In September 2020, against the backdrop of the coronavirus pandemic and with the global economy still reeling, China pledged that it would achieve carbon neutrality by 2060. In this two-part series, researchers from the Columbia School of International and Public Affairs explore China's progress towards that goal along two vectors: "brown" and "green." Part 2 ("green") examines how tree cover has changed in two provinces as China conducts major tree-planting efforts.
Throughout 2020, China highlighted its plans to promote environmentally sustainable growth. At the end of 2020, China's President Xi Jinping went before the United Nations and encouraged a green recovery from the pandemic, while in December, he committed China to meet a series of environmental targets--peaking CO2 emissions, increasing the renewable energy share to 25%, and increasing China's forest stock by 6 billion cubic meters from 2005 levels--by 2030. This last target--increasing forest cover--saw China continue tree-planting programs throughout 2020 in spite of the pandemic. With some municipalities still under lockdown, official press releases assured that China and Xi Jinping would forge ahead on environmental efforts, and in Jiangxi and Hebei provinces, workers who were unable to travel home due to lockdowns were even re-tasked with planting trees.
China's afforestation and reforestation efforts date back to the 1970s when Deng Xiaoping--reversing Mao-era development policies that devastated Chinese forests--implemented tree-planting programs that still run today. One of the largest and most widely publicized is the Green Great Wall program. Originally designed to halt soil erosion and desertification in Northern China, the massive project covers a perimeter of 400km and will not end until 2050. Another major program is the Grain to Green Program, which pays farmers to plant trees in most of China's provinces. Officially, China's progress under these programs (and others) has been substantial. Since 2000, official data indicates China planted 88.9 million hectares of trees, representing an area larger than France, and increased the country's forested area by almost 5%. If anything, China's new planting targets are even more ambitious, aiming to increase forested areas further 1% by 2025. These new trees are now part of China's overall climate targets and the latest (Fourteenth) Five-Year Plan.
These developments gave global environmentalists some grounds for optimism. Yet there are also grounds for skepticism. Progress in one aspect of China's environment can be undermined by backsliding or stagnation in other aspects; we examined this dynamic in Article 1, as we outlined how China continued to rapidly approve and construct new power plants in 2020.
Whether or not China will be able to meet these targets is an open question, but past tree-planting efforts--evaluated through the use of remote sensing data--provide some crucial context. Data from two provinces show possible discrepancies between China's past stated performance and conditions on the ground, and highlight some of the challenges China will face in moving forward on its environmental goals when the pandemic subsides.
Analyzing plant cover
Satellite remote sensing data can help clarify whether official statistics are matched by increased greening on the ground. We analyze satellite data to calculate Normalized Difference Vegetation Index (NDVI) values, which indicate the presence and density of live vegetation in a given area. Although the NDVI alone cannot explain why changes in vegetation cover happen--whether due to natural growth or human intervention--changes in the NDVI over time provide strong evidence of whether an area is growing greener or browner over time.
We compared NDVI results to Chinese afforestation statistics, in particular, the total area of afforestation claimed for each province in China's annual Statistical Yearbooks. If forest planting is taking place in a given area, significant greening should be visible in two to three years as China predominantly plants rapid growth trees such as poplars and Chinese firs. Moreover, as newly planted saplings mature, their growth should register as changes in the NDVI.
We focused specifically on tree-planting efforts in two provinces, with different climatic zones: Inner Mongolia, a large and mostly arid area in the North that comprises a significant portion of the Green Great Wall; and Jiangxi, a province in the South, in a more humid zone, where the "Grain to Green Program" is active. The Grain to Green program, initiated in 1999, pays farmers to plant trees and convert farmland to forest, providing subsidies for seeds and other inputs.
We deployed open-source imagery and remote sensing data analyzed through Google Earth Engine. First, to get a general sense of changes over time, we utilize daily NDVI data from NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) module to compare NDVI levels throughout the years 2001 (the first complete year available), 2010, and 2020. As our study focuses on changes in recent greening, we then deploy open-source imagery from the Landsat 8 satellite to develop a measure of NDVI changes. We obtain imagery from each province across two periods: 2013-2016 and 2017-2020, and calculate corresponding NDVI values for each period. The NDVI provides a range from -1 to +1, corresponding to vegetation cover. The more positive the NDVI, the more extensive the vegetation cover: sparse vegetation is approximately 0.4-0.5, while dense forest approaches 1.0.
We take the maximum NDVI values from each period and calculate the change between periods. We then identify areas in each province where NDVI changed significantly. A change of +0.3 in the NDVI represents significant greening, and across a short time horizon, likely requiring human intervention. If tree planting was successfully implemented, we would expect the growth of these newly planted trees to register as substantial changes in the NDVI.
What We Found
Inner Mongolia has gotten greener over time. Drawing on MODIS data (Figure 1), we clearly see that the region's average NDVI has improved from 2001 to 2020, especially during the summer peak and green season.
Figure 1: MODIS NDVI values by day of year (1-365): Inner Mongolia (average)
However, the provincial story is less straightforward. Looking at our measure of NDVI changes across Inner Mongolia (Figure 2), we do see areas of NDVI improvement (green), but the region also contends with significant desertification and erosion (seen as negative changes in NDVI, in red), alongside other land development.
Moreover, areas of significant change--where NDVI changed by more than 0.3 between periods--were limited. These areas are visualized in bright green in Figure 3, and represent a fraction of the total green area change seen above.
We drilled down on a few of these areas and obtained satellite imagery as available for each period. Two examples are pictured below (Figures 4 and 5). We aggregate this analysis in Figure 6, coding apparent changes in 9 sites as due to planting or natural forest growth (green, five of nine sites), agricultural activity (yellow, three of nine sites), or water effects our NDVI measures (blue, one of nine sites).1 The confounding role of agricultural activity on the NDVI suggests positive NDVI changes may not be due to tree planting. This includes what appears to be new irrigation systems, as seen in Figures 4 and 5.
Figure 4: Inner Mongolia, Ordos
Figure 5: Inner Mongolia, Unlanqab
Because of Jiangxi's southern climate, with more natural greenery and a less distinguished winter season, average NDVI across the province shows less seasonality and a less clear upward trend since 2001. Though average NDVI levels at the same time of year are generally higher than they were in 2001 and 2010, there is substantial variation by year.
Figure 7: MODIS NDVI values by day of year (1-365): Jiangxi (average)
Comparing NDVI changes since 2017 (Figure 8), we see more areas of negative change in the NDVI in a specific band in the west of Jiangxi; this is likely due to specific image sampling issues during the 2017-2020 period. Excluding changes in this band, areas of major NDVI change (greater than +0.3 in the index, Figure 9) appear limited.
We examined 27 of these sites in Jiangxi (Figure 10). Like Inner Mongolia, the greening in these sites does not appear solely due to tree-planting or other development. Of the 27 sites, roughly half (13) appear to have greened due to agricultural activity, compared to 11 sites with greening that can be plausibly attributed to planting or other development. One such greening site is shown in Figure 11. The change in the remaining three sites can be attributed to water, where the NDVI is picking up water-adjacent greening and/or reflection from the water's surface.
Figure 11: Jiangxi, Ganzhou
Explaining the Official Numbers and Future Prospects
Our analysis of Inner Mongolia and Jiangxi shows each province experienced a marginal increase in NDVI from 2017-2020 compared to the period from 2013-2016, representing a small increase in plant and forest cover. However, major changes in NDVI were limited to only a few areas, representing a small fraction of each province's territory. Moreover, changes in forest cover appear less significant than would be expected based on official government tree-planting statistics. Even in the study sites where NDVI analysis found increases in vegetation, the changes do not appear caused by increasing tree cover. Instead, some of the changes were driven by other factors, including crop development, algae formation on lakes, and natural greening alongside rivers.
One potential explanation for the relatively limited change in NDVI may be related to tree die-off. Planted trees may have not reached maturity and have low survival rates. This survival problem may be particularly acute in arid Inner Mongolia, where poplar trees were the planted tree of choice. Poplars grow quickly, making them well suited for local officials who want to demonstrate significant afforestation progress within tight timeframes. But poplars also require significant amounts of water, limiting their self-sustainability and making them particularly ill-suited for Inner Mongolia's arid climate.
However, an alternate explanation is that government tree-planting statistics are inflated. External studies confirm that such inflation has happened before. Data from Global Forest Watch (GFW) found that previous efforts to increase forest cover in Inner Mongolia and Jiangxi were limited during the early 2000s: from the entire period between 2000 and 2012, GFW estimated forest cover in Inner Mongolia increased by around 54,700 hectares (around 4,560 hectares/year), while Jiangxi cover increased by 137,000 hectares (around 11,400 hectares/year). These estimates stand in stark contrast to provincial government statistics from the period. Officials reported that 718,562 hectares of trees were planted in Inner Mongolia and 267,032 hectares in Jiangxi in 2009 alone, a level of tree planting that is simply not supported by outside analysis.
Recent province-level statistics make similar claims. For example, 2017 statistics report that Inner Mongolia and Jiangxi province manually planted 346,309 and 89,405 hectares of trees, respectively. This would represent a combined area larger than the state of Rhode Island (around 314,000 hectares). Although area calculations are outside the scope of our immediate analysis, the limited areas of major greening seen in our NDVI change index, as well as our spot analysis of several sites do cast some doubt that China's tree planting--at least in these provinces--has been fully carried out at levels suggested by government data.
1. High reflectance from bodies of water can influence NDVI values of adjacent regions.