G. 2094

Page 1

Global Research journal of Natural Science

& Technology (GRJNST)

Volume: 04 - Issue 3 (2026), 2094

ISSN P: 2790-7643 ISSN E: 2790-7651

www.grjnst.net

https://doi.org/10.53762/grjnst.04.03.15

Remote Sensing–Based Assessment of Forest Carbon Sequestration and Land

Use Change Dynamics under Climate Variability

Received: 30 March 2026. Accepted: 01 May 2026. Published: 30 May 2026

Muhammad Hassan Ali (Corresponding Author)

Department of Forestry and Range Management,

Shaheed Benazir Bhutto University of Veterinary and Animal Sciences,

Sarkrand, Sindh, Pakistan

soilscience1070@gmail.com

Fateh Ali Chohan

Department of Forestry and Range Management,

Shaheed Benazir Bhutto University of Veterinary and Animal Sciences,

Sarkrand, Sindh, Pakistan

fateh.ali0335@gmail.com

Asad ullah

Department of Forestry and Range Management, Shaheed Benazir Bhutto University of

Veterinary and Animal Sciences, Sarkrand, Sindh, Pakistan

saad.janii.94@gmail.com

Kalsoom

Department of English literature University of Sindh Jamshoro, Sindh, Pakistan

mahachohan554@gmail.com

Muhammad Ismail Chohan

Department of Forestry and Range Management,

Shaheed Benazir Bhutto University of Veterinary and Animal Sciences,

Sarkrand, Sindh, Pakistan

imail40@gmail.com

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

Commons Attribution 4.0 International (CC BY 4.0) license, which permits unrestricted use and distribution

G. 2094

Page 2

Walliullah Lagahri

Department of Forestry and Range Management,

Shaheed Benazir Bhutto University of Veterinary and Animal Sciences,

Sarkrand, Sindh, Pakistan

raiswaliullah@gmail.com

Muhammad Afzal

Department of Forestry and Range Management,

Shaheed Benazir Bhutto University of Veterinary and Animal Sciences,

Sarkrand, Sindh, Pakistan

afzalshar66@gmail.com

Abstract:

Forest ecosystems are among the most significant terrestrial carbon reservoirs

and play a crucial role in regulating the global carbon cycle through carbon

sequestration. However, rapid land use and land cover change (LULC),

combined with increasing climate variability, and has significantly altered the

stability and efficiency of these ecosystems. This review paper examines the

application of remote sensing technologies in assessing forest carbon

sequestration and monitoring land use dynamics under changing climatic

conditions. The study highlights how urbanization, agricultural expansion,

deforestation, and wildfire activities contribute to carbon emissions and

ecosystem degradation across tropical, coastal, semi-arid, and urban forest

landscapes. Special emphasis is placed on the integration of optical, LiDAR,

and Synthetic Aperture Radar (SAR) systems for estimating above-ground

biomass (AGB), soil organic carbon (SOC), and vegetation structural

characteristics. The review further explores the role of vegetation indices, multi-

sensor fusion, and machine learning algorithms such as Random Forest (RF),

Support Vector Machines (SVM), XGBoost, and Convolutional Neural

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 3

Networks (CNNs) in improving biomass estimation accuracy and carbon

accounting. Additionally, the paper discusses the influence of climatic variables,

including temperature rise, altered precipitation regimes, droughts, and

wildfires, on forest carbon sequestration potential and ecosystem resilience.

Emerging technologies such as ESA Biomass, NISAR missions, and Digital

Twin Earth systems are also evaluated for their transformative potential in

global forest monitoring and carbon management. The findings suggest that

integrating advanced remote sensing with artificial intelligence and climate

modeling provides an effective framework for sustainable forest management,

climate mitigation strategies, and accurate carbon monitoring at regional and

global scales.

Keywords: Remote Sensing, Forest Carbon Sequestration, Land Use Change,

Climate Variability, Above-Ground Biomass, LiDAR, SAR, Carbon

Accounting, Machine Learning, Forest Monitoring

Introduction

The terrestrial biosphere constitutes one of the most significant and dynamic

components of the global carbon cycle, with forest ecosystems alone storing

approximately 45% of all terrestrial carbon. These ecosystems function as critical

natural sinks, sequestering an estimated 3.5 Pg C yr⁻¹ from the atmosphere, thereby

playing a pivotal role in mitigating the progression of anthropogenic climate change (Xu,

2025). However, the stability and efficacy of this sequestration capacity are currently

under unprecedented pressure from the twin drivers of land use and land cover change

(LULC) and intensifying climate variability. The transformation of natural landscapes

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 4

driven by urbanization, agricultural expansion, and industrial resource exploitation

accounts for nearly 25% of human-caused greenhouse gas (GHG) emissions

(Merrikhpour et al., 2025). Concurrently, the increasing frequency of extreme climatic

events, including droughts and wildfires, has begun to compromise the historical

resilience of these carbon reservoirs, potentially shifting them from net sinks to net

sources of atmospheric carbon (Xu et al., 2021).

In this context, remote sensing (RS) technologies have emerged as indispensable tools

for the monitoring and assessment of forest carbon dynamics at multiple spatial and

temporal scales. Since the launch of Landsat-1 in 1972, Earth observation has evolved

from simple land cover classification to sophisticated, multi-sensor frameworks capable

of quantifying three-dimensional forest structure, above-ground biomass (AGB), and soil

organic carbon (SOC) (Wulder et al., 2019). Modern assessments now integrate high-

resolution optical imagery, active microwave radar, and laser-scanning systems to provide

a comprehensive, spatially explicit understanding of how LULC and climate variability

interact to shape the future of terrestrial carbon sequestration (H. Nguyen et al., 2019).

Spatiotemporal Dynamics of Land Use Change and Carbon Flux

Land use and land cover change represents the second-largest source of global CO₂

emissions, trailing only the combustion of fossil fuels. Between 2001 and 2023, the

global forest system was a net sink of −5.5 ± 8.1 Gt CO₂e yr⁻¹, a figure that obscures

the volatile balance between −14.5 ± 7.7 Gt CO₂e yr⁻¹ of carbon removals and 9.0 ±

2.7 Gt CO₂e yr⁻¹ of gross emissions. The magnitude of these fluxes is heavily influenced

by regional development trajectories and the specific nature of land transitions (Lu et al.,

2014).

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 5

Urbanization and Peri-Urban Carbon Depletion

Rapid urban growth, particularly in developing metropolitan basins, catalyzes significant

environmental transformations that often lead to a net increase in carbon emissions

(Ashaolu et al., 2019). In the Kağıthane basin of Istanbul, for example, the expansion of

residential and industrial areas has been driven by a 16.59% population increase over the

past decade. Analysis using Sentinel-1 and Sentinel-2 data processed on the Google

Earth Engine (GEE) platform reveals that while some vegetation growth is observed in

managed urban green spaces, there is a systemic decline in natural forest cover and barren

lands. This shift is projected to increase regional carbon emissions by up to 13%

between 2035 and 2095 (Kocaman & Ağaçcıoğlu, 2025).

Beyond direct biomass loss, urbanization modifies local hydrological dynamics,

including peak discharge patterns and surface runoff, which indirectly affects the health

and carbon uptake of surrounding vegetation (Ashaolu et al., 2019). The conversion of

natural land cover to built-up environments alters the surface energy balance, often

leading to urban heat island (UHI) effects that can be accurately mapped using thermal

infrared remote sensing. These thermal anomalies exacerbate physiological stress on

remaining urban trees, potentially reducing their long-term carbon sequestration

potential (Yang, 2025).

Agricultural Expansion and Coastal Ecosystem Vulnerability

Tropical coastal ecosystems represent some of the most carbon-dense environments on

Earth, yet they are increasingly threatened by the expansion of industrial agriculture. In

Phang Nga Bay, southern Thailand, extensive land use transitions between 2000 and

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 6

2020 were primarily driven by the replacement of evergreen and para rubber forests with

oil palm plantations (Aratrakorn et al., 2006). This conversion led to a net carbon

storage loss of approximately

occurring in upland forested areas between

2014).

, with the most intensive degradation

and meters in elevation (Alongi,

The role of "blue carbon" ecosystems, specifically mangroves, is particularly critical in

these coastal landscapes. Despite covering only about one-fifth of the total area in Phang

Nga Bay, mangroves consistently contributed over

of the regional carbon storage.

This highlights a critical spatial disparity: while agricultural encroachment causes

widespread low-intensity carbon loss across upland areas, the localized destruction of

mangrove or swamp forests results in massive, immediate carbon releases (Murphy, Hall,

& Jintana, 2020). Integrating the InVEST (Integrated Valuation of Ecosystem Services

and Tradeoffs) model with high-resolution satellite data has allowed researchers to

demonstrate that localized gains in mangrove sequestration can occasionally offset some

conversion losses, provided that targeted conservation strategies are implemented (Sharp

et al., 2018).

Table 1. Remote Sensing–Based Evaluation of Land Use Transitions and Associated

Carbon Dynamics

Land

Use

Transition Net

Carbon Primary

RS Key Drivers

Category

Impact

Monitoring

(Relative)

Strategy

Forest to Urban

High

Loss Optical (Sentinel- Pop.

Growth,

(Direct

+ 2) + Thermal Infrastructure

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 7

Indirect)

Forest to Oil Palm

High

Loss

Biomass SAR (Sentinel-1) Commercial

+ GEE Agriculture

Coastal

Forest

to Critical Carbon LiDAR + Multi- Food

Security,

Aquaculture

Release

spectral

Economics

Reforestation/Afforestation Gradual

Accumulation

Time-series

Carbon Credits,

Policy

(NDVI/EVI)

Forest to Grassland/Pasture Moderate-High

Loss

Landsat

GFC Livestock, Small-

scale farming

(Hansen et al.)

Long-term Variations in Tropical and Semi-Arid Forests

In regions like Southeast Vietnam, the period between 1990 and 2020 saw significant

fluctuations in carbon services due to ecological degradation and urban expansion.

While ecological succession and forest restoration efforts partially compensated for

some losses, the combined anthropogenic impact outweighed the natural recovery

capacity, leading to a net decline in total carbon storage. This trend is echoed globally,

with tropical deforestation particularly in Brazil and Indonesia accounting for nearly half

of the global forest loss due to land conversion.

In contrast, India's semi-arid and dry forests, located in regions like Rajasthan and

Gujarat, present a different challenge for carbon sequestration assessment. These

ecosystems possess lower biomass accumulation and carbon storage capacity compared

to humid tropical forests due to water scarcity and high temperatures. However, they

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 8

often contain resilient below-ground carbon pools that contribute significantly to soil

organic carbon (SOC) under extreme climatic conditions. Remote sensing of these

drylands requires a shift in focus from canopy greenness to structural parameters and

litter deposition rates, as fast-growing plantation species like Teak or Eucalyptus are

increasingly used for reforestation on wastelands.

Remote Sensing Technologies for Carbon Accounting

The evolution of remote sensing has provided a multi-layered framework for assessing

forest carbon, moving from broad land cover mapping to the precise quantification of

individual tree components (Pendleton et al., 2012)

Optical Remote Sensing and Spectral Limitations

Optical sensors, including Landsat 8/9 and Sentinel-2, remain the foundation of global

monitoring due to their consistent spectral libraries and high revisit frequencies. These

platforms allow for the calculation of vegetation indices such as the normalized

difference vegetation index (NDVI), the enhanced vegetation index (EVI), and the

photochemical reflectance index (PRI). These indices serve as proxies for canopy

greenness, leaf area index (LAI), and photosynthetic vigor, which are empirically linked

to above-ground biomass (AGB) (Li et al., 2021).

However, a fundamental challenge with optical remote sensing is the "saturation" effect.

In high-biomass tropical forests, the spectral response of vegetation indices often reaches

a plateau once the canopy is fully closed, typically at biomass levels between 150 and

200 Mg ha⁻¹. This makes it difficult to distinguish between moderately dense and very

old-growth forests using optical data alone (Sinha et al., 2019). To mitigate this, newer

missions such as Sentinel-2 utilize red-edge spectral bands, which have shown higher

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 9

sensitivity to chlorophyll content and nitrogen status, thereby improving AGB and soil

organic carbon (SOC) estimations even in dense canopies (Jagadish et al., 2024).

Active Systems: LiDAR and SAR

Active remote sensing technologies light detection and ranging (LiDAR) and synthetic

aperture radar (SAR) have revolutionized biomass estimation by providing information

on the vertical and three-dimensional structure of forests (Choi, 2024). LiDAR systems

use laser pulses to measure the distance between the sensor and the Earth's surface,

generating high-resolution "point clouds" that represent the forest's vertical profile.

Airborne laser scanning (ALS) is currently considered the most accurate technology for

extracting major forest attributes such as tree height, canopy density, and volume for

above-ground biomass (AGB) estimation. On a finer scale, terrestrial laser scanning

(TLS) and close-range sensing from unmanned aerial vehicles (UAVs) allow for the

measurement of diameter at breast height (DBH) and trunk shape with centimeter-level

precision (Xu et al., 2023). A meta-analysis of close-range sensing accuracy indicates

that ground-based LiDAR remains the "gold standard" for single-tree and plot-level

assessments, though UAV-based systems are more efficient for stand-scale analysis

(Fayad et al., 2016).

SAR systems, on the other hand, use microwave energy to penetrate cloud cover and,

depending on the wavelength, the forest canopy itself. The sensitivity of SAR to biomass

is primarily a function of its wavelength:

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 10

 X-band and C-band (e.g., Sentinel-1): These shorter wavelengths interact

primarily with leaves and small branches in the upper canopy. They are effective

for forest cover change detection but saturate at relatively low biomass levels.

 L-band (e.g., ALOS PALSAR-2, NISAR): This longer wavelength penetrates

deeper into the canopy, interacting with larger branches and trunks. It provides a

more robust proxy for AGB and is less prone to saturation than optical or C-

band SAR (Le Toan et al., 2024).

 P-band (e.g., ESA Biomass mission): With a wavelength of approximately 70 cm,

P-band SAR can penetrate the entire canopy to interact with the main trunks and

the ground surface, making it the most effective tool for measuring high-biomass

forests up to 500 Mg ha⁻¹ (Ho Tong Minh et al., 2016).

Multi-Sensor Fusion and Advanced Modeling

The most accurate forest carbon assessments are increasingly based on the "fusion" of

multi-source data. By integrating optical spectral data with 3D structural information

from LiDAR or SAR, researchers can overcome the limitations of individual sensors.

For example, combining Sentinel-2 spectral bands with Digital Elevation Models

(DEMs) and LiDAR-derived canopy heights has been shown to significantly improve

the accuracy of machine learning models for AGB prediction (Ali, 2025).

This fusion approach is particularly relevant for meeting the Measurement, Reporting,

and Verification (MRV) standards required by international climate frameworks such as

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 11

the UNFCCC and REDD+. Integrated models can leverage the broad spatial coverage

of satellite imagery and the high-precision "truth" provided by field plots and airborne

LiDAR to generate wall-to-wall maps of carbon stocks and fluxes (Santos, 2025).

Table 2. Comparison of Remote Sensing Sensor Types for Forest Structural and

Biomass Assessment

Sensor Type Specific

Structural Variable Accuracy

Primary

Platform/Band

Measured

Range

(R2)

Limitation

Optical

LiDAR

Sentinel-2 (Red- Canopy

0.55

0.82

– Spectral

Saturation

edge)

Greenness/LAI

ALS (Airborne)

Tree

0.85

0.95

– High

Height/Canopy

Profile

Acquisition

Cost

SAR

L-band

Large

0.60

0.80

– Signal

De-

(ALOS/NISAR)

Branches/Trunks

correlation

P-band (Biomass)

Main Stem/Trunk 0.75

Volume 0.90

– Ionospheric

Interference

Passive

SMOS/SMAP

(VOD)

Vegetation Optical 0.50

Depth 0.70

– Coarse Spatial

Microwave

Res ( >9 km

)

Climate Variability and its Impact on Sequestration Potential

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 12

Climate variability exerts a profound influence on the physiological processes and spatial

distribution of forest vegetation, thereby altering its carbon sequestration potential

(CSP). CSP is defined as the difference between an ecosystem's maximum carbon

carrying capacity (CCC) and its actual carbon stock, representing the potential for future

carbon uptake (Xu, 2025).

Temperature and Precipitation Thresholds

Ongoing climate changes, characterized by shifting temperature regimes and altered

precipitation patterns, can severely impact forest structure and function. In the mountain

ecosystems of Yunnan Province, China, simulation models have shown that the

suitability of forest habitats is primarily limited by the minimum temperature of the

coldest month (TMW) and total seasonal precipitation (PRS). For example, a 1 °C

increase in temperature combined with a 20\% decrease in precipitation could reduce

the potential distribution area of major forest types by 12.41\% (Iheaturu, 2026).

Interestingly, the combined effect of increased temperature and decreased precipitation

can, in some specific cases, increase the CSP of certain forest types by accelerating

biomass turnover, though this is often a transient response. Generally, however, frequent

and severe drought events such as the "flash droughts" observed in southeastern

Australia or the record-low water levels in the Amazon River in 2023 pose a significant

threat to forest health. These events cause unusually rapid drying of soil and vegetation,

leading to widespread tree mortality and reduced carbon uptake (Tian et al., 2023).

Biogeophysical and Biogeochemical Feedback Loops

The interaction between forests and the climate system occurs through two primary

feedback pathways: biogeochemical (BGC) and biogeophysical (BGP). The

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 13

biogeochemical pathway involves the exchange of greenhouse gases, primarily CO₂,

between the forest and the atmosphere. Deforestation and forest degradation release

stored carbon, increasing atmospheric CO₂ concentrations and contributing to global

warming (Boysen et al., 2020). Earth system models (ESMs) estimate that historical

land use and land cover change (LULC)-induced carbon emissions have resulted in a

global warming of approximately 0.21 ± 0.14 °C. The biogeophysical pathway involves

changes in the physical characteristics of the land surface, such as albedo, surface

roughness, and evapotranspiration (ET) (Amali et al., 2024).

Albedo: Converting dark, absorbent forests to reflective pastures or snow-covered

grasslands increases surface albedo, reflecting more solar radiation back into space. This

has a cooling effect, particularly in high-latitude boreal regions (Jiao et al., 2023).

Evapotranspiration: Forests are highly efficient at transferring water from the soil to the

atmosphere through ET, a process that cools the local environment. In tropical regions,

the loss of ET due to deforestation generally leads to significant local warming that can

overcompensate for any albedo-related cooling (Boysen et al., 2020).

Surface Roughness: Forests have high surface roughness, which promotes atmospheric

turbulence and efficient heat transfer (Prestele et al., 2016). Reducing this roughness

through land clearing can lead to higher temperature gradients at the surface.

The net effect of forest land use change is often a delicate balance between these two

pathways. On a global scale, the BGC temperature effects historically dominate the BGP

effects, meaning that the overall impact of historical land use change has been to warm

the climate. However, at regional scales, particularly in the tropics and high latitudes, the

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 14

BGP effects can be equal in magnitude to the BGC effects, highlighting the need for

spatially explicit climate mitigation strategies (Amali et al., 2024).

Table 3. Biogeochemical and Biogeophysical Feedback Pathways Associated with Land

Use Change

Feedback Pathway Mechanism

Climate

Impact Climate

Impact

(Global)

(Regional)

Biogeochemical

(BGC)

CO_{2}

Consistent

Warming

Loss

Varies with biomass

from density

Release/Storage

Biogeophysical

(BGP)

Albedo Shift

Negligible to Slight Cooling in Boreal

Cooling

(High Albedo)

Biogeophysical

(BGP)

Evapotranspiration

Surface Roughness

Slight

Warming Intense Warming in

Tropics

from Loss

Biogeophysical

(BGP)

Slight

Warming Localized Heat Flux

change

from Loss

The Wildfire-Climate Feedback Loop

Climate change is increasingly driving a dangerous "two-way street" relationship with

land use through the intensification of wildfires. Warming temperatures and longer

periods of drought create hotter, drier conditions that escalate fire risk. Emissions from

forest fires have increased by 60\% globally since 2001, with fire now accounting for

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 15

one-third of all land cover change in some regions (Jiao et al., 2023). This creates a self-

reinforcing feedback loop: climate change fuels fires, which release vast quantities of

CO_2, which in turn accelerates further warming. Boreal and humid tropical regions,

which contain the world's last great tracts of natural forest, have seen the most dramatic

increases in fire-driven forest loss, with a strong correlation r^2 = 0.85 in tropical

regions) between global temperature anomalies and fire-induced forest loss (Papucci,

2026).

Machine Learning Frameworks for Biomass Estimation

The transition from traditional forest inventories to remote sensing-based assessment

has been accelerated by the application of advanced machine learning (ML) and deep

learning (DL) algorithms. These methods are particularly effective at capturing the non-

linear and high-dimensional relationships between remote sensing features and forest

carbon stocks (Fayad et al., 2016).

Algorithm Selection and Performance Metrics

A wide range of machine learning (ML) algorithms is currently utilized in the literature,

with random forest (RF), support vector machines (SVM), and extreme gradient

boosting (XGBoost) being the most prominent (Donato et al., 2011).

figure:2 Characteristic extraction" leads to the first concentric arc segment, "Problem

Characteristic Space".

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 16

 Random Forest (RF): RF is the most frequently used algorithm, appearing in

approximately 88% of recent studies. It is an ensemble method that constructs

multiple decision trees and averages their predictions, making it robust against

outliers and noisy predictors. In Xinjiang, China, the RF model combined with

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 17

topographic and meteorological data significantly improved above-ground

biomass

(AGB)

estimation

accuracy

across

diverse

forest

types,

achieving

values greater than 0.65 (Cohen & Goward, 2004).

 XGBoost: While RF is common, XGBoost has recently shown superior

performance in approximately 75% of the studies where it was directly compared

with other methods. In a study of Larix principis-rupprechtii plantations in

northern China, XGBoost achieved an

(RMSE) of 0.73 Mg ha⁻¹ using Sentinel-2 data, outperforming both SVM

) and RF ( ). XGBoost's success is attributed to its efficient

of 0.82 and a root mean square error

(

handling of non-linearities and its regularization parameters that prevent

overfitting (Lü et al., 2023).

 Deep Learning (DL): Convolutional neural networks (CNNs) are increasingly

applied to extract textural and structural features from high-resolution imagery

(e.g., UAV-RGB or PlanetScope). DL models can reduce biomass estimation

errors by 5% to 20% compared to traditional regression methods, particularly in

complex, heterogeneous stands (Xu, 2025).

Feature Selection and Variable Importance

The performance of ML models is highly dependent on the selection of characteristic

variables. Feature selection algorithms, such as the Boruta algorithm or the Least

Absolute Shrinkage and Selection Operator (LASSO), are used to identify the most

relevant predictors from a pool of spectral bands, vegetation indices, and texture features

(Yazar et al., 2023).

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 18

In large-scale provincial models, climate (e.g., mean annual temperature, precipitation),

topography (e.g., slope, elevation), and texture factors often emerge as more significant

than individual spectral bands. For instance, adding Digital Elevation Model (DEM)

data to optical imagery frequently makes the DEM the most important predictor

variable, as it accounts for the environmental gradients that govern tree growth and

biomass accumulation (Amali et al., 2024).

Table 4. Performance Comparison of Machine Learning and Process-Based Models in

Carbon and Biomass Assessment

Model

Type

Best

Performance Primary Data Source Key Feature for Accuracy

(R2)

XGBoost

0.82

Sentinel-2

Red-edge + Vegetation

Indices

RF

0.75

UAV-RGB/LiDAR Texture + Height metrics

SVM

CNN

0.79

Landsat-9

NIR and SWIR bands

0.85 – 0.98

High-res

Imagery

Drone Spatial

pattern/Segmentation

InVEST

N/A

(Process- LULC + Biomass Carbon pool coefficients

maps

based)

The 2025 Technological Frontier: Digital Twins and Next-Gen Missions

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 19

The field of forest carbon monitoring is currently entering a transformative phase

characterized by the launch of dedicated biomass satellites and the operationalization of

"Digital Twin" Earth systems.

The ESA Biomass and NISAR Missions

The year 2025 marks the launch of two critical radar missions designed to quantify

global forest structure with unprecedented precision (Orlov et al., 2024).

ESA BIOMASS Mission (scheduled launch: April 29, 2025): This mission will utilize a

P-band synthetic aperture radar (SAR) to deliver global maps of forest biomass and tree

height every seven months. Its unique wavelength will allow it to "see" through dense

tropical canopies to measure the primary woody biomass, achieving a projected canopy

height root mean square error (RMSE) of 1–2 m and an above-ground biomass (AGB)

RMSE of 15–25 Mg ha⁻¹ (Prestele et al., 2016).

NASA-ISRO NISAR Mission (scheduled launch: July 30, 2025): NISAR is a dual-

frequency (L-band and S-band) SAR mission. The L-band is highly sensitive to large

branches and trunks, while the S-band is responsive to upper canopy foliage. The

integration of these two frequencies will extend the dynamic range of biomass retrieval

and improve coherence stability for long-term monitoring (Prestele et al., 2016).

Destination Earth (DestinE) and Forest Digital Twins

The European Commission’s Destination Earth (DestinE) initiative is building a highly

accurate digital replica of the Earth system, powered by high-performance computing

(HPC) and AI. A flagship component of this system is the Forest Digital Twin (Forest

DTC), led by organizations like VTT and ECMWF.

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 20

The Forest DTC operates at a 10 -meter resolution, primarily utilizing Sentinel-2 data

to provide a comprehensive description of the forest subsystem. It integrates several

specialized models:

 PRELES: A light use efficiency model that outputs Gross Primary Production

(GPP), ET, and Net Ecosystem Exchange (NEE) using daily weather data.

 CROBAS: A tree growth model that uses stand variables (species, density, DBH)

to simulate biomass and litterfall.

 YASSO15: A soil carbon model that estimates soil respiration and carbon

accumulation based on litterfall and climate (Jiao et al., 2021).

This "living digital twin" allows users to run "what-if" scenarios, testing the impact of

different climate pathways and forest management strategies (e.g., selective logging vs.

afforestation) on future carbon sequestration. Such systems are revolutionary for carbon

markets, as they provide a transparent, investible, and auditable ledger of forest health

that surpasses traditional multi-year verification cycles (Boysen et al., 2020)

Methodological Challenges and Strategic Implications

Despite these technological leaps, several persistent challenges continue to limit the

absolute accuracy of remote sensing-based carbon assessments (Amali et al., 2024).

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 21

Figure 1. Comparison of ground-based traditional inventories and remote sensing

estimation methods within the forest carbon accounting framework.

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 22

Scale Inversion and Ground-Truthing

A significant trade-off exists between spatial scale and estimation accuracy. While

ground-based LiDAR offers sub-centimeter precision at the single-tree level, its accuracy

tends to diminish when scaled up to plot-level or stand-level analyses due to cumulative

errors in single-tree segmentation and the interconversion of variables like DBH and

height. Furthermore, there is an "R&D gap" in ground-truthing for diverse ecological

zones; most high-precision allometric models are developed for temperate or plantation

forests, leaving significant uncertainties when applied to natural tropical or semi-arid

ecosystems (Amali et al., 2024).

Non-Permanence Risks and Policy Alignment

For forest carbon projects to remain viable in global markets, the risk of "non-

permanence" the potential for sequestered carbon to be re-released due to wildfire,

drought, or illegal logging must be addressed. Remote sensing allows for the detailed

analysis of multi-year datasets to spot patterns in precipitation and temperature, helping

developers identify whether a location is experiencing increasing drought stress or

shifting toward conditions unsuitable for long-term forest growth (Orlov et al., 2024).

Additionally, there is a need to align satellite-based flux estimates with National

Greenhouse Gas Inventories (NGHGIs) used under the Paris Agreement. Current

bottom-up models and atmospheric top-down observations often show a gap in

estimated anthropogenic land use fluxes, primarily due to differing definitions of

"natural" versus "anthropogenic" forest land. Reconciling these differences using Earth

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 23

observation-based frameworks like Global Forest Watch is essential for building global

confidence in forest carbon accounting (Lu et al., 2014).

Conclusions

Remote sensing–based assessment of forest carbon sequestration and land use change

dynamics has become an essential approach for understanding the interaction between

terrestrial ecosystems and climate change. The review demonstrates that forests function

as major carbon sinks, yet their sequestration capacity is increasingly threatened by

urbanization, agricultural expansion, deforestation, droughts, and wildfire disturbances.

Land use and land cover changes significantly alter ecosystem structure and contribute to

greenhouse gas emissions, thereby intensifying global warming and ecological instability.

Advanced remote sensing technologies, including optical sensors, LiDAR, and SAR

systems, have greatly enhanced the ability to monitor forest structure, biomass

distribution, and carbon dynamics across multiple spatial and temporal scales. The

integration of multi-sensor data with machine learning and deep learning algorithms has

further improved the accuracy of above-ground biomass estimation and carbon stock

mapping. Moreover, climate variability strongly influences forest productivity,

evapotranspiration, and carbon sequestration potential, creating complex biogeophysical

and biogeochemical feedback mechanisms within the Earth system. Emerging

technologies such as ESA Biomass, NISAR, and Forest Digital Twin systems represent a

major advancement in global carbon monitoring and sustainable forest management.

Despite these developments, challenges related to scale, sensor limitations, ground

validation, and non-permanence risks remain significant. The study concludes that

combining remote sensing, artificial intelligence, ecological modeling, and policy-driven

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 24

conservation strategies is essential for effective carbon accounting, climate mitigation,

and long-term ecosystem sustainability under changing environmental conditions.

References

Cohen, W. B., & Goward, S. N. (2004). Landsat's role in ecological applications of

remote sensing. BioScience, 54(6), 535. https://doi.org/10.1641/0006-

3568(2004)054[0535:lrieao]2.0.co;2

H. Nguyen, T., Jones, S., Soto-Berelov, M., & Haywood, A. (2019). Landsat time-series

for estimating forest aboveground biomass and its dynamics across space and

time:

A

review.

Remote

Sensing,

12(1),

98.

https://doi.org/10.3390/rs12010098

Lu, D., Chen, Q., Wang, G., Liu, L., Li, G., & Moran, E. (2014). A survey of remote

sensing-based aboveground biomass estimation methods in forest ecosystems.

International

Journal

of

Digital

Earth,

9(1),

63-105.

https://doi.org/10.1080/17538947.2014.990526

Lü, F., Song, Y., & Yan, X. (2023). Evaluating carbon sink potential of forest

ecosystems under different climate change scenarios in Yunnan, Southwest China.

Remote Sensing, 15(5), 1442. https://doi.org/10.3390/rs15051442

Merrikhpour, H., Badamfirooz, J., Zarandian, A., & Mousazadeh, R. (2025). Assessing

carbon storage and economic valuation of stocked carbon function based on land

use/land cover using the InVEST model. Environmental Health Engineering and

Management Journal, 12(4).

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 25

Wulder, M. A., Loveland, T. R., Roy, D. P., Crawford, C. J., Masek, J. G., Woodcock,

C. E., ... & Lymburner, L. (2019). Current status of Landsat program, science,

and applications. Remote Sensing of Environment, 225, 127-147.

https://doi.org/10.1016/j.rse.2019.02.015

Xu, L., Saatchi, S. S., Yang, Y., Yu, Y., Pongratz, J., Bloom, A. A., ... & Schimel, D.

(2021). Changes in global terrestrial live biomass over the 21st century. Science

Advances, 7(27), eabe9829. https://doi.org/10.1126/sciadv.abe9829

Xu, X. (2025). Biochar application for soil carbon sequestration and greenhouse gas

mitigation in forest ecosystems: A bibliometric analysis using CiteSpace. MDPI.

Ashaolu, E. D., Olorunfemi, J. F., & Ifabiyi, I. P. (2019). Assessing the spatio-temporal

pattern of land use and land cover changes in Osun drainage basin, Nigeria.

Journal

of

Environmental

Geography,

12(1-2),

41–50.

https://doi.org/10.2478/jengeo-2019-0005

Kocaman, B., & Ağaçcıoğlu, H. (2025). Assessment of the impact of land use/land

cover changes on carbon emissions using remote sensing and deep learning: A

case study of the Kağıthane basin. Sustainability, 17(23), 10690.

https://doi.org/10.3390/su172310690

Yang, H. R. (2025). Exploring the complex relationships between influential factors of

urban land development patterns and urban thermal environment: A study on

downtown

Shanghai.

Sustainability,

17(19),

8547.

https://doi.org/10.3390/su17198547

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 26

Yazar, M., Daloglu Cetinkaya, I., Iban, M. C., & Bilgilioglu, S. S. (2023). The green

divide and heat exposure: Urban transformation projects in Istanbul. Frontiers in

Environmental

Science,

11,

Article

1265332.

https://doi.org/10.3389/fenvs.2023.1265332

Alongi, D. M. (2014). Carbon cycling and storage in mangrove forests. Annual Review

of Marine Science, 6, 195-219.

Aratrakorn, S., Thunhikorn, S., & Donald, P. F. (2006). Changes in bird communities

following conversion of lowland forest to oil palm and rubber plantations in

southern Thailand. Bird Conservation International, 16(1), 71-82.

Donato, D. C., Kauffman, J. B., Murdiyarso, D., Kurnianto, S., Stidham, M., &

Kanninen, M. (2011). Mangroves among the most carbon-rich forests in the

tropics. Nature Geoscience, 4(5), 293-297.

Murphy, D. J., Hall, J., & Jintana, V. (2020). Land use change and carbon losses in

Southeast Asian tropical peatlands. Wetlands Ecology and Management, 28(3),

457-472.

Pendleton, L., Donato, D. C., Murray, B. C., Crooks, S., Jenkins, W. A., Sifleet, S., ... &

Baldera, A. (2012). Estimating global “blue carbon” emissions from conversion

and degradation of vegetated coastal ecosystems. PLoS ONE, 7(9), e43542.

Sharp, R., Tallis, H. T., Ricketts, T., Guerry, A. D., Wood, S. A., Chaplin-Kramer, R.,

... & Vogl, A. L. (2018). InVEST 3.7.0 user’s guide. The Natural Capital

Project, Stanford University.

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 27

Fayad, I., Baghdadi, N., Guitet, S., Bailly, J.-S., Hérault, B., Gond, V., El Hajj, M., &

Tong Minh, D. H. (2016). Aboveground biomass mapping in French Guiana by

combining remote sensing, forest inventories and environmental data.

International Journal of Applied Earth Observation and Geoinformation, 52,

502–514. https://doi.org/10.1016/j.jag.2016.07.015

Jagadish, B., Behera, M. D., Prakash, A. J., Paramanik, S., Ghosh, S. M., Patnaik, C., &

Das, A. (2024). Indicating saturation limits of multi-sensor satellite data in

estimating aboveground biomass of a mangrove forest. Journal of the Indian

Society of Remote Sensing, 52, 2483–2500. https://doi.org/10.1007/s12524-

024-01968-1

Li, C., Zhou, L., & Xu, W. (2021). Estimating aboveground biomass using Sentinel-2

MSI data and ensemble algorithms for grassland in the Shengjin Lake Wetland,

China. Remote Sensing, 13(8), 1595. https://doi.org/10.3390/rs13081595

Sinha, S., Mohan, Brun, S., Das, A. K., Sharma, L. K., Jeganathan, C., Santra, A., Santra

Mitra, S., & Nathawat, M. S. (2019). Multi-sensor approach integrating optical

and multi-frequency synthetic aperture radar for carbon stock estimation over a

tropical deciduous forest in India. Carbon Management, 11(1), 39–55.

https://doi.org/10.1080/17583004.2019.1686931

Choi, C. (2024). Combining TanDEM-X Interferometric SAR and GEDI Lidar

Measurements for Improving Forest Height, Structure and Biomass Estimates.

Deutsches Zentrum für Luft- und Raumfahrt (DLR).

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 28

Ho Tong Minh, D., Le Toan, T., Rocca, F., Tebaldini, S., Villard, L., Réjou-Méchain,

M., Phillips, O. L., Feldpausch, T. R., Dubois-Fernandez, P., Scipal, K., & Chave,

J. (2016). SAR tomography for the retrieval of forest biomass and height: Cross-

validation at two tropical forest sites in French Guiana. Remote Sensing of

Environment, 175, 138-147. https://doi.org/10.1016/j.rse.2015.12.037

Le Toan, T., Villard, L., Ho Tong Minh, D., Doblas, J., Mermoz, S., Ferro-Famil, L.,

Koleck, T., Bouvet, A., Planells, M., & Polidori, L. (2024). Tackling high

biomass in tropical forests through the BIOMASS mission. The International

Archives of the Photogrammetry, Remote Sensing and Spatial Information

Sciences, XLVIII-3-2024, 287-293. https://doi.org/10.5194/isprs-archives-

xlviii-3-2024-287-2024

Xu, Y., Lu, Z., Bürgmann, R., Hensley, S., Fielding, E., & Kim, J. (2023). P-band SAR

for ground deformation surveying: Advantages and challenges. Remote Sensing of

Environment, 287, 113474. https://doi.org/10.1016/j.rse.2023.113474

Ali, J. (2025). Remote sensing and integration of machine learning algorithms for

above-ground biomass estimation in Larix principis-rupprechtii Mayr

plantations: a case study using Sentinel-2 and Landsat-9 data in northern China.

Frontiers in Environmental Science.

Iheaturu, C. J. (2026). Remote sensing of tropical forest recovery. Remote Sensing of

Environment, 335, 115257.

Jiao, Y., Wang, B., Yao, X., Wang, S., Chi, T., & Meng, Y. (2023). Forest emissions

reduction assessment using optical satellite imagery and space LiDAR fusion for

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15

G. 2094

Page 29

carbon

stock

estimation.

Remote

Sensing,

15(5),

1410.

https://doi.org/10.3390/rs15051410

Papucci, E. (2026). Review of forest biomass assessments based on remote sensing

reveals progress in methodological quality but major challenges remain. Forestry.

Santos, D. T. D. (2025). Multi-sensor data fusion for operational monitoring of

deforestation and forest carbon stocks in tropical ecosystems. International

Journal of Science, Architecture, Technology, and Environment, 2(12).

Tian, L., Wu, X., Tao, Y., Li, M., Qian, C., Liao, L., & Fu, W. (2023). Review of

remote sensing-based methods for forest aboveground biomass estimation:

Progress,

challenges,

and

prospects.

Forests,

14(6),

1086.

https://doi.org/10.3390/f14061086

Xu, X. (2025). Estimation of forest aboveground biomass using Sentinel-1/2

synergized with extrapolated parameters from LiDAR data and analysis of its

ecological driving factors. Remote Sensing.

GRJNST, Volume: 04 - Issue 3 (2026) / ISSN P: 2790-7643

Article ID: 2094

https://doi.org/10.53762/grjnst.04.03.15