The Impact of Land-Use Changes on the Spatiotemporal Dynamics of Net Primary Productivity in Harbin, China

Abstract

As the global population continues to rise, the impact of urbanization on land utilization and ecosystems are growing more pronounced, particularly within the expanding area of Asia. The land use/land change (LULC) brought by urban expansion directly impacts plant growth and ecological productivity, altering the carbon cycle and climate regulation functions of the region. This research focuses on Harbin City as a case study, employing an enhanced version of the Carnegie–Ames–Stanford Approach (CASA) model to analyze the spatial–temporal variations in vegetation Net Primary Productivity (NPP) across the area from 2000 to 2020. The findings indicate that Net Primary Productivity (NPP) in Harbin exhibited notable interannual variability and spatial heterogeneity. From 2000 to 2005, a decline in NPP was observed across 60.75% of the area. This reduction was predominantly concentrated in the central and eastern areas of the city, where forested landscapes are the dominant feature. In contrast, from 2010 to 2015, 92.12% of the region saw an increase in NPP, closely related to the overall improvement in NPP across all land-use types. Land-use change significantly influenced NPP dynamics. Between 2000 and 2005, 54.26% of NPP increases stemmed from the transition of farmland into forest, highlighting the effectiveness of the “conversion of farmland back to forests” policy. From 2005 to 2010, 98.6% of the area experienced NPP decline, mainly due to forest and cropland degradation, especially the unstable carbon sink function of forest ecosystems. Between 2010 and 2015, NPP improved across 96.86% of the area, driven by forest productivity recovery and better agricultural management. These results demonstrate the profound and lasting impact of land-use transitions on the spatiotemporal dynamics of NPP.

1. Introduction

Carbon neutrality has become a strategic goal in response to the profound changes occurring within the global climate system. It has been actively accepted and pursued by all countries worldwide. Vegetation’s net primary productivity (NPP) is a crucial component in the global carbon cycle. Therefore, exploring vegetation’s NPP is an endeavor of considerable value and critical significance and is essential to achieving the aspiration of carbon neutrality. Population projections show that Central and South Asia are expected to become the most populous region in the world by 2037 [1]. In 2023, India surpassed China’s population for the first time, emerging as the world’s most populous nation, and this demographic shift is projected to persist over the coming decades [2]. As population increases dramatically, unrestricted expansion of built-up areas becomes the norm, directly driving land use/cover (LULC) change [2,3,4].

Harbin, the capital of Heilongjiang Province in northeastern China, is an exemplary model for examining the impacts of urbanization and land-use changes on NPP. The city is an industrial and agricultural hub, with its economy driven by key industries such as equipment manufacturing, food production, pharmaceuticals, and petrochemicals. However, as Harbin progresses through industrialization, the city faces increasing pressure on land and natural resources, resulting in a reduction in arable land. In 2010, Harbin’s per capita arable land area was only 0.19 hectares, which is lower than the provincial average, signaling the challenges of balancing urban growth with land conservation [5]. Ecologically, Harbin has adopted progressive policies to balance development with sustainability. The city has implemented a “protective development” strategy, including the establishment of ecological red lines, the integration of green infrastructure, and compensation mechanisms for environmental restoration. These measures aim to preserve ecological stability while accommodating urban expansion. Harbin’s approach to balancing industrial growth with ecological preservation provides valuable insights for cities in developing countries, particularly in the Global South, where similar “development-protection” challenges exist [6]. The study of NPP dynamics in such regions is crucial for comprehending the long-term environmental consequences of urbanization.

Notable features of urban sprawl include rapid reduction in vegetated areas [7,8], sprawling growth [9,10], increased economic activity at high altitudes [11,12,13], land cover changes in agricultural areas [14,15,16,17], and increased urban heat island effect [18,19]. These changes seriously challenge environmental, ecological, economic, and social systems [20].

Vegetation changes profoundly record the consequence of human activities [21]. Land-use change, especially urbanization, directly alters ecosystem structure, processes, and functions [22]. In urbanized areas, the loss or change in vegetation presents a considerable risk to biodiversity [23]. The net primary productivity (NPP) serves as a crucial indicator for assessing the carbon-fixation capacity of vegetation. It embodies the equilibrium between the carbon assimilated through photosynthesis and the carbon consumed during plant autotrophic respiration [24]. NPP plays a vital role in global carbon processes, being a crucial factor in carbon dynamics worldwide [25]. Land-use change, especially urban sprawl, is a key factor influencing NPP [22,26,27]. Factors such as logging, fires, population pressure, infrastructure development, poor drainage, and rapid urbanization directly or indirectly affect the distribution and change in NPP [16,26]. In addition, the transformation of forests into other land categories, changes in wetlands, degradation of agricultural soils, and changes in rangeland productivity have also profoundly affected the spatial–temporal variations in NPP [28,29]. Consequently, the dynamics of land use and NPP have become critical elements in the study of regional ecological environments.

In 1975, Helmut Lieth and Robert H. Whittaker systematically defined the concept of NPP for the first time as the quantity of CO₂ assimilated per unit of time, per unit of area by a plant via CO₂ assimilated by photosynthesis minus the remaining organic matter consumed by autotrophic respiration, a crucial component in the carbon cycle [30]. Currently, three primary models are used for estimating NPP: climate models [31], process-based models [32], and empirical models [33]. The Carnegie–Ames–Stanford Approach (CASA) model is among the most commonly employed models in NPP research because of its simple structure, few input parameters, easy calculation, and capacity to precisely model the impact of light and moisture on vegetation [34]. The CASA model has been extensively applied in simulating NPP across diverse ecological systems covering different regions of the world [35].

Research shows that from the 1980s to the early 21st century, global terrestrial NPP demonstrated a general upward trajectory. Nevertheless, the pace of growth differed between various regions and ecosystem categories [36,37]. In recent years, NPP trends have become more complex as global climate change has intensified [31]. For example, NPP growth slows down at high latitudes, while in the tropics, NPP declines due to intense weather conditions including droughts and heatwaves [36,38]. Furthermore, human activities, especially nitrogen input, alterations in land management, and greenhouse gas emissions, intensify the effects on NPP [39]. The present study demonstrates that changes in NPP display a clear nonlinear pattern, i.e., they do not show a direct linear relationship but fluctuate in response to changing environmental conditions [40]. These fluctuations reflect the complex interactions between NPP and several factors such as climate, land use, vegetation type, soil nutrients, etc. [36].

Global warming and precipitation patterns have significantly impacted NPP, especially changes in plant species distribution, growth cycle, and rate [41]. Climate change has the potential to reduce biodiversity and impair soil quality, thereby subsequently influencing NPP [42]. As global climate change continues to intensify, the rise in carbon dioxide levels, one of its key characteristics, may enhance plant photosynthesis, thereby promoting an increase in NPP [43]. However, increased greenhouse gas concentrations may also hurt climate change, affecting NPP [44]. Topographic factors such as altitude, slope, and hydrological conditions also profoundly affect vegetation growth and NPP [45,46]. Therefore, NPP dynamics result from the combined effects of multiple factors, necessitating a holistic assessment of their interactions [47].

Recent research has emphasized the substantial influence of LULC alterations on NPP, especially within urbanized areas. For example, the transformation of natural lands into city or agricultural use often leads to substantial decreases in vegetation cover, directly affecting NPP dynamics. Research has demonstrated that urbanization and land-use changes not only modify the landscape but also affect local and regional climatic conditions, thereby complicating the response of NPP to these alterations [48]. Similarly, a thorough review of LULC across mainland Southeast Asia during the last thirty years offers critical insights into the intricate interplay among land use, climate change, and NPP [49]. Land-use changes exert a variety of effects on NPP. The growth of urban areas, agricultural intensification, and various anthropogenic actions directly modify the distribution of plant life and influence primary productivity [50]. Land-use change indirectly affects NPP by changing surface albedo, soil moisture [51], etc. It was found that different land types respond differently to NPP and that policy interventions can also effectively change land-use patterns, affecting NPP [52,53].

In this study, the CASA model was chosen as the main tool for vegetation NPP research due to its advantages of requiring minimal input parameters, being strongly driven by remote sensing data, and offering high efficiency in dynamic analysis. Through the consistent resampling of 30 m land use data alongside additional remote sensing and climate data to a unified precision, the precise assessment of high-resolution NPP spatial distribution was achieved, substantially enhancing the granularity of urban ecological monitoring in developing countries.

This study identifies three key patterns in Harbin’s Net Primary Productivity (NPP): First, policy-driven NPP fluctuations occurred, rising 92% (2010–2015) after ecological programs but collapsing 79.5% (2015–2020) when policies ended, reflecting unstable governance in developing regions. Second, spatial patterns show northern forests (Shangzhi, Wuchang) remained high-productivity zones, while southern cities (Daoli, Nangang) and farmlands (Shuangcheng, Bayan) consistently underperformed, pointing to urban sprawl and intensive farming as main causes. Third, NPP stability depends on two thresholds: land-use intensity (loss > 0.8) and scale (rate > 3% + loss/gain ratio > 3:1). Crossing either threshold causes irreversible damage, worsened by cold climates and dense urban development in black soil areas.

2. Materials and Methods

2.1. Description of the Study Region

Harbin, the capital of Heilongjiang Province, is situated in the south-central region of the province in northeastern China, at coordinates 125°42′–130°10′ E and 44°04′–46°40′ N. The city lies within the central expanse of northeastern China, with elevations ranging between 110 and 180 m above sea level, covering a municipal area of approximately 53,100 square kilometers (Figure 1). In this study, Harbin refers to the entire administrative region of Harbin Prefecture-level City, which includes all its districts (such as Daoli, Nangang, Dao Wai, Ping-fang, Songbei, Hulan, Acheng, and Shuangcheng), as well as county-level cities (such as Shangzhi and Wuchang) and counties (such as Binxian, Fangzheng, Yilan, Bayan, Mulan, Tonghe, and Yanshou). The terrain around Harbin is varied, with low hills in the south and east and relatively flat in the north and west. The Songhua River runs through the city from southeast to northwest. Harbin has a temperate continental monsoon climate with long, cold winters and warm, short summers, with annual precipitation of about 500 to 600 mm and high annual evaporation. The region has a wide variety of landforms, with vegetation consisting mainly of temperate coniferous forests, deciduous broadleaf forests and grasslands, and various kinds of soil, including black soils, meadow soils, and swampy soils. Cultivated land and urban construction land are the region’s primary forms of land use.

2.2. Data Sources

2.2.1. NDVI Data

The Normalized Difference Vegetation Index (NDVI) data employed in this research came from the Center for Resource and Environmental Science and Data (https://www.resdc.cn/, accessed on 28 August 2024), offering a spatial resolution of 250 m by 250 m. This dataset was released by Jixi Gao [52,53], based on 250 m 16 days NDVI and 250 m 16 days pixel reliability of Aqua/Terra-MODIS satellite sensor MOD13Q1, along with land-use data. The data acquisition process involved several steps, including the preliminary reconstruction of similar ground object noise pixels in single-phase images, S-G filtering of long time series images [54], retention of high-quality pixels, maximum synthesis method for 16 days per month, and stitching according to the Chinese range.

2.2.2. Land-Use/Land-Cover Data

The land-use/cover data came from Earth System Science data, featuring a spatial resolution of 30 m by 30 m. The dataset was released by Jie Yang et al. at Wuhan University [55], and they collected training samples by combining stabilized samples from the China Land-Use/Cover Dataset (CLCD) and visually interpreted samples from satellite time-series data, Google Earth, and Google Maps [55]. Furthermore, multiple temporal indicators from all accessible Landsat data were constructed and fed into a random forest classifier to obtain the classification outcomes. The data consisted of farmland, forest, shrub, grassland, water, snow and ice, bare ground, impervious surface, and wetland.

2.2.3. Temperature and Precipitation Data

The temperature and precipitation data came from the National Tibetan Plateau Science Data Center (http://data.tpdc.ac.cn, access on 2 September 2024), with a spatial resolution of 1 km by 1 km; the dataset was released by Shouzhang Peng Scholar [56] and generated by combining the global 0.5° climate dataset from the Climatic Research Unit (CRU) and the global high-resolution climate dataset from WorldClim through the Delta approach. A spatial downscaling process was carried out to create the downscaled data in China. To ensure the data’s reliability, it was validated using data from 496 independent meteorological observation points, and the validation confirmed its credibility.

2.2.4. Solar Radiation Data

The solar radiation data that were utilized in this paper were retrieved from the Google Earth Engine (GEE) platform. The GEE search function was used to retrieve and filter the datasets that meet the research requirements and obtain key information such as their temporal extent and spatial resolution. The required data (based on temporal and spatial extent) were filtered using Python 3.7 code in the GEE code editor, and export parameters were configured in GeoTIFF format. The data is exported to Google Drive through the GEE export tool and further spatially analyzed and processed in ArcGIS. This method ensures data extraction accuracy and format compatibility to meet the research needs for high-resolution solar radiation data.

The projected coordinate system of all data was converted to WGS_1984_UTM_Zone_52N, and the spatial resolution was resampled to 30 m.

2.3. Research Methods

2.3.1. NPP Estimation

In this paper, we estimated NPP by means of the improved Carnegie–Ames–Stanford Approach (CASA) model by Wenquan Zhu et al. [57] The CASA model is a typically light energy utilization model that operates based on remote sensing technology and meteorological data [58]. The model uses remote sensing data and incorporates climate factors such as temperature, precipitation, and solar radiation, along with land cover physiological parameters. NPP is described as the product of photosynthetically active radiation (APAR) and light energy consumption [59]. The CASA model has demonstrated its suitability for NPP estimation across different vegetation types at different scales with higher accuracy [60], and the formula of the model is as follows [61]:

$$\begin{array}{c}\mathrm{N}\mathrm{P}\mathrm{P}\left(x,t\right)=\mathrm{A}\mathrm{P}\mathrm{A}\mathrm{R}\left(x,t\right)\times \mathsf{\epsilon }\left(x,t\right)\end{array}$$

(1)

where APAR (x, t) denotes the photosynthetically active radiation absorbed by image element x in month t (g Cm⁻² month⁻¹), and ε (x, t) denotes the actual light energy utilization rate absorbed by image element x in month t (g C/MJ).

APAR can be calculated using Equation (2) [62]:

$$\begin{array}{c}\mathrm{APAR}(x,t)=\mathrm{SOL}(x,t)\times \mathrm{FPAR}(x,t)\times 0.5\end{array}$$

(2)

where SOL (x, t) denotes the total solar radiation at image element x in month t (g Cm⁻² month⁻¹), and FPAR (x, t) is the proportion of incident light and effective radiation absorbed by the vegetation.

$$\begin{array}{c}\mathsf{\epsilon }\left(\mathrm{x},\mathrm{t}\right)=T_{\epsilon 1}\left(\mathrm{x},\mathrm{t}\right)\times T_{\epsilon 2}\left(\mathrm{x},\mathrm{t}\right)\times W_{\epsilon }\left(\mathrm{x},\mathrm{t}\right)\times {\epsilon }_{max}\end{array}$$

(3)

where $\epsilon$ is the efficiency of the conversion of the photosynthetically active radiation into organic carbon, $T_{\epsilon }$ is the temperature stress, and $W_{\epsilon }$ is the water stress.

2.3.2. Characterization of the Spatial Distribution of NPP

In this study, the spatial autocorrelation model was employed to examine the spatial distribution patterns of vegetation’s net primary productivity and observe whether there is a spatial aggregation feature in its distribution. Spatial autocorrelation, a type of spatial distribution feature, is utilized to unveil the existence of relationships among spatial data. It is typically categorized into global spatial autocorrelation and local spatial autocorrelation.

• Characterization of the global spatial distribution of NPP

In this study, we make use of the global spatial autocorrelation to reveal the interdependence of net primary productivity of vegetation in neighboring regions [63], and the overall trend of spatial autocorrelation of net primary productivity of vegetation in Harbin City was revealed by calculating Moran’s I index, which was calculated as follows [64]:

$$\begin{array}{c}\mathrm{M}\mathrm{o}\mathrm{r}\mathrm{a}\mathrm{n}’\mathrm{s}\mathrm{I}={\displaystyle \frac{\mathrm{n}{\sum }_{i=1}^n{\sum }_{j=1}^n{W}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{\left({\sum }_{i=1}^n{\sum }_{j=1}^n{W}_{ij}\right){\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}}\end{array}$$

(4)

where n is the number of image elements; $x_i$ and $x_j$ represent the spatial NPP at i and j; ${\mathrm{W}}_{\mathrm{ij}}$ represents the spatial weight matrix; and $\overline{x}$ is the average value of NPP. Moran’s, I index varies from −1~1, and less than 0 indicates a negative correlation. More than 0 indicates a positive correlation, and the larger the absolute value, the more pronounced the spatial autocorrelation is. The significance of the results can be tested by the Z score, and it is only significant when p < 0.05; if p ≥ 0.05, it is not meaningful, and it is not possible to judge whether there is a spatial correlation, when $\left|\mathrm{Z}\right|$ ≥ 2.58, the spatial distribution of NPP is highly significant spatial autocorrelation.

• Characterization of the local spatial distribution of NPP

In this study, local Moran’s I is used to reflect the local spatial correlation characteristics of NPP, which can be used to recognize significant statistical hot spots, cold spots, and spatial outliers [65]. The formula of local Moran’s I is as follows:

$$\begin{array}{c}\mathrm{M}\mathrm{o}\mathrm{r}\mathrm{a}{\mathrm{n}}^{‘}\mathrm{s}{I}_i={\displaystyle \frac{\mathrm{n}\left(x_i-\overline{x}\right){\sum }_{j=1}^n{W}_{ij}\left(x_i-\overline{x}\right)}{{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}}\end{array}$$

(5)

where ${\mathrm{I}}_{\mathrm{i}}$ is the local Moran’s I value of the ith image element, n is the total number of image elements, ${\mathrm{x}}_{\mathrm{i}}$ and ${\mathrm{x}}_{\mathrm{j}}$ are the NPP values of the ith and jth image elements, $\overline{x}$ is the average value of the NPP of all image elements, and ${\mathsf{\omega }}_{\mathrm{ij}}$ is the spatial weight between the ith and jth image elements. The result of local spatial autocorrelation analysis generates the LISA clustering map, which can obtain five spatial attribute relationships: high–high clustering, low–low clustering, high–low anomalies, low–high anomalies, and non-significant.

2.3.3. NPP Stability Analysis

In this study, NPP stability was analyzed using the CV index. The CV index, known as the Coefficient of Variation (CV), is a statistical measure of the degree of dispersion of data. It is expressed as the ratio of the standard deviation of a variable to its mean value, usually in the form of a percentage. The larger the value of CV, the greater the degree of dispersion of the NPP of the exact likelihood, i.e., the greater the degree of deviation of the NPP relative to the mean value; and the smaller the value of CV, the less the degree of dispersion of the same likelihood of the NPP, and the more stable it is. The formula calculates the CV:

$$\begin{array}{c}CV={\displaystyle \frac{\sqrt{\frac{1}{\mathrm{n}-1}{\sum }_{i=1}^n\left(x_i-\overline{x}\right)}}{\overline{x}}}\end{array}$$

(6)

where CV is the coefficient of variation in net primary productivity (NPP) of vegetation, i is the year, $x_i$ is the NPP in year i, and $\overline{x}$ is the mean value of NPP from 2000 to 2020.

2.3.4. Analysis of the Impact of Land-Use Change on NPP

• Conversion between land-use types

This study utilizes a land-use transfer matrix to quantify the transfer direction and area transferred between different land-use types.

The land-use transition matrix applies the Markov Model to analyze changes in land use [66]. It can numerically illustrate the transformation among various land categories and indicate the transition rate among them. Its mathematical expression is.

$$\begin{array}{c}{S}_{ij}=\left(\begin{array}{ccc}{S}_{11}& \dots & S_{1n}\\ ⋮& \ddots & ⋮\\ S_{n1}& \dots & S_{nn}\end{array}\right)\end{array}$$

(7)

where S represents the area of land-use type; n is the land-use type; and S_ij denotes the area transferred from land-use type i to land-use type j.

• Amount of change in NPP due to land-use change

The amount of change in NPP due to land-use transfer is calculated according to the direction of transfer and region of transfer of land-use types calculated from the land-use transfer matrix, combined with the average NPP for each land-use type each year. The calculation formula is as follows:

$$\begin{array}{c}∆{NPP}^{\left(t\to t+∆t\right)}=\left({\overline{NPP}}_i^{\left(t+∆t\right)}-{\overline{NPP}}_j^{\left(t\right)}\right)S_{ij}^{\left(t\to t+∆t\right)}\end{array}$$

(8)

where S_ij represents the region shifted from land-use type i to land-use type j; $\overline{{NPP}_i}$ denotes the average value of NPP for the ith land-use type; and t denotes the tth year.

The methodologies used in the current study are shown in the Figure 2.

3. Results

3.1. Model Accuracy Verification

The dependability of the predicted NPP outcomes can be confirmed through on-site inspections or secondary approaches (Figure 3). Because of the large size of the study area and the complexity of vegetation types, it is not easy to obtain NPP from field observations, and the latter was used in this study to verify the NPP outcomes estimated grounded in the CASA model. The MOD17A3 product has gained widespread acceptance globally, and it can act as a benchmark for estimating NPP to assess the accuracy of these estimated NPPs indirectly. Thirty thousand valid MOD17A3 NPP values were chosen at random in the research region and correlated with the assessed NPP at the relevant time and place, with a strong correlation of R² of 0.73 (p < 0.01), which makes the simulation results reliable in general.

Verification results show that the R² value reaches 0.75, indicating that the NPP estimation results have high reliability. However, this inter-model validation method relies on the accuracy of MOD17A3 products, while MOD17A3 itself may have an underestimation in high-latitude areas and sparsely vegetated urban areas [67,68]. Therefore, the effect of remote sensing data uncertainty on NPP estimation also needs to be considered in practical applications. In addition, the sensitivity of the input parameters of the CASA model to the results is also an issue of concern. The dependence of the model on photosynthetically active radiation (PAR) and temperature may lead to bias in the estimation results under extreme climatic conditions [69].

3.2. Spatial and Temporal Distribution of Net Primary Productivity of Vegetation

NPP in Harbin City gradually increased from the east to the west, and its spatial distribution was closely related to land-use/cover type, revealing significant differences in NPP among different land types [70,71] (Figure 4).

3.2.1. Temporal Variation Characteristics of NPP

In Harbin, from 2000 to 2020, the annual NPP (net primary productivity) of various land-use types demonstrated different changing patterns (Figure 5). The NPP of wetlands fluctuated considerably, with a decline followed by a recovery in 2020. City NPP showed a downward trend, stabilizing overall. Barren land exhibited a fluctuating pattern, with some increase but little overall change. The NPP of water bodies slightly decreased, with minimal change over the period. Grassland NPP steadily increased, showing a noticeable growth trend. Forest NPP remained relatively stable with slight declines but stayed at a high level, while farmland NPP showed a slight decrease with some fluctuation. Overall, grassland and wetlands showed significant NPP increases, while water bodies and barren land experienced declines, and forests and farmlands remained relatively stable.

Comparing NPP in Harbin City between 2000 and 2020, we found significant regional differences and interannual fluctuations (

Table 1. Annual vegetation NPP change statistics.

NPP Change.	2000 to 2005		2010 to 2005		2015 to 2010		2020 to 2015
	Area/km2	Percentage	Area/km2	Percentage	Area/km2	Percentage	Area/km2	Percentage
$-1000<∆NPP$ ≤ −750	0.00	0.000%	0.00	0.000%	0.00	0.000%	50.23	0.095%
$-750<∆NPP$ ≤ −500	4.72	0.009%	23.80	0.045%	3.48	0.007%	43.69	0.082%
$-500<∆NPP$ ≤ −250	259.87	0.489%	215.01	0.405%	33.51	0.063%	559.17	1.052%
$-250<∆NPP$ ≤ 0	32,015.25	60.251%	41,266.63	77.661%	4148.08	7.806%	41,604.78	78.297%
$0<∆NPP$ ≤ 250	20,848.84	39.236%	11,619.81	21.868%	48,768.07	91.778%	10,772.94	20.274%
$250<∆NPP$ ≤ 500	7.47	0.014%	10.68	0.020%	180.76	0.340%	94.26	0.177%
$500<∆NPP$ ≤ 750	0.75	0.001%	0.97	0.002%	0.98	0.002%	11.55	0.022%
$750<∆NPP$ ≤ 1000	0.00	0.000%	0.00	0.000%	2.01	0.004%	0.28	0.001%

and Figure 6). Between 2000 and 2020, NPP in Harbin City experienced considerable volatility. From 2000 to 2010, the city’s NPP displayed an overall decline, with 60.75% of the area experiencing a decrease in NPP between 2000 and 2005, and this decrease was predominantly concentrated in the central and eastern regions of the city. Conversely, the western part of the city showed an increase in NPP. When considering the distribution of land-use types in Harbin, it is evident that the central and eastern regions are predominantly covered by forest land, where NPP declined during this period. In contrast, the rise in NPP in the western part of the city is due to the fact that cultivated land and urban development areas are dominant there, both of which experienced an increase in NPP during the same time frame. Subsequently, between 2005 and 2010, the declining trend in NPP became more severe, with the area of impact expanding to 78.11%, particularly in the central and western regions. Meanwhile, the area with an increase in NPP shrank to 21.29% in the northeast. This result demonstrates, during this period, all land-use types, except for barren land, experienced a decline in NPP. This trend suggests that the areas of NPP decline are expanding over time, while the growth areas are gradually concentrating towards the urban fringe. However, from 2010 to 2015, there was a significant positive shift in NPP in Harbin, with 92.12% of the area realizing an increase in NPP. This shift happened because, during the same period, the yearly mean NPP of all Land categories in Harbin showed an overall improvement. This change signifies the recovery and enhancement of ecosystem productivity. Unfortunately, this positive trend has not been sustained. Between 2015 and 2020, NPP again showed a decreasing trend, affecting 79.52% of the regions, while only 20.27% showed an increase in NPP. This decline is mainly due to the decrease in the annual average NPP of both forest and farmland during this period. This may reflect the sensitivity of ecosystems to environmental changes and the dynamics of ecosystem productivity over time.

3.2.2. Spatial Variation Characteristics of NPP

To reveal the spatial distribution characteristics of vegetation net primary productivity (NPP), this study used the global Moran index to analyze the spatial aggregation or dispersion patterns of NPP distribution in Harbin. The analysis results showed that the global Moran’s I values were 0.83, 0.85, 0.85, 0.87, and 0.74 in the five study years of 2000, 2005, 2010, 2015, and 2020, respectively, which were close to or exceeded 0.8, indicating that the spatial distribution of NPP had a significant positive correlation (

Table 2. Spatiotemporal autocorrelation of net primary production between 2000 and 2020.

	2000	2005	2010	2015	2020
Moran I	0.83	0.85	0.85	0.87	0.74
Z-score	744.96	763.00	759.37	774.10	783.16
p-value	0.00	0.00	0.00	0.00	0.00

). The corresponding Z-scores were 744.96, 763.00, 759.37, 774.10, and 783.16, which far exceeded the critical values, further verifying the strength of spatial autocorrelation. Meanwhile, the p-values for all years are less than 0.01, indicating that these spatial autocorrelation results are statistically significant, proving that Harbin NPP exhibits a significant spatial clustering effect during the study period.

The global Moran’s I value (close to or over 0.8), which has been maintained at a high level for a long time, suggests that the NPP in Harbin has a strong positive spatial correlation, showing a clear aggregation pattern. This result is consistent with the “spatial autocorrelation” theory in ecology, which suggests that ecosystems in neighboring areas tend to be in similar states [72]. However, Moran’s I value decreased to 0.74 in 2020, indicating that the spatial aggregation effect of NPP was weakened. This change may be related to changes in ecosystem patterns under the cumulative impact of land-use transformation, urban expansion, as well as climate shifts [73]. This trend suggests that although NPP in Harbin City generally shows spatial aggregation, the spatial autocorrelation has weakened in the context of land use and climate change in recent years.

The study performed a local spatial autocorrelation analysis of the NPP values in Harbin City during the study period to uncover the detailed characteristics of its spatial arrangement. The analysis shows that the spatial variation in NPP in Harbin City presents a distinct pattern characterized by high–high aggregation and low–low aggregation. Over the period from 2000 to 2020, the proportions of NPP high–high aggregation areas are 37.37%, 37.22%, 36.08%, 36.29%, and 34.89%, respectively (

Table 3. NPP spatial autocorrelation variations.

	2000		2005		2010		2015		2020
LISA-CL	Area/km2	Percentage	Area/km2	Percentage	Area/km2	Percentage	Area/km2	Percentage	Area/km2	Percentage
Not Significant	11,425.27	21.60%	14,069.15	26.62%	14,675.05	27.75%	11,056.72	20.92%	10,149.64	19.23%
High–high	19,768.48	37.37%	19,675.97	37.22%	19,081.94	36.08%	19,183.67	36.29%	18,419.49	34.89%
Low–low	21,184.06	40.05%	18,514.52	35.03%	18,596.77	35.17%	22,088.64	41.79%	23,689.94	44.88%
Low–high	404.95	0.77%	503.11	0.95%	399.60	0.76%	433.72	0.82%	459.49	0.87%
High–low	112.01	0.21%	95.11	0.18%	129.76	0.25%	95.10	0.18%	67.21	0.13%

). These areas are mainly distributed in the north, center, and south of the urban area, particularly in the north and southeast, forming a dense agglomeration with topography dominated by mountains and forests (e.g., Shangzhi City, Wuchang City) with high natural vegetation cover and well-preserved ecosystems (Figure 7). Studies have shown that the NPP of forest ecosystems is significantly higher than farmland or urban areas [74], which is more likely to form high NPP aggregation [75]. In addition, these regions have lower levels of urbanization, lower population densities, and weaker industrial and agricultural development intensity. For example, Mulan County and Bin County have a sparse distribution of lodging and limited expansion of tourism and construction land, reducing vegetation destruction, thus presenting the characteristics of a high concentration of NPP [76].

3.3. Spatial Stability of the NPP

To explore the spatial stability of NPP, the coefficient of variation (CV index) of NPP within the research region was calculated based on the image metric scale. The degree of NPP fluctuation was classified into five grades by drawing on the grading standards of the CV index used by the previous authors to analyze the patterns of fluctuating changes in NPP within the study area. (

Table 4. Area statistics of variation coefficient of NPP in Harbin.

Coefficient of Variation	Degree of Variation	Area/km2	Percentage of Total Area
CV ≤ 0.05	Low volatility fluctuations	17,351.80	32.89%
0.05 < C ≤ 0.1	Relatively low volatility fluctuations	29,071.08	55.11%
0.1 < CV ≤ 0.15	Moderate volatility fluctuations	3869.12	7.33%
0.15 < CV ≤ 0.2	Relatively high volatility fluctuations	857.27	1.63%
CV > 0.2	High volatility fluctuations	1603.24	3.04%

). The mean CV index value of NPP in the study area over the period from 2000 to 2020 was 0.077, suggesting that the overall NPP change in the study area is small [75]. The figure shows the distribution across space of the CV index of NPP, from which it can be seen that the CV index of NPP is negligible in the majority of the study area (Figure 8). The low fluctuation area with a CV index < 0.1 covers 46,400 km², accounting for 88% of the total area of the study area, indicating that most of the NPP in the study area is in a relatively stable state. The areas with medium fluctuation change in NPP (0.1 < CV ≤ 0.15) and relatively high fluctuation change (0.15 < CV ≤ 0.2) are mainly in the northwestern plains, with the areas of 3869.12 km² and 857.27 km², accounting for 7.33% and 1.63% of the study area. The distribution of areas with medium and relatively high fluctuation change is more dispersed, primarily focused in the construction land, which is associated with land-use and human activities [76]. The areas with high fluctuation changes (CV > 0.2) are primarily focused in the Songhua River basin and scattered in the southeast of the research region, accounting for 3.04%, which may be related to changes in hydrothermal conditions in the river basin, and these changes may affect vegetation productivity [77,78].

3.4. Impact of Land-Use Change on Net Primary Productivity of Vegetation

Figure 9 illustrates how land composition varies across different land-use categories over time.

Land-use changes in Harbin City from 2000 to 2020 showed significant stage characteristics. The dynamic attitude index of land-use types in each period was 2.81% (2000–2005), 3.33% (2005–2010), 3.65% (2010–2015), and 11.91% (2015–2020), indicating that the reorganization of the land-use pattern in the later period accelerated significantly. Cultivated land, watersheds, and construction land show a continuous expansion. In contrast, forest land, grassland, bare land, and wetland areas shrink, revealing the double driving effect of urbanization and agricultural activities.

The evolution of NPP in Harbin City from the year 2000 through 2020 showed significant spatial and temporal heterogeneity. In the study period, changes in NPP for different land-use types and their spatial transitions constituted the double driving factors of regional NPP fluctuations (

Table 5. Effects of land-use change on vegetation net primary productivity.

Time Period	Direction of Land-Use type Transfer	Transfer Area /(m2)	NPP Change /(106 g C·m−2 year−1)
From 2000 to 2005	Farmland	26,613,424,800	9577.91
	Farmland to Forest	506,003,400	127,667.66
	Farmland to City	208,542,600	−22,980.45
	Forest to Farmland	598,843,800	−168,461.41
	Forest	22,733,523,900	−667,571.27
	Forest to City	2,985,300	−1169.84
	Grassland to Farmland	9,126,000	505.28
	Grassland to Forest	244,800	75.23
	Grassland	45,814,500	469.93
	Water to Farmland	20,208,600	5588.85
	Water to Forest	1,774,800	937.99
	Water	491,998,500	−21,386.88
	Barren to Farmland	1,255,500	354.40
	Barren	9,774,000	−598.84
	Barren to City	2,019,600	346.81
	City to Farmland	230,400	28.14
	City to Water	13,285,800	−2629.02
	City	1,479,651,300	17,149.83
	Wetland to Farmland	6,720,300	−761.70
	Wetland to Forest	260,100	36.05
	Wetland	20,295,900	−209.57
From 2005 to 2010	Farmland	26,141,801,400	−673,897.27
	Farmland to Forest	652,275,000	137,432.77
	Farmland to City	197,883,900	−26,735.21
	Forest to Farmland	506,565,000	−140,685.55
	Forest to City	4,698,900	−1818.72
	Grassland to Farmland	27,427,500	530.20
	Grassland	47,063,700	−156.60
	Grassland to City	9,365,400	−842.85
	Water to Farmland	36,887,400	10,854.08
	Water	507,076,200	15,603.38
	Water to City	10,017,900	1852.53
	Barren to Farmland	1,003,500	318.88
	Barren to City	2,998,800	625.07
	Barren	4,772,700	365.71
	City to Farmland	244,800	20.75
	City to Water	36,645,300	−8803.79
	City	1,674,740,700	−41,115.23
	Wetland to Farmland	3,505,500	−451.49
	Wetland	16,790,400	−494.87
From 2010 to 2015	Farmland	26,203,603,500	1,485,193.82
	Farmland to Forest	207,911,700	77,891.18
	Farmland to City	215,888,400	−11,906.86
	Forest to Farmland	1,323,404,100	−237,944.26
	Forest	22,058,695,800	3,047,623.25
	Forest to City	5,264,100	−1535.16
	Grassland to Farmland	17,127,000	1358.81
	Grassland to Forest	1,515,600	602.14
	Grassland	37,251,900	2775.50
	Water to Farmland	15,832,800	6043.37
	Water	759,534,300	24,006.55
	Water to City	14,390,100	3883.42
	Barren to Farmland	765,000	227.83
	Barren	2,959,200	125.99
	Barren to City	1,003,500	186.64
	City to Farmland	260,100	43.18
	City to Water	18,133,200	−3338.07
	City	1,881,312,300	101,918.51
	Wetland to Farmland	5,019,300	−214.04
	Wetland	11,771,100	1047.69
From 2015 to 2020	Farmland	24,588,205,200	−60,3741.48
	Farmland to Forest	1,755,548,100	411,459.59
	Farmland to City	1,056,136,500	−111,895.26
	Forest to Farmland	1,915,096,500	−655,941.54
	Forest	20,254,353,300	−1,692,861.74
	Forest to City	50,373,000	−21,353.35
	Grassland to Forest	5,983,200	1431.23
	Grassland	2,794,500	30.63
	Grassland to City	8,897,400	−899.67
	Water to Farmland	164,236,500	53,465.26
	Water	602,616,600	17,195.02
	Water to City	61,877,700	15,107.11
	City to Farmland	893,205,000	77,956.92
	City to Forest	46,801,800	16,203.18
	City	1,135,511,100	6681.50
	Wetland to Farmland	2,641,500	−412.61
	Wetland to Forest	8,636,400	887.20
	Wetland	450,000	−10.19

Full data reference attachment (Table A1, Table A2, Table A3 and Table A4).

).

From 2000 to 2005, 54.26% of the NPP growth area, 77.25% of which originated from the ecological transition from farmland to forest land, indicated that the early stage of the farmland-to-forest policy had a significant contribution to the enhancement of carbon sinks [77,78]. This shift in land use not only increased carbon sequestration capacity but also helped to restore water retention capabilities, enhance soil stability, and improve biodiversity. The conversion of farmland to forest promoted ecosystem services by stabilizing the local climate and fostering a more balanced water cycle [79]. However, of the 45.74% NPP decline area in the same period, 73.77% was caused by the decline of productivity of the forest land itself (e.g., the decline of carbon sequestration capacity caused by the aging of mature forests), followed by the conversion of forest to farmland (18.62%) and the expansion of land for construction (2.71%). Not only do these shifts in land use reduce carbon storage, but they also endanger soil fertility and disrupt the regulation of water. Additionally, the conversion of forests to agricultural land and the growth of urban areas further break up habitats, posing risks to biodiversity and ecosystem stability [80].

Between 2005 and 2010, the NPP decline widened significantly to 98.6%, with a structural shift in the dominant drivers. The contribution rate of NPP decline in cultivated land increased to 34.81% (possibly related to continuous cropping obstacles and excessive fertilization), and the contribution rate of NPP decline in forest land was 48.43% (reflecting the unstable carbon sink function of secondary forest during recovery). These changes are indicative of an intensifying pressure on ecosystem services, particularly water regulation and soil health. The overuse of cropland and the decline of forest ecosystems during this period led to a loss of carbon sequestration potential, further exacerbating challenges related to climate regulation [81]. Although the direct contribution of construction land expansion is only 1.52%, the habitat fragmentation caused by it indirectly aggravates the vulnerability of the ecosystem [82,83,84,85].

From 2010 to 2015, the area of NPP improvement rebounded to 96.86%, with 64.11% of the increment originating from forest land productivity recovery (e.g., plantation forests entering a rapid growth period) and 31.24% from optimization of cropland management (e.g., promotion of conservation tillage technology). During this period, policy interventions like the project of returning farmland to forest and improved agricultural practices greatly enhanced carbon sink capacity, improved water conservation, and boosted soil fertility [86]. These efforts contributed positively to the recovery of ecosystem services and regional stability [87].

From 2015 to 2020, the decline of NPP dominated again (91.05%). Its driving mechanism is characterized by multiple features: the decline of forest land itself contributes 53.37% (frequent occurrence of pests and diseases and extreme weather events), forest-tillage conversion contributes 20.68% (overexploitation of economic forests), and degradation of arable land contributes 19.04% (decline in land strength due to black soil loss). These changes highlight a growing challenge for regional ecosystem services. The decline in forest net primary productivity not only reduces carbon sequestration but also weakens water regulation and biodiversity conservation functions [88]. The overexploitation of economic forests and the degradation of arable land further affect soil fertility, decreasing its ability to retain water and support diverse species [86]. Significantly, the contribution of urban land expansion increased to 4.23%, suggesting that the squeezing effect of urbanization on ecological space is intensifying. This urban expansion leads to habitat fragmentation, further reducing ecosystem resilience and threatening the long-term stability of ecological services [89,90].

In summary, NPP changes are closely tied to land-use type changes. Both land-use changes and changes in vegetation net primary productivity directly affect regional ecosystem services like carbon sequestration, water regulation, and biodiversity. Land-use planning should prioritize the restoration and preservation of forest ecosystems, as well as sustainable agricultural practices, to ensure the long-term stability of ecosystem services and to enhance carbon sink capacity. In the future, policy making should aim to achieve an equilibrium between economic advancement and ecological protection to maintain the integrity of these essential services.

4. Discussion

4.1. Drivers of NPP Temporal Variability: Policy Volatility, Urban Expansion, and Climate Feedbacks

The dramatic decline in net primary productivity (NPP) across central and eastern Harbin between 2000 and 2010 was driven by rapid urban expansion and fragmented governance. Implementation of the Harbin Urban Master Plan (2004–2020), which prioritized industrial growth, resulted in a 1.8% annual increase in construction land, predominantly through conversion of arable and forested areas [91]. This pattern aligns with NPP degradation observed in rapidly urbanizing cities of the Global South, such as Jakarta [92]. Despite China’s introduction of the Returning Farmland to Forests policy in 2003, its initial effectiveness was undermined by insufficient economic incentives for farmers and institutional conflicts between land management sectors. For instance, overlapping jurisdictional objectives in the Shuangcheng District delayed ecological restoration efforts until 2010.

In contrast, post-2010 NPP recovery was closely linked to policy refinements and climatic variability. The 2007 amendment to the Returning Farmland to Forests program, which raised subsidies to 70 RMB per mu and extended compensation periods to 8 years for ecological forests, significantly improved stakeholder participation. However, the subsequent relaxation of these policies after 2015 triggered a sharp regional NPP decline (79.52%), highlighting the transient efficacy of top-down restoration strategies in developing economies compared to incremental approaches in developed nations [93]. Concurrently, climatic drivers—particularly rising temperatures and shifting precipitation patterns in Northeast China—modulated interannual NPP fluctuations, underscoring the compound effects of anthropogenic and natural factors on vegetation productivity [94].

This study elucidates the high sensitivity and inherent unsustainability of urban ecological restoration in developing economies. Empirical evidence reveals a 92.12% surge in regional Net Primary Productivity (NPP) during 2010–2015 under intensive policy interventions, followed by a precipitous 79.52% decline post-2015 with policy deregulation. The identified “pulse-compression” pattern fundamentally diverges from the incremental restoration trajectories observed in developed nations, exposing institutional contradictions between land-use conversion and ecological governance in transitional economies. The Harbin case demonstrates that post-2015 NPP resurgence closely coupled with infrastructure-driven urbanization phases, empirically validating the law of diminishing marginal effectiveness in policy implementation.

4.2. Mechanisms of NPP Spatial Aggregation Patterns

During the analysis period, the proportions of NPP low–low aggregation areas were 40.05%, 35.03%, 35.17%, 41.79%, and 44.88%, respectively. These areas mainly include the central urban Harbin regions (e.g., Daoli and Nangang Districts) and their suburbs (e.g., Hulan District). Rapid urbanization in these areas has resulted in a large amount of natural vegetation being replaced by construction land, and urban expansion has significantly reduced vegetation cover, resulting in a lower NPP [95]. In addition, plain areas such as Bayan County and Shuangcheng District are dominated by farmland, where monocropping and frequent farming activities lead to loss of soil organic matter, reducing the carbon sink capacity of the ecosystem [96]. Meanwhile, excessive use of chemical fertilizers and pesticides may further inhibit vegetation growth, and these factors together contribute to the low NPP aggregation in these regions.

In addition, low–high and high–low aggregation types accounted for less than 1% throughout the study period, indicating that most neighboring spatial units in the study area had slight differences in NPP and a flat spatial gradient. This is mainly because the eastern mountainous regions of Harbin (e.g., Shangzhi and Wuchang) are dominated by forest cover and generally have high NPP. In contrast, the western plain areas (e.g., Hulan and Shuangcheng districts) are dominated by farmland and urban areas, with overall low NPP. The two regions are characterized by intense internal geographic homogeneity and fewer spatially abrupt changes in natural or anthropogenic drivers. This results in very few boundary areas of low–high and high–low aggregation.

4.3. Land-Use Conversion Dynamics and Agri-Ecological Trade-Offs in NPP Variation

Recently, land-use patterns in Harbin have been influenced by multiple factors, including rapid urbanization, agricultural development, ecological protection policies, climate change, and adjustments in economic activities. Urban expansion has significantly occupied cropland and forest land to meet population growth and infrastructure construction needs [97]. The interaction between agricultural expansion and the farmland-to-forest policy has led to fluctuations in the area of cultivated land [45]. Ecological restoration measures such as wetland protection have mitigated regional environmental degradation to a certain extent [98]. In addition, climate change has further changed the distribution pattern of land types by affecting soil suitability and precipitation distribution [45]. Economic restructuring has intensified the conversion of land from agriculture to industry and services, and the synergistic effect of these factors has complicated the regional land-use dynamics and had far-reaching impacts on the Net Primary Productivity (NPP) and ecosystem functions [98].

Between 2000 and 2020, the impact of land-use change on NPP in Harbin City showed significant dynamics, and the conversion between land types became the main driver of NPP fluctuations. This study found that cropland conversion to forest land contributed the most to NPP growth, which is consistent with many studies. For example, Wu et al. (2022) showed that a similar cropland conversion to forest land significantly increased regional NPP and improved soil quality in the Tibetan Plateau region [97]. What is unique in this study is the emphasis on the negative impacts on NPP caused by the reverse conversion of forested land to cropland. This finding highlights the trade-off between agriculture and ecological conservation in Harbin City. In contrast, other regions, such as the southern hilly areas, usually emphasize the environmental benefits of woodland to grassland or wetland conversion [45,98]. The frequent inter-conversion of cropland and forest land is the core of land-use change in Harbin City and the primary driver of NPP fluctuations. Conversion of forest to farmland boosts food production may result in an immediate increase in soil degradation and biodiversity decline, thereby weakening long-term ecosystem service capacity. Conversely, converting cropland to forested land improves carbon storage capacity and NPP but may affect food security. Although the contribution of urban construction land expansion to NPP reduction is relatively low (2.71% to 4.23%), its long-term cumulative effect cannot be ignored. Urban expansion is usually accompanied by permanent vegetation loss, soil confinement, and heat island effects, which threaten regional ecosystem productivity [45,98]. The contribution of urban expansion increased from 2015 to 2020, reflecting the continued pressure on ecosystems from urbanization.

4.4. Dual Critical Thresholds and Risk Early-Warning Mechanisms in Land Conversion Dynamics

This study found that Harbin triggered the ecological risk threshold (NPP loss/gain ratio reached 6.16:1) when the land conversion rate was only 3.65% (

Table 6. Land-use conversion and NPP profit and loss dynamics in Harbin (2000–2020).

Period	Land Conversion Rate (%)	Proportion of NPP Loss Area (%)	Proportion of NPP Increase Area (%)	NPP Loss per Unit of Conversion Rate (Loss/Conversion Rate)	NPP Increase per Unit of Conversion Rate (Increase/Conversion Rate)
2000 to 2005	2.81	1.76	1.05	0.63	0.37
2005 to 2010	3.33	1.94	1.39	0.58	0.42
2010 to 2015	3.65	3.14	0.51	0.86	0.14
2015 to 2020	11.91	6.26	5.65	0.53	0.47

). The ecological loss intensity per unit conversion rate (0.86) surged by 48% compared to the previous stage, significantly lower than the 52.32% development intensity threshold proposed in the study of Hebei Province, China [99]. From 2015 to 2020, the ecological restoration effect began to appear under a high conversion rate (11.91%), and the share of gain area successfully rebounded to 5.65%. The loss/gain ratio decreased to 1.11:1. This change confirmed that the ecological quality of the Songhua-Liao River Basin first declined and then increased [100].

Therefore, this study innovatively proposed a double critical threshold for land-use conversion:

Intensity Threshold: when the NPP loss per unit conversion rate exceeds 0.8 (e.g., 2010–2015), land development has entered a high ecological risk zone. This threshold is lower than in temperate cities, mainly due to the superimposed effect of freeze–thaw sensitivity of cold black soils and compact urbanization [101,102].

Scale Threshold: the self-healing capacity of the system’s net primary productivity of vegetation collapses when the land conversion rate exceeds 3%, and the loss/gain ratio is >3:1.” The rapid early warning indicator of “loss/gain ratio > 3:1” can effectively circumvent the dilemma of missing NPP monitoring data in the global South [103].

5. Conclusions

Under the global carbon-neutral framework, the Paris Agreement’s dual-carbon goal demands enhanced regional carbon sink capacity. While the EU improved carbon sequestration efficiency by 50% through forest optimization, China’s farmland-to-forest restoration contributed 1482.62 Tg of carbon sinks. However, developing countries face the “growth-carbon reduction” dilemma, with ecological governance vulnerable to policy fluctuations and market rebound. This study analyzes spatiotemporal dynamics of vegetation net primary productivity (NPP) in Harbin (2000–2020) using an improved CASA model, spatial autocorrelation, and land-use transfer matrices. It proposes a spatial planning paradigm to reconcile development–ecology conflicts in developing countries. The primary findings of this research are outlined below:

NPP Fluctuations and Policy-Driven Volatility: Harbin’s NPP exhibited a three-phase “decline-rise-decline” trend. Policy-driven restoration (2010–2015: 92.12% regional growth) and post-policy decline (2015–2020: 79.52% regional reduction) highlight the “impulse-type” vulnerability of ecological governance in developing economies.

Spatial Aggregation Patterns: NPP showed persistent high-intensity clustering (Moran’s I ≥ 0.74), forming a “high-north, low-south” structure. Forested northern mountains (Shangzhi, Wuchang) were stable high-value cores, while urban expansion zones (Daoli, Nangang) and degraded farmland (Shuangcheng, Bayan) formed low-value clusters, confirming urbanization and agricultural intensification as dominant stressors.

Dual Critical Thresholds: Land-use change impacts NPP via intensity threshold (unit conversion loss >0.8) and scale threshold (conversion rate > 3% + loss/gain ratio > 3:1). Exceeding either threshold collapses ecosystem resilience, revealing compounded risks from frost-thaw sensitivity and compact urbanization in cold black soil regions.

Drawing from the analytical results, this study suggests the subsequent focused policy suggestions:

Zonal Management: Establish forest carbon sink compensation in northern ecological reserves, enforce a 3% land conversion cap in urban buffers, and link crop rotation with carbon subsidies in western agricultural zones.

Dynamic Monitoring: Create a loss/gain ratio platform to freeze land approvals (if intensity threshold breached) and mandate triple reforestation. Integrate scale thresholds into ecological assessments with a “one-vote veto” mechanism.

Limitations and Future Directions:

The findings and suggestions of this study have valuable implications for other cities in Northeast China, especially those with abundant forest resources similar to Harbin. In order to increase the accuracy of NPP quantification, future research will concentrate on improving the accuracy of remote sensing data, which will enable more precise forecasts of NPP changes. Additionally, we plan to investigate the interactions between policy fluctuations, urban expansion, and climate change, exploring how these factors dynamically and collectively influence NPP. This will be a key area of focus in our future studies.

This study primarily focuses on the supply-side mechanisms of vegetation carbon sinks, while neglecting the dynamics of demand-side carbon emissions and the synergy between carbon sources and sinks. Data constraints have also limited the analysis of spatiotemporal carbon source-sink matching and cross-regional compensation. Future work should aim to develop bidirectional carbon source-sink models and deploy AI-driven remote sensing technologies for intelligent carbon regulation to enhance ecosystem stability.