Next Article in Journal
Optimal Sensor Placement Using Learning Models—A Mediterranean Case Study
Previous Article in Journal
Monitoring Regional-Scale Surface Deformation of the Continuous Permafrost in the Qinghai–Tibet Plateau with Time-Series InSAR Analysis
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Investigation of Long-Term Forest Dynamics in Protected Areas of Northeast China Using Landsat Data

1
College of Earth and Environmental Sciences, Lanzhou University, Lanzhou 730000, China
2
National Tibetan Plateau Data Center, Key Laboratory of Tibetan Environmental Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China
3
Academy of Plateau Science and Sustainability, Qinghai Normal University, Xining 810016, China
4
Institute of Forest Resources Information Techniques, Chinese Academy of Forestry, Beijing 100091, China
*
Author to whom correspondence should be addressed.
Remote Sens. 2022, 14(13), 2988; https://doi.org/10.3390/rs14132988
Submission received: 21 May 2022 / Revised: 12 June 2022 / Accepted: 21 June 2022 / Published: 22 June 2022

Abstract

:
Forest dynamics, including forest loss and gain, are long-term complex ecological processes affected by nature and human activities. It is particularly important to understand the long-term forest dynamics of protected areas to evaluate their conservation efforts. This study adopted the Landsat tree-canopy cover (TCC) method to derive annual TCC data for the period 1984–2020 for the protected areas of northeast China, where protection policies have been carried out since the end of the 20th century, e.g., the Natural Forest Conversion Program (NFCP). A strong correlation was found between the TCC estimates derived from Landsat and LiDAR observations, suggesting the high accuracy of TCC. Forest loss and gain events were also identified from the time series of TCC estimates. High correlations were reported for both forest loss (Producer’s accuracy = 85.21%; User’s accuracy = 84.26%) and gain (Producer’s accuracy = 87.74%; User’s accuracy = 88.31%), suggesting that the approach can be used for monitoring and evaluating the effectiveness of the NFCP and other forest conservation efforts. The results revealed a fluctuating upward trend of the TCC of the protected area from 1986 to 2018. The increased area of TCC was much larger than the decreased area, accounting for 59.68% and 40.34%, respectively, suggesting the effectiveness of protection policies. Since the NFCP was officially implemented in 1998, deforestation was effectively curbed, the area of forest loss was significantly reduced (slope: −14.24%/year), and the area of forest gain significantly increased (slope: 4.13%/year). We found that regional forest changes were mainly manifested in “forest gain after loss (forest recovery)” and “forest gain with no loss (forest newborn)”, accounting for 40.29% and 37.28% of the total area of forest change, respectively. Moreover, the forest gain area far exceeds the forest loss area, reaching 11.24 million hectares, suggesting a successful forest recovery due to forest protection.

1. Introduction

Forests are considered the most extensive terrestrial ecosystems, accounting for about 1/3 of the total land area [1]. Forests are also an important component of the global biosphere, playing an essential role in the global carbon and biomass cycle. The health of forests is closely related to the sustainability of regional ecosystems and human societies [2]. Under the combined influence of human activities, climate change, and various natural disasters [3,4,5], forests have experienced complex and profound changes at different scales. Studies have shown that global forest cover has decreased dramatically in the past few decades [6,7,8]. At the same time, increasing forest disturbance has had a huge impact on carbon storage potential [9,10,11].
With the further increase in the understanding of the importance of environmental conservation, governments of various countries have begun implementing various relevant policies. On 25 September 2015, 193 member states formally adopted 17 sustainable development goals (SDGs) at the United Nations Sustainable Development Summit. The 15th goal clearly states that by 2020, the implementation of sustainable management of all types of forests should be promoted, deforestation halted, degraded forests restored, and afforestation and reforestation substantially increased globally. During the recent United Nations Framework Convention on Climate Change at the 26th Conference of the Parties (COP26), leaders of more than 100 countries pledged to stop and reverse the trend of deforestation and land degradation by 2030 and invest more funds in protecting and restoring forests. The concept of environmental conservation has gradually gained popular support. As early as 1998, after a catastrophic flood disaster in the Yangtze and Songhua river basins, China began to implement the Natural Forest Conversion Program (NFCP) on a pilot basis and officially put it into practice in 2000 [12,13,14]. Logging of natural forests in the upper reaches of the Yangtze River and the middle and upper reaches of the Yellow River, and timber output in key state-owned forest areas, such as in northeast China and Inner Mongolia, was greatly reduced. At the same time, measures such as forest tending and natural regeneration have also been gradually implemented, aiming to drastically curb the deterioration of the environment, protect biodiversity, and promote sustainable social and economic development. The second phase of the NFCP ended in 2020, and the protection of natural forest resources has shifted to a long-term conservation phase. It is particularly urgent to comprehensively assess the changes in the natural forest protection project areas.
In past decades, remote sensing technology has developed rapidly and has been widely used as an important approach for obtaining spatio-temporal information, providing new opportunities for forest change monitoring for large areas [15]. Landsat has been the longest terrestrial Earth observation satellite series. Its earliest image can be traced back to 1972 and its entire archive can be obtained for free on a global scale. The rich data archive makes it possible to monitor long-term forest changes [16,17,18,19]. The characteristics of forest change at global [20,21,22] and regional [23] scales have been discussed, but these studies either had a short time range or focused mainly on the analysis of forest loss events and do not provide a complete analysis of forest change processes. In addition, previous studies mostly used various indices to monitor forest change, e.g., Normalized difference vegetation index (NDVI) and Normalized burn ratio (NBR) [24,25]. However, these vegetation indices do not have a clear physical meaning and cannot effectively represent the current forest cover state, are prone to interference by other types of vegetation, and lead to large errors. To solve this problem, forest change studies based on canopy cover have been gradually carried out [26]. DiMiceli et al. [27] developed the MODIS Vegetation Continuous Fields dataset, which has a spatial resolution of 250 m and is widely used in regional and global assessments. However, many land cover changes occur within patches of 250 m or less [28]. Based on this, Sexton et al. [29] constructed a method for extracting tree-canopy cover (TCC) based on long-term Landsat data and published a global dataset with a resolution of 30 m. Feng et al. [30] validated the dataset and showed that the forest cover rate was 91% accurate, and the forest change was over 88% accurate, satisfying the requirements for assessing forest spatio-temporal variation characteristics.
Northeast China has the largest relatively continuous forest region in the country and one of its most important carbon sink areas. It is also a key area for the implementation of the NFCP. In this paper, we derived the annual TCC of the NFCP in northeast China from 1984 to 2020 based on long time series of Landsat images and identified forest change. We combined the spatio-temporal changes of TCC with forest loss and gain events to comprehensively analyze forest change characteristics in the NFCP of northeast China over the past three decades, providing the scientific basis for its forest management and sustainable development.

2. Study Area

The NFCP in northeast China is mainly distributed in the Daxing’anling, Xiaoxing’anling and Changbai Mountains. Located between 41°21′–53°33′N, 118°47′–133°58′E (Figure 1), with an altitude range of 9–2572 m above sea level and a total area of about 46.56 million hectares. The administrative area includes 26 counties/cities in Heilongjiang Province, 14 counties/cities in Jilin Province, and 8 counties/cities in Inner Mongolia. It has a temperate continental monsoon climate with an average annual temperature range of −4.7 °C–10.07 °C [31] and average annual precipitation range of 310–750 mm, 75% to 85% of which is concentrated from June to October. The main geomorphic types include mountains, hills, plains, and plateaus [32]. This area is a key forest region of China, with vast primeval forests, including evergreen needleleaf, deciduous needleleaf, deciduous broadleaf, and mixed forests [33], providing a good living environment for wild animals and plants, with abundant biodiversity.

3. Data and Methods

3.1. Remote Sensing Data

We chose the Landsat Collection 2 level-1 images acquired by Landsat 5 TM, Landsat 7 ETM + and Landsat 8 OLI during the leaf-on season (May to September) from 1984 to 2020. Images with >90% of cloud cover were excluded. A total of 7564 Landsat images were selected for the 64 World Reference System (WRS)-2 tiles covering the study area. The selected Landsat images were downloaded from the EarthExplorer interface (https://earthexplorer.usgs.gov/, (accessed on 1 June 2021)) and radiometrically calibrated to reflectance [18].
The MODIS Vegetation Continuous Field (VCF) annual dataset with a spatial resolution of 250 m was retrieved from the National Aeronautics and Space Administration (https://ladsweb.modaps.eosdis.nasa.gov/, (accessed on 1 June 2021)). MODIS VCF dataset, publicly available worldwide, provides three surface cover components: percent tree cover, percent non-tree cover, and percent bare. In this research, we selected the percent tree cover data and obtained 21 years of data for the study area from 2000 to 2020.
The Global Ecosystem Dynamics Investigation (GEDI) LiDAR and Ice, Cloud and Land Elevation Satellite–2 (ICESat-2) observations were used to validate the estimated TCC values. The GEDI dataset is publicly available from EARTHDATA (https://earthdata.nasa.gov/, (accessed on 1 October 2021)) and provides new spaceborne vegetation canopy structure information. The canopy cover inversion of GEDI directly uses the cover parameter of L2B product, and this parameter represents the percent of the ground covered by canopy material in a vertical projection [34]. This study used the observation data of GEDI in the Greater Khingan Mountains for July 2019, with a total of 13,144 observation points (Figure 1). The ICESat-2 dataset comes from the National Snow and Ice Data Center (https://nsidc.org/, (accessed on 1 October 2021)). We obtained the ATL08 products of the study area for July and August 2019. The products provide relevant canopy and topography indicators that can be used to invert vegetation cover. ATL08 canopy cover is defined as the ratio of canopy photons to the total number of signal photons (terrain + canopy photons) within each 100 step [35]. To remove the influence of atmosphere and terrain, we filtered the data according to their cloud cover and terrain slope and retained 6116 observation points (Figure 1).
The Digital Surface Model (DSM) dataset was obtained from the Japan Aerospace Exploration Agency (https://www.eorc.jaxa.jp/, (accessed on 1 October 2021)) with a spatial resolution of 30 m.
All the data were reprojected to an Albers Equal Area Conic projection (WGS84).

3.2. Tree-Canopy Cover Estimation

We used the method proposed by Sexton et al. [29] establishing a regression tree model between the long-term Landsat images and the MODIS VCF to downscale the tree cover data and obtained annual TCC images for the study area from 1984 to 2020. Tree cover can be expressed as a piecewise linear function of surface reflectivity and temperature:
C i , t =   f X i , t +   ε
where X is the vector of the estimated surface reflectivity and temperature; ε is the estimation error generated by a regression tree model f applied to X . The subscript i denotes the pixel’s position in space, and the pixel index t refers to its position in time, indexed by year. The model was fit to spatiotemporally coincident training data as response and spectral measurements using a CatBoost regression tree algorithm (https://catboost.ai/, (accessed on 1 June 2021)) from the coincident Landsat images as covariates and then applied to each complete Landsat image to produce the map of estimates. The Landsat based tree-canopy cover estimates within a year at each pixel were composed following the best pixel composite recommended by [29].

3.3. Forest Change Detection

The bi-temporal class-probabilities model was used for forest change detection. Sexton et al. [36] proposed this method in 2015, and it has been widely used in forest change analysis. When comparing the forest cover probability detected at two times, we may find the following four dynamic categories: stable forest (FF), stable non-forest (NN), forest gain (NF) and forest loss (FN) [36]. By calculating these dynamic probabilities, the change state of a forest at this location can be determined.
Firstly, the probability of a certain location belonging to the forest land cover class is:
p F   = def   p c   >   c * = c * 100 p c dc
where p F is the probability that this point belongs to forest, and c * is the threshold of TCC, and 30% TCC was chosen in this study.
To calculate the probability of each forest dynamic at that location, only the following joint probability is required:
p FF i =   p F i , t 1 , F i , t 2 =   p F i , t 1 ×   p F i , t 2
p NN i =   p N i , t 1 , N i , t 2 = 1   p F i , t 1 × 1   p F i , t 2
p NF i =   p N i , t 1 , F i , t 2 = 1   p F i , t 1 ×   p F i , t 2
p FN i =   p F i , t 1 , N i , t 2 =   p F i , t 1 × 1   p F i , t 2
where p FF i , p NN i , p NF i , p FN i represent the change process from t 1 to t 2 t 1 + 1 as stable forest (FF), stable non-forest (NN), probability of forest growth (NF), and forest loss (FN), respectively.
To ensure the extraction accuracy of forest changes, we removed the first and last two years of the time series, and only analyzed the forest changes in protected areas from 1986 to 2018.

3.4. Forest Change Accuracy Validation

Samples were randomly selected from the extent of the national forest protection area in northeast China and then visually interpreted to provide references for validating the accuracy of the detected forest changes.

3.4.1. Point Selection

Stratified sampling was adopted to increase the performance of point selection across the region. The region was divided into two strata, i.e., changed and non-changed areas, to ensure enough representation for the relatively smaller changed area. The inclusion probability for each stratum was calculated as:
p i | S = n s N s
where p i | S is the probability of each pixel being sampled, and its value is the ratio of the desired number of pixels ( n S ) to the total number of pixels in the stratum ( N S ).
Following the commendation by Feng et al. [30], 500 samples were expected to be selected for each of the non-changed and changed strata (Figure 1). Based on our calculation, the region consisted of ~91 million pixels in the changed stratum and ~493 million pixels in the non-changed stratum. Hence, the p i | S was 0.0001% for the non-changed stratum and 0.0005% for the changed stratum. A total of 1000 samples were collected.

3.4.2. Visual Interpretation

Google Earth Pro was used to visually interpret the samples. Its historical images can be used to capture the process of earth’s surface change in the past decades. To ensure the reliability of sample point interpretation, we also extracted the Normalized difference vegetation index (NDVI) time series of each point from the long-time series of Landsat images using Google Earth Engine (https://earthengine.google.com/, (accessed on 1 November 2021)) and used it to interpret auxiliary data. Figure 2 shows a point (121.637831°E, 51.286481°N) located in Genhe City, Inner Mongolia Autonomous Region. A historical image on Google Earth Pro shows that a forest fire in 2003 caused extensive forest loss in this area, and the NDVI time series diagram also shows that NDVI experienced a sharp decline in the same year (Figure 3). Therefore, we interpreted that at this point, forest loss occurred in 2003, and forest recovery started afterwards and fully recovered in 2016.

3.4.3. Validation Metrics

The forest change accuracy assessment was based on a confusion matrix using overall accuracy (OA), producer’s accuracy (PA), and user’s accuracy (UA) as validation metrics [37]. Large differences in the number of pixels of disturbed and undisturbed strata can result in different sampling probabilities for each pixel. To eliminate this difference, we normalized the inclusion probability of each observation point to make it proportional to the area of each stratum [30,38]. The calculation formulas of a disturbed stratum weight w i and a non-disturbed stratum weight w j are as follows:
w i = p j | S j p i | S i +   p j | S j
w j = p i | S i p i | S i +   p j | S j
where p i | S i and p j | S j represent the probability of sampling each point in the disturbed and the undisturbed strata, respectively.

4. Results

4.1. Spatial Distribution of Tree-Canopy Cover

The TCC estimated in this research was in strong correlation with GEDI and ICESat-2 observations, and the R-square were 0.61 and 0.60, respectively (Figure 4), effectively reflecting the forest cover status in this region. From 1986 to 2018, the forest cover in the NFCP in northeast China was relatively high (Figure 5), with an average TCC of 44.67%, and its spatial distribution was mainly bimodal (Figure 6a). The TCC values of 0–6% and 59–76% accounted for 17.21% and 32.85% of the areas, respectively. According to the definition of forest in the United Nations Framework Convention on Climate Change [39], the forest area of the NFCP in northeast China was 31.13 million hectares, continuously distributed in Daxing’anling and Xiaoxing’anling Mountains. The distribution of TCC in Changbai Mountains was more heterogeneous than it is in the Daxing’anling Mountains due to the stronger influence of human activities such as urban expansion. The highest values were mainly distributed at longitudes between 120.7–125.8°E and 127.5–130.3°E (Figure 6b), with a mean value of TCC greater than 40% in these areas. The latitudinal variation of TCC was mainly characterized by a “V” shape (Figure 6c). The forest density was higher in the latitudinal range of 41.3–44.7°N, and the TCC value exceeded 40%. Then it decreased with the increase in latitude, reaching the lowest value at 46.1°N, where TCC was only 12.49%. The TCC variability increased with the increase in latitude above 46.1°N. In terms of altitude, high TCC areas were mainly distributed in the 400–1900 m elevation zone (Figure 6d). The mean TCC in these areas was greater than 45%, and the maximum was 67.59% in the 1650–1700 m elevation range.

4.2. Spatio-Temporal Changes of Tree-Canopy Cover

The analysis of the temporal and spatial change patterns of TCC can comprehensively reflect the state of forest change. We found that from 1986 to 2018, the average TCC in protected areas was 37.19–48.35%, mainly showing a fluctuating upward trend (Figure 7); the change slope was 0.12%/year, and forest cover continued to improve, especially after 2005. Spatially, forest fires in Daxing’anling in 1987 and 2003 had a great impact on forest cover (Figure 8). In addition, in the surrounding area of the Northeast Plain, TCC mainly showed a decreasing trend (Figure 8). Statistically, the area of TCC that increased was much larger than the one that decreased, accounting for 59.68% and 40.34%, respectively.

4.3. Forest Loss and Gain

The extraction accuracy of forest changes in this study was relatively high compared with manual interpretation (Table 1), effectively reflecting the state of regional forest change. From 1986 to 2018, the forest in the NFCP in northeast China underwent drastic and complex changes. For a clearer understanding of the forest change process, we divided forest change into four categories based on whether the forest had experienced loss and gain: ‘Forest loss without gain’, ‘Forest gain without loss (afforestation)’, ‘Forest gain after loss (reforestation)’ and ‘Forest loss after gain’ (Figure 9). During the study period, ‘Forest gain after loss (reforestation)’ had the largest distribution area, more than 5.12 million hectares of forests recovered after loss (Figure 9b), mainly in the 300–600 m elevation zone (Figure 10). Forest losses caused by fires in 1987 and 2003 largely recovered in Daxing’anling Mountains (Figure 9a). These were followed by ‘Forest gain without loss (afforestation)’, accounting for 37.68% of the total area of forest change, more than 4.79 million hectares of new forest were added (Figure 9b). Compared with reforestation, afforestation was distributed over a wider range of elevations (300–900 m elevation zone) (Figure 10). Especially in areas with higher altitudes and less impacted by human activities, there were also large areas of new forests, which may be closely related to the significant warming in northeast China since the 1990s. There were 1.48 million hectares of forests that did not recover after the loss (Forest loss without gain) (Figure 9b), which were mainly distributed at the edges of the Northeast Plain and some areas of Changbai Mountain (Figure 9a), with an elevation range of 200–500 m (Figure 10). Most of these losses occurred before 1998 and were closely related to human activities such as farmland reclamation. Only 1.32 million hectares of forest were disturbed after recovery (Forest loss after gain) (Figure 9b), mainly distributed in the 300–500 m elevation zone (Figure 9 and Figure 10).
The characteristics of forest loss and gain were also analyzed separately. Over 33 years, the forest loss area (including ‘Forest loss without gain’, ‘Forest gain after loss’, and ‘Forest loss after gain’) of the NFCP in northeast China exceeded 7.92 million hectares (Table 2), mainly distributed in the 200–600 m elevation zone (Figure 11), covering a total area of 5.85 million hectares, accounting for 66.62% of the total loss area. The average annual loss area exceeds 240 thousand hectares, accounting for 0.52% of the protected areas. The area of forest loss peaked in 2003 (Figure 12a), reaching 548.7 thousand hectares. In 1987, 1997, 1998, and 2004, the area of forest loss exceeded 350 thousand hectares, of which 1987 and 2003 were mainly caused by large-scale forest fires. From 6 May to 2 June 1987, extreme forest fires occurred in the northern forest region of the Daxing’anling Mountains, with an area of 1.14 million hectares of burned and sparse forests. Our research shows that the fire caused a substantial loss of 207.4 thousand hectares of forest area. The degree of forest gain (including ‘Forest gain without loss’, ‘Forest gain after loss’, and ‘Forest loss after gain’) of the NFCP in northeast China was much greater than the forest loss, with more than 11.24 million hectares, mainly distributed in the 300–1000 m elevation zone (Figure 11). The average annual forest gain area was 340.52 thousand hectares, accounting for 0.73% of the protected areas. By comparison, except for areas above 1600 m, the forest gain area in the other elevation zones were greater than the forest loss. The gain area in 2008 was the largest, reaching 509.85 thousand hectares (Figure 12b). In 2009, 2011, and 2012, the gain area exceeded 480 thousand hectares (Figure 12b). Among the 48 counties/cities in the NFCP in northeast China, Olunchun Autonomous Banner had the largest forest loss and gain areas (Figure 13), 1.33 million hectares and 1.74 million hectares, respectively.
The implementation of the NFCP in 1998 effectively promoted forest restoration in northeast China. Before 1998, deforestation was mainly motivated by economic interests [40]. The forest loss area increased continuously with a slope of 7.98%/year (Figure 12a), and the average annual forest loss area was 241.17 thousand hectares (Table 2). After the NFCP was officially implemented in 1998, deforestation was effectively curbed, and the area of forest loss was significantly reduced (slope: −14.24%/year) (Figure 12a), the annual average area of forest loss decreased to 239.54 thousand hectares (Table 2). At the same time, the area of forest gain increased significantly (slope: 4.13%/year) (Figure 12b), and the average annual restoration area was as high as 361.47 thousand hectares (Table 2), which was much larger than before the implementation of the NFCP.

5. Discussion

5.1. The Impact of Forest Protection Policies on Forest Change

As early as 1978, forest management was identified as a public policy focus, and China began to implement comprehensive forest protection, combining logging with afforestation to realize sustainable forestry practices [40]. In 1984, China officially started to implement forest management strategies based on a five-category classification system of forest resources: Shelterbelt/windbreak Forest; Special-use Forest; Timber Forest; Economic Forest, and Fuel Forest [12], trying to adopt a more comprehensive management approach to forest use. A “cutting quota” was also introduced in 1987, which formally stipulated that the annual timber production should not exceed the approved annual cutting quota [12]. However, none of these schemes could effectively maintain the country’s forest resources, and forest logging was still the main focus of China’s forest enterprises. At the same time, illegal logging and resulting deforestation were rampant. Our research shows that between 1986 and 1998, fluctuations in forest loss areas increased. Between 1998 and 2008, after the implementation of the NFCP, forest areas increased rapidly, which shows that forests in the protected areas can be restored quickly, effectively curbing deforestation. This is consistent with the results of Yu et al. [40]. The forest gains gradually decreased after 2008, which may have been caused by the rapid rate of forest gain in the early stage reaching a saturation point. After 2000, large-scale forest loss events still occurred in individual years, but most of these events were caused by natural disasters, not related to large-scale deforestation. Curtis et al. [41] classified the drivers of forest loss (i.e., commodity-driven deforestation, shifting agriculture, forestry, wildfire, and urbanization), indicating that the forest loss in the protected areas in northeast China from 2001 to 2015 was mainly caused by forestry activities, accounting for 24.2%, followed by forest fires, which were mainly distributed in the north of the Daxing’an Mountains, accounting for 17.2%, and only 0.2% of the area was caused by farmland reclamation, which was scattered in Changbai Mountain. In the study area, there was no loss caused by commodity-driven deforestation and urbanization. Overall, national forest protection policies in different periods were the main factors affecting forest change. Natural disasters [42,43] such as fire, insects, wind, snow and frost also caused some loss of forest resources.

5.2. Accurate Detection of Forest Change

Time series of Landsat TCC provide a consistent high-resolution dataset that can be effectively used to capture complex forest dynamics. Because of its advantages of spatial resolution and long time range, our dataset is not only sensitive to drastic events such as clear cuts and forest fires, but also responds to processes with weak change characteristics such as forest recovery. Compared with the global forest change products [21], our results not only have a longer time range but also quantify annual forest recovery features. In addition,, we found that the global forest change dataset seriously underestimates the forest loss and gain of the NFCP in northeast China. From 2001 to 2018, this study identified that the forest loss area of the protected areas was 3.96 million hectares, while the global forest change dataset recorded only 763.18 thousand hectares. This was likely caused by a large number of small-scale disturbances that have not been effectively identified in the global forest change dataset, which is consistent with the results of Milodowski et al. [44] in the Amazon. The point at (123.969°E, 50.839°N) is located in Olunchun Autonomous Banner, Inner Mongolia Autonomous Region. Google Earth Pro historical images show significant deforestation here after 2000 (Figure 14), and our research has effectively extracted the scope of forest disturbance, while the global forest change product does not detect forest loss there (Figure 15). This may be due to the fact that global forest change products use global data for modeling, which are inadequate to represent the characteristics of regional forest changes, resulting in large errors [44,45]. In contrast, our research retains a large number of details of forest loss and gain, and most of these small-scale changes can be confirmed through manual interpretation, which demonstrates that our research results can more accurately reflect the state of regional forest change.

5.3. Validation

The validation of Landsat-derived TCC in northeast China was found to be strongly consistent with GEDI and ICESat-2 observations, suggesting that, in general, the data is highly accurate. There are some errors in the GEDI and ICESat-2 observations in the higher TCC range caused by the heterogeneity of the surface and the influence of clouds, which are mainly manifested in the identification of buildings, grasslands, shrubs, and other objects as trees. In particular, there is a large probability of photon irradiation on the tree canopy of ICESat-2 observations, resulting in an overestimation of the observation of canopy cover [35]. Validation revealed that the forest loss and forest gain derived from the long-term TCC detected forest change with high accuracy.

6. Conclusions

Based on the long-time series TCC, we identified the forest loss and gain events of the NFCP in northeast China from 1986 to 2018. The algorithm used in this paper not only effectively detected events with strong interference characteristics, such as clear-cuts and fires, but was also equally sensitive to processes with weak change characteristics such as forest recovery that can be used for monitoring and evaluating the effectiveness of other forest conservation efforts. This research can supplement gaps in global forest change products and can more accurately reflect the state of regional forest changes.
The NFCP in northeast China has experienced a robust process of forest change from 1986 to 2018. The forest protection policies implemented in different periods in China were the main influencing factors of forest change. After the implementation of the NFCP in 1998, the area of forest loss decreased significantly. The continuous and effective implementation of the protection policy is crucial to the improvement of the regional ecological environment. Overall, from 1986 to 2018, the degree of forest gain in protected areas was much greater than forest loss, and forest ecosystems tended to improve under the joint action of human activities, climate change, and natural disaster events.
With the booming of Earth observation satellites in the past decade, more data becomes available that can be fused and integrated with Landsat to overcome the gaps in observations caused by clouds. Further improvements of the spatial and spectral resolutions of remote sensing data can help transform the research paradigm, adding the use of object-oriented methods to explore the characteristics of forest change from a single canopy scale [46], or make greater use of canopy spectral characteristics for analysis.

Author Contributions

J.W. designed and performed the experiments; M.F. and X.L. supervised and designed the research and contributed to the article’s organization; Z.H. and C.W. provided assistance with the data processing. Y.P. and T.Y. provided protected area boundary data. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Natural Science Foundation of China (42171140), the Fundamental Research Funds of CAF(CAFYBB2020ZD002).

Data Availability Statement

The forest change dataset is available at https://www.doi.org/10.11888/Terre.tpdc.272579 (accessed on 12 June 2022).

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. FAO. Global Forest Resources Assessment 2020; FAO: Rome, Italy, 2020; ISBN 978-92-5-132974-0. [Google Scholar]
  2. Griffiths, P.; Kuemmerle, T.; Baumann, M.; Radeloff, V.C.; Abrudan, I.V.; Lieskovsky, J.; Munteanu, C.; Ostapowicz, K.; Hostert, P. Forest Disturbances, Forest Recovery, and Changes in Forest Types across the Carpathian Ecoregion from 1985 to 2010 Based on Landsat Image Composites. Remote Sens. Environ. 2014, 151, 72–88. [Google Scholar] [CrossRef]
  3. Kautz, M.; Meddens, A.J.H.; Hall, R.J.; Arneth, A. Biotic Disturbances in Northern Hemisphere Forests—A Synthesis of Recent Data, Uncertainties and Implications for Forest Monitoring and Modelling: Biotic Disturbances in Northern Hemisphere Forests. Glob. Ecol. Biogeogr. 2017, 26, 533–552. [Google Scholar] [CrossRef] [Green Version]
  4. Westerling, A.L. Increasing Western US Forest Wildfire Activity: Sensitivity to Changes in the Timing of Spring. Philos. Trans. R. Soc. B Biol. Sci. 2016, 371, 20150178. [Google Scholar] [CrossRef] [PubMed]
  5. Allen, C.D.; Breshears, D.D.; McDowell, N.G. On Underestimation of Global Vulnerability to Tree Mortality and Forest Die-off from Hotter Drought in the Anthropocene. Ecosphere 2015, 6, art129. [Google Scholar] [CrossRef]
  6. Goldewijk, K.K. Estimating Global Land Use Change over the Past 300 Years: The HYDE Database. Glob. Biogeochem. Cycles 2001, 15, 417–433. [Google Scholar] [CrossRef]
  7. Turner, M.G.; Baker, W.L.; Peterson, C.J.; Peet, R.K. Factors Influencing Succession: Lessons from Large, Infrequent Natural Disturbances. Ecosystems 1998, 1, 511–523. [Google Scholar] [CrossRef]
  8. Seidl, R.; Thom, D.; Kautz, M.; Martin-Benito, D.; Peltoniemi, M.; Vacchiano, G.; Wild, J.; Ascoli, D.; Petr, M.; Honkaniemi, J.; et al. Forest Disturbances under Climate Change. Nat. Clim. Chang. 2017, 7, 395–402. [Google Scholar] [CrossRef] [Green Version]
  9. Nabuurs, G.-J.; Lindner, M.; Verkerk, P.J.; Gunia, K.; Deda, P.; Michalak, R.; Grassi, G. First Signs of Carbon Sink Saturation in European Forest Biomass. Nat. Clim. Chang. 2013, 3, 792–796. [Google Scholar] [CrossRef]
  10. Seidl, R.; Schelhaas, M.-J.; Rammer, W.; Verkerk, P.J. Increasing Forest Disturbances in Europe and Their Impact on Carbon Storage. Nat. Clim. Chang. 2014, 4, 806–810. [Google Scholar] [CrossRef] [Green Version]
  11. Masek, J.G.; Huang, C.; Wolfe, R.; Cohen, W.; Hall, F.; Kutler, J.; Nelson, P. North American Forest Disturbance Mapped from a Decadal Landsat Record. Remote Sens. Environ. 2008, 112, 2914–2926. [Google Scholar] [CrossRef]
  12. Dai, L.; Zhao, F.; Shao, G.; Zhou, L.; Tang, L. China’s Classification-Based Forest Management: Procedures, Problems, and Prospects. Environ. Manag. 2009, 43, 1162–1173. [Google Scholar] [CrossRef] [PubMed]
  13. Zhang, P.; Shao, G.; Zhao, G.; Le Master, D.C.; Parker, G.R.; Dunning, J.B., Jr.; Li, Q. China’s Forest Policy for the 21st Century. Science 2000, 288, 2135–2136. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  14. Yang, H. China’s Natural Forest Protection Program: Progress and Impacts. For. Chron. 2017, 93, 113–117. [Google Scholar] [CrossRef] [Green Version]
  15. Feng, M.; Li, X. Land Cover Mapping toward Finer Scales. Sci. Bull. 2020, 65, 1604–1606. [Google Scholar] [CrossRef]
  16. Wulder, M.A.; White, J.C.; Loveland, T.R.; Woodcock, C.E.; Belward, A.S.; Cohen, W.B.; Fosnight, E.A.; Shaw, J.; Masek, J.G.; Roy, D.P. The Global Landsat Archive: Status, Consolidation, and Direction. Remote Sens. Environ. 2016, 185, 271–283. [Google Scholar] [CrossRef] [Green Version]
  17. Wulder, M.A.; White, J.C.; Goward, S.N.; Masek, J.G.; Irons, J.R.; Herold, M.; Cohen, W.B.; Loveland, T.R.; Woodcock, C.E. Landsat Continuity: Issues and Opportunities for Land Cover Monitoring. Remote Sens. Environ. 2008, 112, 955–969. [Google Scholar] [CrossRef]
  18. Hansen, M.C.; Loveland, T.R. A Review of Large Area Monitoring of Land Cover Change Using Landsat Data. Remote Sens. Environ. 2012, 122, 66–74. [Google Scholar] [CrossRef]
  19. Vogelmann, J.E.; Xian, G.; Homer, C.; Tolk, B. Monitoring Gradual Ecosystem Change Using Landsat Time Series Analyses: Case Studies in Selected Forest and Rangeland Ecosystems. Remote Sens. Environ. 2012, 122, 92–105. [Google Scholar] [CrossRef] [Green Version]
  20. Hansen, M.C.; Stehman, S.V.; Potapov, P.V.; Fung, I.Y. Quantification of Global Gross Forest Cover Loss. Proc. Natl. Acad. Sci. USA 2010, 107, 8650–8655. [Google Scholar] [CrossRef] [Green Version]
  21. Hansen, M.C.; Potapov, P.V.; Moore, R.; Hancher, M.; Turubanova, S.A.; Tyukavina, A.; Thau, D.; Stehman, S.V.; Goetz, S.J.; Loveland, T.R.; et al. High-Resolution Global Maps of 21st-Century Forest Cover Change. Science 2013, 342, 850–853. [Google Scholar] [CrossRef] [Green Version]
  22. Song, X.-P.; Hansen, M.C.; Stehman, S.V.; Potapov, P.V.; Tyukavina, A.; Vermote, E.F.; Townshend, J.R. Global Land Change from 1982 to 2016. Nature 2018, 560, 639–643. [Google Scholar] [CrossRef] [PubMed]
  23. Ceccherini, G.; Duveiller, G.; Grassi, G.; Lemoine, G.; Avitabile, V.; Pilli, R.; Cescatti, A. Abrupt Increase in Harvested Forest Area over Europe after 2015. Nature 2020, 583, 72–77. [Google Scholar] [CrossRef]
  24. DeVries, B.; Verbesselt, J.; Kooistra, L.; Herold, M. Robust Monitoring of Small-Scale Forest Disturbances in a Tropical Montane Forest Using Landsat Time Series. Remote Sens. Environ. 2015, 161, 107–121. [Google Scholar] [CrossRef]
  25. Hamunyela, E.; Brandt, P.; Shirima, D.; Do, H.T.T.; Herold, M.; Roman-Cuesta, R.M. Space-Time Detection of Deforestation, Forest Degradation and Regeneration in Montane Forests of Eastern Tanzania. Int. J. Appl. Earth Obs. Geoinf. 2020, 88, 102063. [Google Scholar] [CrossRef]
  26. Hansen, M.C.; Townshend, J.R.G.; DeFries, R.S.; Carroll, M. Estimation of Tree Cover Using MODIS Data at Global, Continental and Regional/Local Scales. Int. J. Remote Sens. 2005, 26, 4359–4380. [Google Scholar] [CrossRef]
  27. DiMiceli, C.; Townshend, J.; Carroll, M.; Sohlberg, R. Evolution of the Representation of Global Vegetation by Vegetation Continuous Fields. Remote Sens. Environ. 2021, 254, 112271. [Google Scholar] [CrossRef]
  28. Townshend, J.R.G.; Justice, C.O. Selecting the Spatial Resolution of Satellite Sensors Required for Global Monitoring of Land Transformations. Int. J. Remote Sens. 1988, 9, 187–236. [Google Scholar] [CrossRef]
  29. Sexton, J.O.; Song, X.-P.; Feng, M.; Noojipady, P.; Anand, A.; Huang, C.; Kim, D.-H.; Collins, K.M.; Channan, S.; DiMiceli, C.; et al. Global, 30-m Resolution Continuous Fields of Tree Cover: Landsat-Based Rescaling of MODIS Vegetation Continuous Fields with Lidar-Based Estimates of Error. Int. J. Digit. Earth 2013, 6, 427–448. [Google Scholar] [CrossRef] [Green Version]
  30. Feng, M.; Sexton, J.O.; Huang, C.; Anand, A.; Channan, S.; Song, X.-P.; Song, D.-X.; Kim, D.-H.; Noojipady, P.; Townshend, J.R. Earth Science Data Records of Global Forest Cover and Change: Assessment of Accuracy in 1990, 2000, and 2005 Epochs. Remote Sens. Environ. 2016, 184, 73–85. [Google Scholar] [CrossRef] [Green Version]
  31. Liang, L.; Li, L.; Liu, Q. Precipitation Variability in Northeast China from 1961 to 2008. J. Hydrol. 2011, 404, 67–76. [Google Scholar] [CrossRef]
  32. Huang, L.; Wang, B.; Niu, X.; Gao, P.; Song, Q. Changes in Ecosystem Services and an Analysis of Driving Factors for China’s Natural Forest Conservation Program. Ecol. Evol. 2019, 9, 3700–3716. [Google Scholar] [CrossRef]
  33. Xiao, X.; Boles, S.; Liu, J.; Zhuang, D.; Liu, M. Characterization of Forest Types in Northeastern China, Using Multi-Temporal SPOT-4 VEGETATION Sensor Data. Remote Sens. Environ. 2002, 82, 335–348. [Google Scholar] [CrossRef]
  34. Tang, H.; Armston, J. GEDI L2B Footprint Canopy Cover and Vertical Profile Metrics. 39. Available online: https://lpdaac.usgs.gov/documents/588/GEDI_FCCVPM_ATBD_v1.0.pdf (accessed on 1 October 2021).
  35. Neuenschwander, A.; Pitts, K. The ATL08 Land and Vegetation Product for the ICESat-2 Mission. Remote Sens. Environ. 2019, 221, 247–259. [Google Scholar] [CrossRef]
  36. Sexton, J.O.; Noojipady, P.; Anand, A.; Song, X.-P.; McMahon, S.; Huang, C.; Feng, M.; Channan, S.; Townshend, J.R. A Model for the Propagation of Uncertainty from Continuous Estimates of Tree Cover to Categorical Forest Cover and Change. Remote Sens. Environ. 2015, 156, 418–425. [Google Scholar] [CrossRef] [Green Version]
  37. Brovelli, M.; Molinari, M.; Hussein, E.; Chen, J.; Li, R. The First Comprehensive Accuracy Assessment of GlobeLand30 at a National Level: Methodology and Results. Remote Sens. 2015, 7, 4191–4212. [Google Scholar] [CrossRef] [Green Version]
  38. Sexton, J.O.; Urban, D.L.; Donohue, M.J.; Song, C. Long-Term Land Cover Dynamics by Multi-Temporal Classification across the Landsat-5 Record. Remote Sens. Environ. 2013, 128, 246–258. [Google Scholar] [CrossRef]
  39. Sexton, J.O.; Noojipady, P.; Song, X.-P.; Feng, M.; Song, D.-X.; Kim, D.-H.; Anand, A.; Huang, C.; Channan, S.; Pimm, S.L.; et al. Conservation Policy and the Measurement of Forests. Nat. Clim. Chang. 2016, 6, 192–196. [Google Scholar] [CrossRef]
  40. Yu, D.; Zhou, L.; Zhou, W.; Ding, H.; Wang, Q.; Wang, Y.; Wu, X.; Dai, L. Forest Management in Northeast China: History, Problems, and Challenges. Environ. Manag. 2011, 48, 1122–1135. [Google Scholar] [CrossRef]
  41. Curtis, P.G.; Slay, C.M.; Harris, N.L.; Tyukavina, A.; Hansen, M.C. Classifying Drivers of Global Forest Loss. Science 2018, 361, 1108–1111. [Google Scholar] [CrossRef]
  42. Ji, L.; Wang, Z.; Wang, X.; An, L. Forest Insect Pest Management and Forest Management in China: An Overview. Environ. Manag. 2011, 48, 1107–1121. [Google Scholar] [CrossRef]
  43. Yi, K.; Bao, Y. Estimates of Wildfire Emissions in Boreal Forests of China. Forests 2016, 7, 158. [Google Scholar] [CrossRef] [Green Version]
  44. Milodowski, D.T.; Mitchard, E.T.A.; Williams, M. Forest Loss Maps from Regional Satellite Monitoring Systematically Underestimate Deforestation in Two Rapidly Changing Parts of the Amazon. Environ. Res. Lett. 2017, 12, 094003. [Google Scholar] [CrossRef] [Green Version]
  45. Tropek, R.; Sedláček, O.; Beck, J.; Keil, P.; Musilová, Z.; Šímová, I.; Storch, D. Comment on “High-Resolution Global Maps of 21st-Century Forest Cover Change”. Science 2014, 344, 981. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  46. Brandt, M.; Tucker, C.J.; Kariryaa, A.; Rasmussen, K.; Abel, C.; Small, J.; Chave, J.; Rasmussen, L.V.; Hiernaux, P.; Diouf, A.A.; et al. An Unexpectedly Large Count of Trees in the West African Sahara and Sahel. Nature 2020, 587, 78–82. [Google Scholar] [CrossRef]
Figure 1. The extent of the Natural Forest Conversion Program in northeast China and the validation sample sites.
Figure 1. The extent of the Natural Forest Conversion Program in northeast China and the validation sample sites.
Remotesensing 14 02988 g001
Figure 2. Google Earth Pro historical images at the marked point (121.637831°E, 51.286481°N) reflect forest loss and gain processes.
Figure 2. Google Earth Pro historical images at the marked point (121.637831°E, 51.286481°N) reflect forest loss and gain processes.
Remotesensing 14 02988 g002
Figure 3. NDVI time series at point (121.637831°E, 51.286481°N) from 1986 to 2019.
Figure 3. NDVI time series at point (121.637831°E, 51.286481°N) from 1986 to 2019.
Remotesensing 14 02988 g003
Figure 4. Comparison of TCC estimation with: (a) GEDI; and (b) ICESat-2 observations in 2019. The color represents the density of the points, ranging from low (blue), medium (yellow), to high (red).
Figure 4. Comparison of TCC estimation with: (a) GEDI; and (b) ICESat-2 observations in 2019. The color represents the density of the points, ranging from low (blue), medium (yellow), to high (red).
Remotesensing 14 02988 g004
Figure 5. Spatial distribution of average TCC in the Natural Forest Conversion Program from 1986 to 2018, northeast China.
Figure 5. Spatial distribution of average TCC in the Natural Forest Conversion Program from 1986 to 2018, northeast China.
Remotesensing 14 02988 g005
Figure 6. (a) Histogram of TCC. TCC changes with (b) longitude; (c) latitude; and (d) altitude.
Figure 6. (a) Histogram of TCC. TCC changes with (b) longitude; (c) latitude; and (d) altitude.
Remotesensing 14 02988 g006
Figure 7. Interannual change of TCC during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 7. Interannual change of TCC during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g007
Figure 8. Spatial changes of TCC during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 8. Spatial changes of TCC during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g008
Figure 9. (a) Spatial distribution of forest change and (b) Statistics on change area during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 9. (a) Spatial distribution of forest change and (b) Statistics on change area during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g009
Figure 10. Distribution of forest change types by elevation zone during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 10. Distribution of forest change types by elevation zone during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g010
Figure 11. Distribution of forest loss and gain by elevation zone during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 11. Distribution of forest loss and gain by elevation zone during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g011
Figure 12. Annual forest (a) loss and (b) gain during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 12. Annual forest (a) loss and (b) gain during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g012
Figure 13. Spatial statistics of forest (a) loss and (b) gain in each county/city during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Figure 13. Spatial statistics of forest (a) loss and (b) gain in each county/city during 1986–2018 in the Natural Forest Conversion Program, northeast China.
Remotesensing 14 02988 g013
Figure 14. Google Earth Pro historical images at point (123.969°E, 50.839°N).
Figure 14. Google Earth Pro historical images at point (123.969°E, 50.839°N).
Remotesensing 14 02988 g014
Figure 15. Comparison of forest loss detection results (a) this research; (b) global forest change dataset.
Figure 15. Comparison of forest loss detection results (a) this research; (b) global forest change dataset.
Remotesensing 14 02988 g015
Table 1. Validation of forest change detection accuracy.
Table 1. Validation of forest change detection accuracy.
Accuracy IndexForest Loss (%)Forest Gain (%)
Overall accuracy94.9294.13
Producer’s accuracy85.2187.74
User’s accuracy84.2688.31
Table 2. Statistics of forest change area in different periods.
Table 2. Statistics of forest change area in different periods.
Period (Year)Forest Loss Area (×1000 ha)Forest Gain Area (×1000 ha)
TotalAnnual AverageTotalAnnual Average
1986–19972894.02241.173646.38303.87
1998–20185030.43239.547590.93361.47
1986–20187924.45240.1311,237.31340.52
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Wang, J.; He, Z.; Wang, C.; Feng, M.; Pang, Y.; Yu, T.; Li, X. Investigation of Long-Term Forest Dynamics in Protected Areas of Northeast China Using Landsat Data. Remote Sens. 2022, 14, 2988. https://doi.org/10.3390/rs14132988

AMA Style

Wang J, He Z, Wang C, Feng M, Pang Y, Yu T, Li X. Investigation of Long-Term Forest Dynamics in Protected Areas of Northeast China Using Landsat Data. Remote Sensing. 2022; 14(13):2988. https://doi.org/10.3390/rs14132988

Chicago/Turabian Style

Wang, Jianbang, Zhuoyu He, Chunling Wang, Min Feng, Yong Pang, Tao Yu, and Xin Li. 2022. "Investigation of Long-Term Forest Dynamics in Protected Areas of Northeast China Using Landsat Data" Remote Sensing 14, no. 13: 2988. https://doi.org/10.3390/rs14132988

APA Style

Wang, J., He, Z., Wang, C., Feng, M., Pang, Y., Yu, T., & Li, X. (2022). Investigation of Long-Term Forest Dynamics in Protected Areas of Northeast China Using Landsat Data. Remote Sensing, 14(13), 2988. https://doi.org/10.3390/rs14132988

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop