Attributing the impacts of ecological engineering and climate change on carbon uptake in Northeastern China

Abstract

Context
In the past decades, several ecological engineering (eco-engineering) programs have been conducted in China, leading to a significant increase in regional carbon sink. However, the contribution of different eco-engineering programs to carbon uptake is still not clear, as the location of different programs is difficult to identify, and their impacts are concurrent with climate change.

Objectives
We aim to detect the location of eco-engineering programs and attribute the impacts of eco-engineering and climate change on vegetation dynamics and carbon uptake in Northeastern China during 2000–2020.

Methods
We developed a new framework to detect the location of eco-engineering programs by combining a temporal pattern analysis method and Markov model, and to attribute the impacts of eco-engineering and climate change on vegetation greenness and carbon uptake by combining a neighbor contrast method within a sliding window and trend analysis on the normalized difference vegetation index (NDVI) and gross primary production (GPP).

Results
We identified four main forestry eco-engineering programs: croplands to forest (CtoF), grasslands to forest (GtoF), savannas to forest (StoF), and natural forest conservation (NFC) programs, whose areas accounted for 2.11%, 1.89%, 3.41%, and 1.72% of the total study area, respectively. Both eco-engineering and climate change contributed to the increase in greenness and carbon uptake. Compared to climate change effect, eco-engineering increased NDVI and GPP by 121% and 21.43% on average, respectively. Specifically, the eco-engineering-induced increases in GPP were 54.1%, 9.46%, 8.13%, and 24.20% for CtoF, GtoF, StoF, and NFC, respectively.

Conclusions
These findings highlight the important and direct contribution of eco-engineering on vegetation greening with positive effects on carbon sequestration at a fine scale, providing an important implication for eco-engineering planning and management towards a carbon-neutral future.

Full text

Vol.: (0123456789)
1 3
Landsc Ecol
https://doi.org/10.1007/s10980-023-01679-x
RESEARCH ARTICLE
Attributing theimpacts ofecological engineering
andclimate change oncarbon uptake inNortheastern
China
HuidongLi· WanjingGao· YageLiu· FenghuiYuan· MinchaoWu·
LinMeng
Received: 10 October 2022 / Accepted: 30 April 2023
© The Author(s), under exclusive licence to Springer Nature B.V. 2023
Abstract
Context In the past decades, several ecological
engineering (eco-engineering) programs have been
conducted in China, leading to a significant increase
in regional carbon sink. However, the contribution of
different eco-engineering programs to carbon uptake
is still not clear, as the location of different programs
is difficult to identify, and their impacts are concur-
rent with climate change.
Objectives We aim to detect the location of eco-
engineering programs and attribute the impacts of
eco-engineering and climate change on vegetation
dynamics and carbon uptake in Northeastern China
during 2000–2020.
Methods We developed a new framework to
detect the location of eco-engineering programs by
combining a temporal pattern analysis method and
Markov model, and to attribute the impacts of eco-
engineering and climate change on vegetation green-
ness and carbon uptake by combining a neighbor
contrast method within a sliding window and trend
analysis on the normalized difference vegetation
index (NDVI) and gross primary production (GPP).
Results We identified four main forestry eco-engi-
neering programs: croplands to forest (CtoF), grass-
lands to forest (GtoF), savannas to forest (StoF),
and natural forest conservation (NFC) programs,
whose areas accounted for 2.11%, 1.89%, 3.41%,
and 1.72% of the total study area, respectively. Both
eco-engineering and climate change contributed to
the increase in greenness and carbon uptake. Com-
pared to climate change effect, eco-engineering
increased NDVI and GPP by 121% and 21.43% on
average, respectively. Specifically, the eco-engineer-
ing-induced increases in GPP were 54.1%, 9.46%,
Supplementary Information The online version
contains supplementary material available at https:// doi.
org/ 10. 1007/ s10980- 023- 01679-x.
H.Li(*)· W.Gao· Y.Liu
CAS Key Laboratory ofForest Ecology andManagement,
Institute ofApplied Ecology, Chinese Academy
ofSciences, Shenyang110016, China
e-mail: huidong.li@iae.ac.cn
W.Gao
College ofResources andEnvironment, University
ofChinese Academy ofSciences, Beijing100049, China
F.Yuan
Department ofSoil, Water, andClimate, University
ofMinnesota, SaintPaul, MN, USA
M.Wu
Department ofPhysical Geography andEcosystem
Science, Lund University, 22100Lund, Sweden
M.Wu
Department ofEarth Sciences, Uppsala University,
75105Uppsala, Sweden
L.Meng
Department ofEarth andEnvironmental Sciences,
Vanderbilt University, Nashville37240, USA

Landsc Ecol
1 3
Vol:. (1234567890)
8.13%, and 24.20% for CtoF, GtoF, StoF, and NFC,
respectively.
Conclusions These findings highlight the important
and direct contribution of eco-engineering on vegeta-
tion greening with positive effects on carbon seques-
tration at a fine scale, providing an important impli-
cation for eco-engineering planning and management
towards a carbon-neutral future.
Keywords Ecological engineering· Climate
change· Carbon uptake· Attribution· Northeastern
China
Introduction
Terrestrial ecosystems can capture carbon dioxide
(CO2) from the atmosphere through photosynthesis,
significantly slowing down atmospheric CO2 increase
(Keenan etal. 2016) and mitigating global warming
(Shevliakova et al. 2013). Carbon sinks in terres-
trial ecosystems have continued to increase since the
1970s (Friedlingstein et al. 2020) and offset 30% of
anthropogenic CO2 emissions (Friedlingstein et al.
2022). It was predicted that the global terrestrial eco-
systems can still be a strong carbon sink by the end of
twenty-first century (Shao etal. 2013; Friedlingstein
etal. 2014; Harper etal. 2018; Green etal. 2019). As
the main terrestrial carbon pool, forests store ~ 45%
of terrestrial carbon (Bonan 2008), contributing more
than 80% of terrestrial carbon sinks (Pan etal. 2011).
Forests have become the key terrestrial ecosystems to
maintain the balance between anthropogenic carbon
emissions and ecosystem carbon absorption.
Afforestation through ecological engineering (eco-
engineering) is an efficient solution to enhance car-
bon sink of terrestrial ecosystems and has become
one of the most cost-effective ways to mitigate cli-
mate change (Griscom etal. 2017; Bernal etal. 2018;
Palmer 2021). In America (Domke et al. 2020),
Europe (Kim etal. 2016), Asia (Lu et al. 2018), and
many other regions, different eco-engineering pro-
grams (e.g. such as natural forest protection, shel-
ter forest system, and natural landscape restoration,
etc.) have been successively planned and carried out
(Brancalion and Holl 2020). These eco-engineering
programs have contributed to the greening of the
earth and thus increased global carbon sink in recent
decades (Tagesson etal. 2020; Cabon etal. 2022). It
was estimated that the carbon sequestration of global
forest restoration and afforestation programs was 1.30
PgC a−1, which offset more than 10% of carbon emis-
sions caused by human activities (Pugh etal. 2019).
Moreover, at the global scale, there are more than
900million hectares of area for potential afforestation
programs (Bastin Jean-Francois et al. 2019), which
could further increase 42 PgC a−1 carbon sequestra-
tion (Veldman etal. 2019; Domke etal. 2020).
China is ambitious to achieve the goal of car-
bon neutrality by 2060 and has implemented sev-
eral national eco-engineering programs to increase
the carbon sink of terrestrial ecosystems (Yu et al.
2022a). These eco-engineering programs in China are
dominated by young and middle-aged forests and thus
have huge carbon sequestration potential (Yao etal.
2018). However, the contributions of different eco-
engineering programs to carbon sink are still unclear
(Piao et al. 2022a). One of the most important rea-
sons is that identifying the locations of different eco-
engineering types is difficult. At the regional scale,
many studies divided the natural area and human-
disturbed area mainly based on the extent of planning
eco-engineering programs (Lu etal. 2018; Chen etal.
2021). However, eco-engineering programs are usu-
ally planned based on the functional areas and their
spatial overlap often occurs (He et al. 2015, 2020;
Mao et al. 2019; Wang et al. 2022a). This coarse
dividing method at a regional scale cannot reflect
the real change of land cover and landscape at a fine
scale. At the pixel scale, the traditional identification
method based on the land cover change between the
start and the end years may have large uncertain-
ties, due to the uncertainties of land cover data prod-
ucts (Zhou etal. 2008; Zomlot et al. 2017; Liu etal.
2018). Moreover, forest cover variabilities during the
implementation of eco-engineering further compli-
cate the land cover classification and may result in
repeated shifts between original land cover type (e.g.,
agriculture) and forest type (Zhan et al. 2023). In
addition, eco-engineering programs are usually con-
structed continuously in a region. The identification
based on land cover change only on a pixel scale can-
not reflect the whole regional change process of eco-
engineering (Simanjuntak etal. 2020). Therefore, an
accurate identification method of eco-engineering to
capture the robust and spatially continuous transition
process of land cover information at a regional scale
is needed.

Landsc Ecol
1 3
Vol.: (0123456789)
In eco-engineering regions, carbon sequestration
is jointly affected by eco-engineering and climate
change (Piao et al. 2022b). Eco-engineering can
expand forest area, while climate change represented
by CO2 fertilization and warming effects may enhance
plant photosynthesis in ecosystems (Liu etal. 2019;
Wang etal. 2020). To predict future carbon sink and
guide future eco-engineering planning, it is of great
importance to attribute the impacts of eco-engineer-
ing and climate change on carbon sink (Fang etal.
2015; Chen et al. 2021). However, it is challenging
and the results largely depended on attribution meth-
ods. Some studies attempted to separate the contribu-
tions of eco-engineering and climate change on car-
bon uptake by comparing the paired field sampling
and upscaling the field survey dataset (Deng et al.
2017; Lu et al. 2018; Zhang et al. 2020; Yin et al.
2022), but the estimation accuracy was constrained
by the lack of some key environment information as
well as large landscape heterogeneity. Some studies
have used multi-model simulations to investigate the
effects of environmental change and eco-engineering
on carbon sinks (Piao etal. 2015; Gang et al. 2018).
Their results, however, were constrained by the
uncertainties of model process and parameters (Wang
et al. 2022b). Other studies used regression residu-
als to distinguish the carbon sink changes caused by
different factors (Qi etal. 2019; Chen etal. 2021; Ge
etal. 2021; Liu etal. 2022c), but the results had large
systematic uncertainties rooted in some assumptions
(Shi etal. 2021). An accurate method to attribute eco-
engineering and climate change impacts on carbon
sink is also needed.
In this study, a novel framework was developed to
attribute eco-engineering and climate change impacts
on vegetation greenness and carbon uptake and
applied in Northeastern China. Specifically, we first
identified the location of eco-engineering programs
by combining a temporal pattern analysis method and
Markov model based on the changes in land cover
and vegetation coverage. We then calculated the
trends of NDVI and GPP with and without eco-engi-
neering within a sliding window around the eco-engi-
neering regions. Finally, we separated the impact of
eco-engineering on GPP from climate change using a
neighbor contrast method and examined its relation-
ship with vegetation change. Our study can provide
insight into the role of eco-engineering in increasing
carbon uptake at a fine scale and thus support future
eco-engineering planning and management from a
perspective of carbon neutrality.
Study area anddata
Study area
The study was conducted in Northeastern China
(115–135° E, 39–54° N Fig.1), a region with abun-
dant forest resources and a large potential to absorb
the CO2 in the atmosphere (Yu et al. 2011). Since
1980s, several ecological restoration programs such
as the Three-North Shelterbelt, Natural Forest Con-
servation, and Grain to Green programs have been
successively carried out in Northeastern China (He
et al. 2020; Liu etal. 2022a, b). In the first decade
of twenty-first century, the amount of carbon sink
of these programs reached 97.7 PgC, accounting for
~ 30% of the total carbon sequestration of terrestrial
ecosystems in Northeastern China (Lu et al. 2018),
and this is likely to continue in the future (Yuan etal.
2021).
The study area has a temperate monsoon climate,
spanning from the middle temperate zone to the
cold temperate zone from south to north. The annual
average temperature ranges from − 4 to 11 °C, with
a significant difference between the north and the
south. The annual precipitation changes from 300 to
1000mm, transitioning from humid and semi-humid
areas to semi-arid areas from the southeast to the
northwest. Northeastern China is one of the regions
with the most significant warming in China (Ren
et al. 2005). Its annual average warming rate (0.36
°C/10a) is significantly higher than the national aver-
age (0.25 °C/10a) (Jia and Guo 2011), and the rate of
warming increases with latitude. At the same time,
future climate simulation studies show that the aver-
age temperature of this region will continue to rise
by more than 3 °Cby the end of twenty-first century,
accompanied by a weak trend of humidification (Tao
etal. 2016). The carbon sequestration in this region
has been significantly affected by climate change
(Chen etal. 2021).
Data
We used multiple remote sensing datasets from
MODIS products, including land use/land cover

Landsc Ecol
1 3
Vol:. (1234567890)
(LULC), vegetation continuous fields (VCF), nor-
malized difference vegetation index (NDVI), and
gross primary production (GPP), for the study
period 2000–2020. LULC and VCF data were
used to distinguish the location of eco-engineering
programs. NDVI and GPP were used to show the
changes in vegetation dynamics and carbon uptake
and to disentangle the impacts of eco-engineering
and climate change.
LULC
Yearly LULC data was acquired from MCD12Q1
product collection 6 at a resolution of 500 m
(Sulla-Menashe et al. 2019). The MCD12Q1
data is derived using supervised classifications of
MODIS Terra and Aqua reflectance data and then
uses additional post-processing that incorporates
prior knowledge and ancillary information to fur-
ther refine specific classes. The LULC data product
was obtained based on six different classification
schemes and the one based on International Geo-
sphere-Biosphere Program (IGBP) classification
(17 types) was used in this study.
VCF
Annual VCF data was acquired from MOD44B at
250 m resolution (Majasalmi and Rautiainen 2021).
The data represents the sub-pixel-level surface veg-
etation cover and provides a gradient of three ground
cover components: percent tree cover, percent non-
tree cover, and percent non-vegetated. The VCF algo-
rithm involved a semi-automated process to generate
regression trees with machine learning software using
monthly composites of Terra 250- and 500-m Land
Surface Reflectance data from all seven bands and
Land Surface Temperature.
NDVI
Monthly NDVI data from MOD13A product col-
lection 6 at a resolution of 1 km (Beck et al. 2006)
was used to calculate the annual mean NDVI for the
following analysis. The monthly data is generated
by extracting all 16-day MOD13A2 NDVI products
overlapping with the month and employs a weighted
temporal average. The NDVI data is derived from
atmospherically-corrected reflectance in the red,
Fig. 1 The location and land cover types of the study area. Colors represent the MODIS land cover types in 2001 (see sec-
tion“LULC” for more details)

Landsc Ecol
1 3
Vol.: (0123456789)
near-infrared, and blue wavebands and has been
widely used to study vegetation greenness.
GPP
Annual GPP data was obtained from the
MOD17A3HGF product version 6.1 with a spa-
tial resolution of 500 m (Wei et al. 2022).
MOD17A3HGF product is the gap-filled dataset of
the 8-day MOD17 product, which is derived based on
the original radiation use efficiency logic of Monteith.
The MOD17A3HGF dataset has cleaned the poor-
quality inputs based on the Quality Control (QC)
label for every pixel, e.g., 8-day Leaf Area Index
and Fraction of Photosynthetically Active Radiation
(LAI/FPAR) (Liu etal. 2021).
Methods
This study developed a new framework to attribute
the impacts of eco-engineering and climate change
on greening and carbon uptake at a fine scale
(Fig. 2). The framework includes three key steps:
(1) identify eco-engineering regions by combin-
ing a temporal pattern analysis method and Markov
model, (2) analyze the temporal trends of NDVI
and GPP in the eco-engineering region by combin-
ing a Mann–Kendall test and linear regression, and
(3) quantify the contributions of eco-engineering
and climate change to the change of GPP through a
neighbor contrast method within a sliding window
and attribution calculation.
Identify eco-engineering region
Eco-engineering regions generally experienced sig-
nificant changes in land cover and vegetation cover-
age (Yu etal. 2022a). Based on these changes, we
combined a temporal pattern analysis method and
Markov model to identify the location of eco-engi-
neering region.
Input data
Land cove r Forest coverage GPPNDVI
Step 1: Identify eco-engineering region
Temporal pattern analysisPixe l scale:
Regional scale: Markov model
Location of each ec o-engineering program
Step 2: Analyze ec ol ogical change in eco-engineering region
Linear regressi onTrend:
Sign ificance:Mann -Kendall te st
Productivity change for each eco-engineering program
Step 3: Quantify the contribution of eco-engineering and climate change on productivity
Background zone:
Non-eco-engineer ing region,
forest coverage did no t increase
Other zone: (excluded)
Eco-engineer ing zone The difference of two slopes:
eco-engineer ing induced GPP change
Round slid ing-window
Radius:
10 km
GPP slope of eco-engineering zone:
both eco-engineering and climate
change induced GP P change
Neighbor contrast method
GPP slope of background zone:
clim ate chan ge induced GPP change
Attribution of productivity
Fig. 2 A framework for disentangling the effects of eco-engineering programs and climate change on carbon uptake

Landsc Ecol
1 3
Vol:. (1234567890)
Temporal pattern analysis
We used a temporal pattern analysis (Wang et al.
2012; Hussain et al. 2013) to detect the land cover
change process at pixel scale based on the time series
of land cover data. The temporal pattern analysis
determines the dominant change pattern of land cover
by comparing the maximum frequency of all poten-
tial change patterns with different time spans (Fig.
S1). Specifically, we first selected different time spans
to detect the potential land cover change patterns.
We then calculated the frequency of each land cover
change pattern to estimate its probability. We calcu-
lated the frequencies of all land cover change patterns
under each time span and chose the change pattern
with the maximum frequency as the most possible
change pattern under this time span. Finally, we com-
pared the maximum frequencies of all the time spans
and chose the time span with the maximum frequency
as the optimal one (Fig. S2). The land cover change
pattern with the maximum frequency under the opti-
mal time span was taken as the dominant change pat-
tern of the entire time series. We found that more than
98% and 90% of the maximum frequencies are larger
than 60% and 80%, respectively. There are few pix-
els with the same maximum frequencies under differ-
ent time spans. On these pixels, if the corresponding
land cover change patterns were the same, we chose
this pattern as the dominant one, whereas if the cor-
responding land cover change patterns were different,
we assumed this pixel had no dominant pattern. Time
span is an important parameter of the temporal pat-
tern analysis and largely affects the significance of the
calculated frequency of land cover change pattern. A
larger time span can decrease statistical samples for
frequency calculation, leading to larger standard error
and smaller significance of the calculated frequency.
We conducted a statistical significance analysis of all
potential land cover change patterns under different
time spans ranging from 1 to 20 years and found the
large time span decreased the number of samples and
the significance of land cover change patterns, sub-
sequently increasing the uncertainty of the temporal
pattern analysis. When the time span increased to be
larger than 13 years, there were no enough samples,
which could not guarantee a significant difference
between different land cover change patterns. Thus,
we selected the time spans of 1–13 years for the tem-
poral pattern analysis in this study.
Markov model
We then detected the land cover change process at a
regional scale using the Markov model based on the
results of temporal pattern analysis (Fig. S3), given
that the pixel-level results may be affected by the
uncertainties from raw land cover data and missed
the spatial continuity of eco-engineering (Yu et al.
2022b). Markov model is an effective random process
model to simulate the dynamics of land cover change
process and has been applied in many land cover
change studies (Taylor and Rising 2021; Fu et al.
2022). Markov model can identify the final stable
state of land cover change at a regional scale based
on a state transfer matrix (i.e., the probability that the
region undergoes land cover transition from one type
to another). Through a comparison of the final land
cover state from Markov model and the initial state,
we can determine the dominant land cover pattern at
a regional scale. In this study, we calculated the state
transfer matrix based on the pixel-level land cover
change from 2000 to 2020 within a sliding window
around each pixel and then applied the Markov model
to obtain the final land cover types. Based on the ini-
tial and final land cover types in the sliding window,
we obtained a spatially continuous and robust land
cover change process at a regional scale. A sensitiv-
ity test of sliding window radius ranging from 10
to 30 km was conducted resulting in similar results
despite the varying radius, indicating an insensitivity
to radius chosen. A large radius could significantly
increase the computing time with very few added
values. Therefore, we chose the smallest one (10km)
as the radius to make sure high accuracy and fast
computation.
Division ofeco‑engineering programs
During the study period, Northeastern China imple-
mented many eco-engineering programs (He et al.
2020). Based on the change of land cover from
Markov model and forest coverage trend (the trend
calculation was described in the following section),
we classified the eco-engineering regions into four
major types, including Croplands to forest (CtoF),
Grasslands to forest (GtoF), Savanna to forest (StoF),
and natural forest conservation (NFC). The CtoF,
GtoF, and StoF were defined based on the land cover
changes from croplands, grasslands, and savanna to

Landsc Ecol
1 3
Vol.: (0123456789)
forests, whereas the NFC was defined based on the
consistent forest land cover type but significantly
increased forest coverage.
Analyze ecological change in eco-engineering region
We analyzed the annual variations of VCF, NDVI,
and GPP and calculated their trends in each eco-engi-
neering region. Specifically, we used the Mann–Ken-
dall trend test method to calculate the significance
of the changing trend (Ma etal. 2020) and then used
a linear regression method to calculate the annual
change rate of VCF, GPP, and NDVI. The VCF trend
was used to classify the eco-engineering type of NFC
at a significance of 0.05, whereas the NDVI and GPP
trends were used to attribute the contribution of four
eco-engineering programs on vegetation dynamics
and carbon uptake.
Quantify the contributions of eco-engineering and
climate change on GPP
We used a neighbor contrast method to quantify the
different contributions of eco-engineering and climate
change to GPP through the comparison of ecologi-
cal processes between the eco-engineering zone and
background zone within the sliding window around
each eco-engineering region.
Distinguish neighbor contrast zones
The GPP increase in eco-engineering regions results
from the combined effects of eco-engineering and
climate change. We used a neighbor contrast method
to disentangle the two effects. We first identified the
neighboring pixels with and without eco-engineering.
Specifically, we distinguished neighbor contrast zones
within a 10-km-radius window (same as section
“Markov model”) around the eco-engineering region
and divided all pixels within the window into three
classes: eco-engineering zone, background zone,
and other zone. The eco-engineering zone includes
the pixels that experienced the dominant land cover
change pattern of the eco-engineering program.
While the background zone (i.e., the productivity
variation only results from climate change) includes
the pixels satisfying the conditions that the land cover
did not change and was consistent with the initial land
cover type of the dominant eco-engineering program
within the window, and the forest coverage had no
significant increase. Other pixels within the window
were classified as other zone and excluded in the fol-
lowing analysis.
Attribution ofgreening andGPP trend
We calculated the average temporal trends (i.e.,
annual change rate) of NDVI and GPP for the eco-
engineering and background zones within each
10-km-radius window, separately. The difference of
the annual change rates of NDVI and GPP between
eco-engineering and background zones represents
the variations of vegetation dynamics and productiv-
ity caused by eco-engineering. We first compared the
spatial patterns of NDVI and GPP trends and their
difference between eco-engineering and background
zones in the study area. We then aggregated the NDVI
and GPP trends and their difference between eco-
engineering and background zones across the study
area for each eco-engineering program by statistical
variables, such as quantile and mean values, quanti-
fying the overall contribution of eco-engineering on
vegetation greenness and carbon uptake in Northeast-
ern China. In addition, we examined the relationship
of the differences of NDVI and GPP trends between
eco-engineering and background zones to reveal the
mechanism of eco-engineering impact on carbon
uptake.
Results
Mapping of eco-engineering programs
We found substantial changes in land cover in North-
eastern China between 2000 and 2020, with a net
gain in forest area and loss in croplands, grasslands,
and savannas (Fig.3). Specifically, deciduous broad-
leaf forests showed the largest net increase in area,
i.e., 10187.26 km2 (2.84%) increase, which includes
20,257.38 km2 (5.65%) gain and 10,070.12 km2
(2.81%) loss. The area gain of deciduous broadleaf
forests mainly comes from the land use changes of
croplands (2.07%, 7421.12 km2), grasslands (1.52%,
5460.25 km2), and savannas (1.47%, 5273.25 km2).
Mixed forests showed the second largest net increase
in area, i.e., 10067.68 km2 increase, which mainly

Landsc Ecol
1 3
Vol:. (1234567890)
came from deciduous broadleaf forests (1.78%,
6388.63 km2) and savannas (1.05%, 3764.19 km2).
Croplands showed the largest area loss (3100.875
km2, 5.32%), which were mainly converted to decidu-
ous broadleaf forests (2.07%), croplands/natural veg-
etation mosaics (1.37%), and built-up area (0.75%).
Woody savannas and grasslands also showed large
area loss, 13,402.31 km2 (3.74%) and 13,672.50 km2
(3.82%), respectively. Evergreen needleleaf forests,
shrublands, and permanent wetlands didn’t change
much. These results indicate a significant increase in
forest areas, especially deciduous broadleaf forests,
and a significant loss in croplands, grasslands, and
savannas in Northeastern China from 2000 to 2020.
Based on the major land cover conversions men-
tioned above, we mainly focused on four forestry eco-
engineering programs in the study area, i.e., crop-
lands to forest (CtoF), grasslands to forest (GtoF),
savannas to forest (StoF), and natural forest conser-
vation (NFC) (see details in section “Division of
eco-engineering programs”). These four land cover
conversions occurred in distinct regions (Fig. 4),
accounting for 9.05% of the study area. The areas
experiencing StoF were in the western study region.
Northern study regions mostly show the StoF. CtoF
and NFC mostly occurred in the eastern study region.
The areas of CtoF, GtoF, StoF, and NFC account for
2.11%, 1.89%, 3.41%, and 1.72% of the total study
area, respectively. The central and upper west of the
study region, which were mostly agricultural land and
grassland, respectively, didn’t experience much land
cover change.
Enhanced greening due to eco-engineering
We found eco-engineering programs profoundly
increased the greening trend in Northeastern China
(Fig. 5). Without eco-engineering programs, the
background greening trend was 0.15/100a (slope of
NDVI during 2000–2020, p < 0.05) on average across
all study areas with significant spatial variations
(Fig. 5b). The northern and certain regions in the
east and the south showed largest increasing rates of
NDVI over time. As a comparison, the greening trend
with eco-engineering programs was 0.34/100a on
average (Fig. 5a), which was 0.19/100a (1.21 times)
higher than the background greening trend on average
(Fig. 5c). The largest effect of eco-engineering pro-
grams on the greening speed was found in the north
of the study areas, where woody savannas and shrubs
were the main land cover, and the smaller greening
regions were in the east and the south.
Among the four eco-engineering programs,
the background zones of NFC showed the high-
est NDVI trend (0.22/100a), while the eco-engi-
neering zones of StoF showed the highest NDVI
trend (0.41/100a) with eco-engineering programs
(Fig. 6). The increases in NDVI trend due to
Fig. 3 Land cover changes
in Northeastern China
between 2000 and 2020.
Positive values indicate an
increase, while negative
values indicate a decrease

Landsc Ecol
1 3
Vol.: (0123456789)
Fig. 4 Locations of four
main forestry eco-engineer-
ing programs
Fig. 5 Spatial patterns of mean NDVI trend of a eco-engineering zone, b background zones, and c their difference during 2000–
2020. Only pixels with significant change are shown in the figure (P < 0.05)

Landsc Ecol
1 3
Vol:. (1234567890)
eco-engineering programs (difference between eco-
engineering and background zones) were on aver-
age 0.18, 0.15, 0.24, and 0.13/100a for CtoF, GtoF,
StoF, and NFC, respectively (Fig. 6). There were
larger spatial variations (e.g., interquartile range)
in the increases of NDVI trends for the types of
CtoF and StoF programs, especially when compar-
ing to NFC program.
Increased carbon update due to eco-engineering
We found that eco-engineering programs increased
the GPP trend in Northeastern China (Fig. 7). With-
out eco-engineering programs, the background GPP
trend (p < 0.05) was 0.56 kgC/m2/a on average across
all study areas (Figs. 7b, 8). The western region
showed the largest increasing rate of GPP over time.
As a comparison, the GPP trend with eco-engineering
programs was 0.68 kgC/m2/a on average (Figs.7a, 8),
Fig. 6 NDVI trends in
areas a with and without
eco-engineering programs,
and b their difference
during 2000–2020. The
error bar represents a 95%
confidence interval, and the
black line and two edges
from the box plot represent
the mean, 25th, and 75th
percentiles, respectively
Fig. 7 Spatial patterns of mean GPP trends of a eco-engineering zones, b background zones, and c their difference during 2000–
2020. Only pixels with significant change are shown in the figure (P < 0.05)

Landsc Ecol
1 3
Vol.: (0123456789)
which was 0.12 kgC/m2/a (21.43%) higher than the
background (Figs. 7c, 8). The largest effect of eco-
engineering programs on the GPP trend was found in
the east and some southern parts of the study area.
Among the four eco-engineering programs, the
background zones of GtoF showed the highest GPP
trend (0.86 kgC/m2/a), while the background zones of
StoF showed the lowest GPP trend (0.46 kgC/m2/a).
This is also true for GPP trends with eco-engineering
zones. The increases in GPP trends of eco-engineer-
ing zones due to afforestation were on average 0.29,
0.08, 0.04, and 0.12 kgC/m2/a for CtoF, GtoF, StoF,
and NFC programs, respectively (Fig. 8), account-
ing for 54.1%, 9.46%, 8.13%, and 24.20%, compared
to the trend in background zones caused by climate
change.
The relationship between GPP trend and vegetation
greening trend
We found the magnitude of the increase in GPP
trend (∆GPP trend between eco-engineering and
background zones) was positively correlated with
the increases in VCF trend (∆VCF trend between
eco-engineering and background zones) and NDVI
trend (∆NDVI trend between eco-engineering and
background zones) due to eco-engineering programs
(Fig. 9). Generally, ∆GPP trend showed higher cor-
relation with ∆NDVI trend (Fig.9a), compared to the
∆VCF trend (Fig.9b), especially for CtoF and StoF
areas. The NFC area showed the strongest response
of ∆GPP trend to ∆VCF trend, i.e., 0.71 kgC/m2/%,
while the GtoF area showed the smallest slope (0.47
Fig. 8 GPP trends of areas a with and without eco-engineer-
ing programs, and b their difference during 2000–2020. The
error bar represents a 95% confidence interval, and the black
line and two edges from the box plot represent the mean, 25th,
and 75th percentiles, respectively

Landsc Ecol
1 3
Vol:. (1234567890)
kgC/m2/%). In contrast, the NFC area showed the
smallest slope of ∆GPP trend against ∆NDVI trend,
i.e., 0.16 × 102 kgC/m2, whereas the StoF area
showed the largest slope (0.90× 102 kgC/m2).
We found that the linear relationship between
∆GPP trend and ∆VCF trend was not as large as that
between ∆GPP trend and ∆NDVI trend. This is likely
because VCF mainly represents forest expansion
only in horizontal, while NDVI could indicate forest
expansion and growth both in horizontal and verti-
cal. Interestingly, we found a non-linear increase in
∆NDVI trend that became larger with the increase
in ∆VCF trend (Fig.9c). The reason for the smaller
∆NDVI trend at the beginning might be that the eco-
engineering occurred very recently, and the trees were
still very young. The initial eco-engineering mainly
Fig. 9 Relationships between the changing trends of a GPP
and VCF, b GPP and NDVI, and c NDVI and VCF. The points
show the mean values of x-axis at an interval of 0.01. Linear
regression lines are shown for a, b, and a locally weighted
smoothing line is shown for c. The slope and correlation coef-
ficient are shown from linear regressions

Landsc Ecol
1 3
Vol.: (0123456789)
promoted forest expansion, but did not significantly
increase NDVI. This partly explained the smaller
contribution of eco-engineering to the increased GPP,
compared to climate change. Moreover, this implies
that the newly planted trees in the eco-engineering
region still have a large potential to increase the
greening and carbon uptake in the future, along with
the growth of trees of the current eco-engineering
programs.
Discussions
Magnitude difference in contributions of
eco-engineering and climate change
Accurately attributing the impacts of eco-engineering
and climate change on carbon cycle largely depends
on the specific methods and approaches employed.
There are mainly two existing approaches: field-based
estimation and remote sensing-based estimation.
Based on the paired field samples measurements, Lu
et al. (2018) compared GPP within eco-engineering
regions and surrounding natural ecosystems, and
found eco-engineering programs contributed an
increase in GPP by 50.7% on average across China
and by 29.1% in Northeastern China with the largest
contribution from CtoF program. Remote sensing-
based attribution studies mainly employed regres-
sion residual method. For example, using machine
learning residual method, (Chen et al. 2021) found
the contributions of climate change (48–56%) and
human activities (44–52%) on GPP were similar in
eco-engineering regions in China. While using mul-
tiple regression residual method, Liu et al. (2022c)
found human activities in eco-engineering regions
contributed > 70% to the increases in GPP in China.
Specially, in Northeastern China, Chen et al. (2021)
found human activities drove the 67% increase in
GPP, while Liu etal. (2022c) found human activities
impact can reach 86% in eco-engineering regions.
This study developed a new framework to quan-
tify the contributions of eco-engineering and climate
change on carbon uptake in Northeastern China.
This study found a 21.43% increase in GPP caused
by eco-engineering in Northeastern China, com-
pared to that caused by climate change, with increas-
ing order of CtoF, NFC, GtoF, and StoF. Our results
are consistent with the study based on paired field
samples measurements (Lu etal. 2018). In contrast,
the residual-based attribution method significantly
overestimates the contribution of eco-engineering,
because they only coarsely separated “natural area”
and “human-disturbed area” based on the planned
eco-engineering ozone at a regional scale. Therefore,
these studies fail to accurately capture the actual and
local eco-engineering influences in eco-engineering
regions and did not exclude the potential anthropo-
genic impacts (such as deforestation) in the back-
ground region Chen et al. (2021). Moreover, the
residual method-based studies treat the regression
residuals as the anthropogenic influence and thus may
produce large overestimation in the planned eco-engi-
neering zones that have no human activities Bernier
etal. (2017). In sum, our newly developed framework
represents a more accurate way of attributing the eco-
engineering and climate change impacts, because of
the accurate detection of eco-engineering regions and
calculation of contributions at a fine scale.
Uncertainties
It is worth noting that our study is based on certain
assumptions for the newly developed framework,
which could introduce uncertainties in the attribution
results. Firstly, we assumed the land cover change
from non-forest to forest was solely caused by eco-
engineering, without considering the potential influ-
ence of climate change on land cover change when
detecting the locations of different eco-engineering
types. This may lead to an underestimation of cli-
mate change contribution on carbon uptake. But the
study area is a highly human-managed region (Yu
et al. 2011), and the natural evolution from crop-
land, grassland, and savanna to temperate and cold
temperate forests in the study area is a gradual pro-
cess. Therefore, the short-term influence of climate
change on land cover change is relatively small,
which is difficult to detect by the coarse-resolution
remote sensing data. Secondly, we adopted a “space-
for-time” substitution between eco-engineering and
background regions to attribute the individual influ-
ence of eco-engineering and climate change on car-
bon uptake. In this approach, we didn’t consider the
interaction between eco-engineering areas and their
adjacent regions in a horizontal manner and the inter-
action between eco-engineering and climate change
in a vertical manner (Damgaard 2019). For example,

Landsc Ecol
1 3
Vol:. (1234567890)
in horizontal direction, the attribution ignores the
impact of eco-engineering program on the surround-
ing land cover and the corresponding carbon update
through horizontal flow of water and nutrients. The
potential influence of spatial gradients of species
composition on the temporal changes in biodiversity
and the carbon cycle are considered neither. In verti-
cal direction, the method is based on an assumption
that climate change and its influence on eco-engineer-
ing region and the surrounding background region are
the same thus ignoring the impact of the interaction
between eco-engineering and climate change on the
carbon uptake. However, the impact of these interac-
tions in vertical and horizontal aspects on the carbon
uptake is complicated and thus difficult to quantify
and exclude in this study. Ecosystem process model
that fully considers vertical and horizontal move-
ments such as water and nutrients is a more accurate
approach for such purpose. The future study should
consider applying the process-based ecosystem mod-
els to further evaluate the current results and explain
the mechanism behind.
Conclusions andoutlook
We developed a new framework to attribute the
impacts of eco-engineering and climate change on
vegetation greenness and carbon uptake and used this
framework to distinguish the different impacts of eco-
engineering and climate change on NDVI and GPP in
Northeastern China during 2000–2020. Our results
indicate that 9.05% of the study area experienced eco-
engineering, with four main co-engineering programs
of CtoF, StoF, GtoF, and NFC. Both eco-engineering
and climate change promoted vegetation greening
and increased GPP. On average, eco-engineering pro-
grams further increased the NDVI by 1.21 times and
the GPP by 21.43%, compared to the change caused
by climate change. CtoF showed the largest increase
(54.1%), while StoF showed the smallest (8.13%).
Our findings highlight the important role of eco-engi-
neering in increasing carbon uptake and the need for
future comprehensive studies to quantify the different
changes in forests from different land cover types.
These results could contribute to understanding
where to plant trees to achieve potential climatic ben-
efits in future eco-engineering programs. Moreover,
this new understanding could be incorporated into
the land surface models in projecting the ecosystem
changes within the Earth system (Fisher and Koven
2020). Although land surface models have greatly
advanced in the complexity of representing land
surface biophysics, human activities (afforestation
and forest management) are still underrepresented
in predicting carbon cycles under climate change. It
will bring large uncertainties in carbon estimation,
especially in the regions such as Northeastern China
that experienced profound land cover change due
to eco-engineering programs. Accurately assessing
these impacts could help further improve the efforts
and benefits of eco-engineering programs at local and
regional scales and also could provide a useful guide-
line to achieve the goal of carbon neutrality at multi-
ple scales.
Acknowledgements The work was supported by
National Key Research and Development Program of China
(2022YFF1300501) and Major Program of Institute of Applied
Ecology, Chinese Academy of Sciences (IAEMP202201). MW
is supported by the Swedish National Space Agency (SNSA)
under Grant Dnr 2021-00111.
Author contributions HL wrote the main manuscript text,
did the statistical analysis, and prepared Figs.1, 3, 4, 5, 6, 7,
8, and 9. WG and YL prepared the data and Fig.2 and wrote
parts of the “Introduction” and “Methods” sections. All authors
reviewed the manuscript.
Funding The work was supported by National Key
Research and Development Program of China (Grant No.
2022YFF1300501) and Major Program of Institute of
Applied Ecology, Chinese Academy of Sciences (Grant No.
IAEMP202201), and Swedish National Space Agency (Grant
No. 2021-00111).
Declarations
Competing interests The authors declare no competing inter-
ests.
References
Bastin Jean-Francois, Yelena F, Claude G et al (2019) The
global tree restoration potential. Science 365:76–79
Beck PSA, Atzberger C, Hogda KA et al (2006) Improved
monitoring of vegetation dynamics at very high latitudes:
a new method using MODIS NDVI. Remote Sens Environ
100:321–334
Bernal B, Murray LT, Pearson TRH (2018) Global carbon
dioxide removal rates from forest landscape restoration
activities. Carbon Balanc Manag 13:22

Landsc Ecol
1 3
Vol.: (0123456789)
Bernier PY, Paré D, Stinson G etal (2017) Moving beyond
the concept of “primary forest” as a metric of forest
environment quality. Ecol Appl 27:349–354
Bonan GB (2008) Forests and climate change: forcings,
feedbacks, and the climate benefits of forests. Science
320:1444–1449
Brancalion PHS, Holl KD (2020) Guidance for successful
tree planting initiatives. J Appl Ecol 57:2349–2361
Cabon A, Kannenberg SA, Arain A etal (2022) Cross-biome
synthesis of source versus sink limits to tree growth.
Science 376:758–761
Chen Y, Feng X, Tian H etal (2021) Accelerated increase in
vegetation carbon sequestration in China after 2010: a
turning point resulting from climate and human interac-
tion. Glob Change Biol 27:5848–5864
Damgaard C (2019) A critique of the space-for-time substitu-
tion practice in community ecology. Trends Ecol Evol
34(5):416–421
Deng L, Liu S, Kim DG etal (2017) Past and future carbon
sequestration benefits of China’s grain for green pro-
gram. Glob Environ Change 47:13–20
Domke GM, Sonja S, Oswalt N etal (2020) Tree planting has
the potential to increase carbon sequestration capacity
of forests in the United States. Proc Natl Acad Sci USA
117:24649–24651
Fang J, Yu G, Ren X et al (2015) Carbon sequestration in
China’s terrestrial ecosystems under climate change pro-
gress on ecosystem carbon sequestration from the CAS
Strategic Priority Research Program. Bull Chin Acad
Sci 30:848–857
Fisher RA, Koven CD (2020) Perspectives on the future of
land surface models and the challenges of representing
complex terrestrial systems. J Adv Model Earth Syst.
https:// doi. org/ 10. 1029/ 2018M S0014 53
Friedlingstein P, Meinshausen M, Arora VK et al (2014)
Uncertainties in CMIP5 climate projections due to car-
bon cycle feedbacks. J Clim 27:511–526
Friedlingstein P, O’Sullivan M, Jones MW et al (2020)
Global carbon budget 2020. Earth Syst Sci Data
12:3269–3340
Friedlingstein P, Jones MW, O’Sullivan M etal (2022) Global
carbon budget 2021. Earth Syst Sci Data 14:1917–2005
Fu F, Deng S, Wu D etal (2022) Research on the spatiotempo-
ral evolution of land use landscape pattern in a county area
based on CA-Markov model. Sust Cities Soc 80:103760
Gang C, Zhao W, Zhao T etal (2018) The impacts of land con-
version and management measures on the grassland net
primary productivity over the Loess Plateau, Northern
China. Sci Total Environ 645:827–836
Ge W, Deng L, Wang F, Han J (2021) Quantifying the con-
tributions of human activities and climate change to veg-
etation net primary productivity dynamics in China from
2001 to 2016. Sci Total Environ 773:145648
Green JK, Seneviratne SI, Berg AM etal (2019) Large influ-
ence of soil moisture on long-term terrestrial carbon
uptake. Nature 565:476–479
Griscom BW, Adams J, Ellis PW etal (2017) Natural climate
solutions. Proc Natl Acad Sci USA 114:11645–11650
Harper AB, Powell T, Cox PM etal (2018) Land-use emissions
play a critical role in land-based mitigation for Paris cli-
mate targets. Nat Commun 9:2938
He B, Miao L, Cui X, Wu Z (2015) Carbon sequestration
from China’s afforestation projects. Environ Earth Sci
74:5491–5499
He X, Wang Z, Zheng H (2020) Assessment of ecological
impacts of key ecological projects in Northeast China.
China Science Publishing, Beijing
Hussain M, Chen D, Cheng A et al (2013) Change detec-
tion from remotely sensed images: from pixel-based to
object-based approaches. ISPRS J Photogramm Remote
Sens 80:91–106
Jia J, Guo J (2011) Characteristics of climate change in
Northeast China for last 46 years. J Arid Land Resour
Environ 25:109–115
Keenan TF, Prentice IC, Canadell JG et al (2016) Recent
pause in the growth rate of atmospheric CO2 due to
enhanced terrestrial carbon uptake. Nat Commun
7:13428
Kim N, Yiying C, McGrath Matthew J etal (2016) Europe’s
forest management did not mitigate climate warming.
Science 351:597–600
Liu X, Yu L, Si Y etal (2018) Identifying patterns and hot-
spots of global land cover transitions using the ESA CCI
land cover dataset. Remote Sens Lett 9:972–981
Liu Y, Piao S, Gasser T etal (2019) Field-experiment con-
straints on the enhancement of the terrestrial carbon
sink by CO2 fertilization. Nat Geosci 12:809–814
Liu Z, Liu Y, Wang J (2021) A global analysis of agricultural
productivity and water resource consumption changes
over cropland expansion regions. Agric Ecosyst Environ
321:107630
Liu Y, Li H, Wang A (2022a) Estimating the legacy effect
of post-cutting shelterbelt on crop yield using Google
Earth and Sentinel-2 Data. Remote Sens 14(19):5005
Liu Y, Li H, Yuan F et al (2022b) Estimating the impact of
shelterbelt structure on corn yield at a large scale using
Google Earth and Sentinel 2 data. Environ Res Lett
17:044060
Liu Y, Liu H, Chen Y etal (2022c) Quantifying the contri-
butions of climate change and human activities to veg-
etation dynamic in China based on multiple indices. Sci
Total Environ 838:156553
Lu F, Hu H, Sun W et al (2018) Effects of national eco-
logical restoration projects on carbon sequestration
in China from 2001 to 2010. Proc Natl Acad Sci USA
115:4039–4044
Ma J, Zhang C, Yun W etal (2020) The temporal analysis of
regional cultivated land productivity with GPP based on
2000–2018 MODIS data. Sustainability 12:411
Majasalmi T, Rautiainen M (2021) Representation of tree cover
in global land cover products: Finland as a case study
area. Environ Monit Assess 193:121
Mao D, He X, Wang Z etal (2019) Diverse policies leading to
contrasting impacts on land cover and ecosystem services
in Northeast China. J Clean Prod 240:117961
Palmer L (2021) How trees and forests reduce risks from cli-
mate change. Nat Clim Change 11:374–377
Pan Y, Birdsey RA, Fang J etal (2011) A large and persistent
carbon sink in the world’s forests. Science 333:988–993
Piao S, Yin G, Tan J etal (2015) Detection and attribution of
vegetation greening trend in China over the last 30 years.
Glob Change Biol 21:1601–1609

Landsc Ecol
1 3
Vol:. (1234567890)
Piao S, He Y, Wang X, Chen F (2022a) Estimation of China’s
terrestrial ecosystem carbon sink: methods, progress and
prospects. Sci China Earth Sci 65:641–651
Piao S, Yue C, Ding J, Guo Z (2022b) Perspectives on the role
of terrestrial ecosystems in the ‘carbon neutrality’ strat-
egy. Sci China Earth Sci 65:1178–1186
Pugh TAM, Lindeskog M, Smith B etal (2019) Role of forest
regrowth in global carbon sink dynamics. Proc Natl Acad
Sci USA 116:4382–4387
Qi X, Jia J, Liu H, Lin Z (2019) Relative importance of climate
change and human activities for vegetation changes on
China’s silk road economic belt over multiple timescales.
CATENA 180:224–237
Ren G, Chu Z, Zhou Y etal (2005) Recent progresses in stud-
ies of regional temperature changes in China. Clim Envi-
ron Res 10(4):701–716
Shao P, Zeng X, Sakaguchi K etal (2013) Terrestrial carbon
cycle: climate relations in eight CMIP5 earth system mod-
els. J Clim 26:8744–8764
Shevliakova E, Stouffer RJ, Malyshev S et al (2013) Histori-
cal warming reduced due to enhanced land carbon uptake.
Proc Natl Acad Sci USA 110:16730–16735
Shi S, Yu J, Wang F et al (2021) Quantitative contributions of
climate change and human activities to vegetation changes
over multiple time scales on the Loess Plateau. Sci Total
Environ 755:142419
Simanjuntak CI, Deliar A, Virtriana R (2020) Analysis of dom-
inant land cover class based on land change cluster pattern
in West Java Province. In: Fifth international conferences
of Indonesian Society for Remote Sensing: the revolution
of earth observation for a better human life. IOP Publish-
ing Ltd, Bristol, p. 012015
Sulla-Menashe D, Gray JM, Abercrombie SP, Friedl MA
(2019) Hierarchical mapping of annual global land cover
2001 to present: the MODIS collection 6 land cover prod-
uct. Remote Sens Environ 222:183–194
Tagesson T, Schurgers G, Horion S etal (2020) Recent diver-
gence in the contributions of tropical and boreal forests to
the terrestrial carbon sink. Nat Ecol Evol 4:202–209
Tao C, Jiang C, Sun J (2016) Projection of future changes in
climate in Northeast China using a CMIP5 multi-model
ensemble. Chin J Geophys Chin Ed 59:3580–3591
Taylor CA, Rising J (2021) Tipping point dynamics in global
land use. Environ Res Lett 16:125012
Veldman JW, Aleman JC, Alvarado ST et al (2019) Com-
ment on “the global tree restoration potential.” Science
366:eaay7976
Wang D, Gong J, Chen L et al (2012) Spatio-temporal pat-
tern analysis of land use/cover change trajectories in Xihe
watershed. Int J Appl Earth Obs Geoinf 14:12–21
Wang J, Feng L, Palmer PI etal (2020) Large chinese land car-
bon sink estimated from atmospheric carbon dioxide data.
Nature 586:720–723
Wang M, Zhao J, Wang S, Chen B (2022a) Regional contribu-
tions and climate attributions to interannual variation of
global net ecosystems production by an ECOSYSTEM
processed model driven by remote sensing data over the
past 35 years. Remote Sens 14:3208
Wang Y, Zhang Z, Chen X (2022b) The dominant driving force
of forest change in the Yangtze River Basin, China: cli-
mate variation or anthropogenic activities? Forests 13:82
Wei X, Yang J, Luo P etal (2022) Assessment of the variation
and influencing factors of vegetation NPP and carbon sink
capacity under different natural conditions. Ecol Indic
138:108834
Yao Y, Piao S, Wang T (2018) Future biomass carbon seques-
tration capacity of Chinese forests. Sci Bull 17:1108–
1117. https:// doi. org/ 10. 1016/j. scib. 2018. 07. 015
Yin S, Gong Z, Gu L etal (2022) Driving forces of the effi-
ciency of forest carbon sequestration production: spatial
panel data from the national forest inventory in China. J
Clean Prod 330:129776
Yu D, Zhou L, Zhou W etal (2011) Forest management in
Northeast China: history, problems, and challenges. Envi-
ron Manag 48:1122–1135
Yu G, Zhu J, Xu L, He N (2022a) Technological approaches
to enhance ecosystem carbon sink in China: nature-based
solutions. Bull Chin Acad Sci 37:490–501
Yu Z, Ciais P, Piao S et al (2022b) Forest expansion domi-
nates China’s land carbon sink since 1980. Nat Commun
13:5374
Yuan W, Wu S-Y, Hou S et al (2021) Projecting future veg-
etation change for Northeast China using CMIP6 model.
Remote Sens 13:3531
Zhan C, Liang C, Zhao L etal (2023) Detection and attribution
of vegetation dynamics in the National Barrier Zone of
China by considering climate temporal effects. Int J Appl
Earth Obs Geoinf 116:103140
Zhang Y, Yuan J, You C etal (2020) Contributions of National
Key Forestry Ecology Projects to the forest vegetation car-
bon storage in China. For Ecol Manag 462:117981
Zhou Q, Li B, Kurban A (2008) Trajectory analysis of land
cover change in arid environment of China. Int J Remote
Sens 29:1093–1107
Zomlot Z, Verbeiren B, Huysmans M, Batelaan O (2017)
Trajectory analysis of land use and land cover maps to
improve spatial-temporal patterns, and impact assessment
on groundwater recharge. J Hydrol 554:558–569
Publisher’s Note Springer Nature remains neutral with regard
to jurisdictional claims in published maps and institutional
affiliations.
Springer Nature or its licensor (e.g. a society or other partner)
holds exclusive rights to this article under a publishing
agreement with the author(s) or other rightsholder(s); author
self-archiving of the accepted manuscript version of this article
is solely governed by the terms of such publishing agreement
and applicable law.