ground Biomass
Above-ground biomass can be measured or estimated both destructively and non-destructively.In the destructive method, sometimes also known as the harvest method, the trees are actually cut down and weighed.Sometimes a selected sample of trees are harvested and estimations for the whole population are based on these, especially where there is uniformity in tree size, for example a pine plantation.The destructive method of biomass estimation is limited to a small area due to the destructive nature, time, expense and labour involved.It is also not suitable where there may be threatened flora and fauna.
The non-destructive methods include the estimation based on allometric equations or through remote imagery.Allometric equations have been developed through the use of tree dimensions [9][10][11][12][13][14][15], such as diameter at breast height (dbh) and tree height, however these are not very useful in heterogenous forests.Allometric equations are most useful in uniform forests or plantations with similar aged stands.

Role of Remote Sensing in Mapping Above-Ground Biomass
While biomass derived from field data measurements is the most accurate, it is not a practical approach for broad-scale assessments.This is where Remote Sensing has a key advantage.It can provide data over large areas at a fraction of the cost associated with extensive sampling and enables access to inaccessible places.Data from Remote Sensing satellites are available at various scales, from local to global, and from a number of different platforms.There are also different types of data, such as optical, radar and LiDAR, with each one having certain advantages over the others [15].
Optical Remote Sensing probably provides the best alternative to biomass estimation through field sampling due to its global coverage, repetitiveness and cost-effectiveness.Optical Remote Sensing data is available from a number of platforms, such as IKONOS, Quickbird, Worldview, SPOT, Sentinel, Landsat and MODIS.The spatial resolutions vary from less than one metre to hundreds of metres.Optical Remote Sensing data has been used by numerous researchers for biomass estimation [16][17][18][19][20][21][22][23][24][25][26][27].
LiDAR is a relatively new technology that has found favour in biomass estimation.It has the ability to sample the vertical distribution of canopy and ground surfaces, providing detailed structural information about vegetation.This leads to more accurate estimations of basal area, crown size, tree height and stem volume.A number of studies have established strong correlations between LiDAR parameters and above-ground biomass [45][46][47][48][49][50][51][52][53][54][55].

Purpose of this Special Issue
Vegetation biomass plays a crucial role in understanding and monitoring ecosystem response and its contribution to the global carbon cycle.The recognition of forests as potential sinks of atmospheric carbon has resulted in numerous studies being conducted in estimating above-ground biomass or carbon stocks across varying scales.In addition, grassland biomass quantification is critical in understanding rangeland productivity as a resource for animal grazing.However, uncertainties in the Remote Sensing of AGB are high due to vegetation structural variations, heterogeneity of landscapes, seasonality and disproportionate data availability, among others.Recent developments in high resolution space-borne and air-borne satellite data have provided an opportunity to better estimate and map AGB across different spatial and temporal scales.The use of drones and UAVs has opened up avenues for super-fine resolution biomass estimation for targeted applications.Recent sensors, such as the Worldview series, now provide meter level spatial resolution while Sentinel and Landsat 8 provide free data for the whole world, opening up accessibility and more applications of Remote Sensing data, including for biomass estimation.
Remote sensing is a constantly evolving technology with new applications and methods being regularly introduced.This special issue was a call for the latest innovative methods and applications to map AGB at different scales.The range of topics included, but was not limited to, algorithm development and implementation, accuracy assessment, scaling issues (local-regional-global biomass mapping), integration of microwave (i.e., LiDAR) and optical sensors, forest biomass mapping, rangeland productivity and abundance (grass biomass, density, cover), bush encroachment biomass, seasonality and long term biomass monitoring, and climate change impacts and temporal monitoring.

Summary of Papers Published in this Special Issue
This special issue details results from a total of 15 papers that unpack the importance of Remote Sensing in biomass estimation and mapping across different spatial scales.The rich spectral, temporal and spatial information contained in satellite images has seen an improvement on productivity mapping under complex environmental conditions.Data sets used range from field spectrometers, multispectral and multi temporal images as well as microwave derived images.The data sets were combined with advanced machine learning algorithms and other state of the art processing techniques to reveal spatial and temporal biomass patterns.
A number of papers in this issue incorporated phenology in biomass estimation using Remote Sensing.Schucknecht et al. [56] used in-situ spatiotemporal biomass production and Remote Sensing to build a biomass model for the period 2001 to 2015.The phenology based seasonal NDVI was used as a proxy for biomass production and the model successfully predicted biomass at the end of the growing season.In a related study, NDVI Land Surface phenology derived from MODIS was used to model biomass in seasonal wetlands.The method was robust across different environmental conditions [57].The use of high spatial resolution images in estimating carbon stocks across seasons was evaluated across two seasons in an abandoned agricultural land [58].Pixel-scale vegetation indices derived from the dry season images yielded higher correlations with biomass than those derived from the wet season.The result confirms the saturation problem encountered using NDVI during biomass peak.Long term trends in biomass production are also critical in understanding the mechanism, direction and magnitude of climatic effects.Feng et al. [59] simulated potential productivity and actual productivity using Remote Sensing as well as climatic and anthropogenic data sets in Northern Tibetan Plateau between 1993 and 2011.Their results showed the importance of precipitation in regulating Net Primary Productivity.
Microwave Remote Sensing has also gained popularity in vegetation mapping in recent years.This issue presents three papers that used LiDAR and RADAR data to model vegetation biomass across different biomes.In a comparative study of modelling the biophysical attributes between deciduous and conifer forests, low density footprint LiDAR data successfully revealed spatial patterns of stem volume and total dry biomass [60].Vaglio Laurin et al. [61] tested the potential of ALOS2 and NDVI to estimate above ground biomass.The objective was to solve the saturation problem experienced at low to moderate biomass levels when using JERS and ALOS SAR.The integrated SAR and NDVI improved estimation of biomass, while only a slight saturation was experienced at higher forest biomass levels.Pasture biomass was also modelled by integrating optical and X-Band radar data in an open savannah woodland [62].The objective was to minimise the saturation problem experienced when using optical data as well as difficulties in separating dry matter from litter.TerraSAR-X (TSX) and image data from Landsat TM data from both dry and wet seasons could predict standing dry matter with high accuracy.
Optical multispectral data, with varying resolutions, was also used to estimate biomass.A multi-source satellite data approach was used to evaluate the potential of Remote Sensing in reducing biomass estimation error in an Alpine Meadow grassland [63].Results showed that filtered MODIS NDVI data reduces biomass estimation errors and the error increases with the increasing spatial scale of investigation.Addition of Laser Altimeter data to Landsat TM in a multivariate modelling framework improved forest biomass predictions in Northern China [64].Sibanda et al. [65] integrated texture metrics and the red edge derived indices from simulated Worldview-3 data to estimate biomass.Results showed the robustness of the model in reducing errors of prediction across all management treatments.
A total of 3 studies in this issue also applied hyperspectral data to improve biomass estimation.Jin et al. [66] used indices generated from the field spectroscopic data to calibrate the AquaCrop model for wheat yield and biomass estimation.In another study [67], fusion of Ultrasonic Sward Height and with narrow band normalized spectral index and simulated Worldview −2 data improved biomass prediction accuracy in grasslands with heterogeneous sward structure.Cheng et al. [68] showed the importance of dry matter indices in predicting biomass in canopy components of paddy rice as compared to chlorophyll indices, which saturate at high biomass levels.
Finally, this issue reports the utility of multi-angle data in improving biomass predictions.A study using multi angle CHRIS/PROBA data [69] showed that off nadir vegetation indices could predict forest biomass more accurately than the nadir derived indices.The result underscores the importance of Bidirectional Reflectance Distribution Function (BRDF) as a source of information than a source of uncertainty.
In summary, this Special Issue explored the role of Remote Sensing in estimating grassland, forest and woody biomass using a plethora of data and processing methods.Seasonality information was successfully built into biomass models with improved accuracies.The fusion of microwave and multispectral/hyperspectral data also reduced uncertainty errors in biomass estimation, especially in environments with complex canopy structure.Of critical importance is that the special issue highlighted methods and data sets that solves the problem of saturation in biomass estimation using the conventional vegetation indices [70].The issue provides a platform for day to day methods and approaches to operationalize Remote Sensing in vegetation productivity management.