Sir

We have compared the official estimates of agricultural land and rates of agricultural land conversion with those derived from Landsat thematic mapper satellite images for 10 counties in the Pearl River Delta, which is one of the fastest-developing regions in China. Ground- based field assessments verify the high accuracy of our techniques in estimating the area of agricultural land and its change through time1,2.

Our results indicate that there is significantly more agricultural land than reported in official statistics3. Although this underreporting is well documented4, particularly using coarse resolution (1-km) satellite data sets5, our study is the first to use high-resolution satellite imagery to quantify this bias.

Satellite-derived estimates of total agricultural area in the ten counties for 1990 is 6,724 km2 — this is 115% greater than the 3,119 km2 reported in government yearbooks ( Fig. 1a). Similarly, satellite- derived estimates for the total area of agricultural land converted to other uses between 1990 and 1996 is about 11% greater than the 789 km2 reported in statistical yearbooks (Fig. 1b).

Figure 1: a, Estimates of agricultural land in 10 counties of the Pearl River Delta, 1990, derived from satellite imagery (blue) and official yearbooks (red).
figure 1

b, Estimates of agricultural land conversion from 1990 to 1996 in 10 counties of the Pearl River Delta, derived from satellite imagery (blue) and official yearbooks (red).

Full size image

One interesting effect is that while the satellite-based method estimates more overall conversion of agricultural land, the fraction of the agricultural land converted is smaller than official estimates.

With the world's largest population and one of the fastest-growing economies, China's ability to feed itself has important domestic and international implications. Agricultural self-sufficiency is a concern because economic development is rapidly converting China's small per-capita stock of agricultural land to other uses6,7. Analyses of the severity of this problem are based largely on official statistics for the stock of agricultural land and its rate of conversion, but these statistics may be biased by institutional factors.

Starting with the Great Leap Forward in 1958, Chinese farmers have had strong incentives to understate the extent of their agricultural land. Grain quotas were based on agricultural acreage, so understating agricultural area reduced the quota a farmer was expected to provide.

Although this production quota has been eliminated, farmers still have incentives to understate their agricultural land. A 1985 moratorium in the region limited the amount of agricultural land converted for other uses. This directive, together with a desire to evade taxes on land leased for commercial purposes, causes farmers to understate the area of agricultural land that has been converted to other uses.

Despite this under-reporting, we are not suggesting that the satellite-derived estimates are unbiased. Remote-sensing estimates may overestimate the amount of agricultural land because the resolution of the imagery (30×30 m) is too coarse to differentiate among small irrigation ditches, dirt paths and the other types of land use that co-exist with agriculture. Moreover, remote-sensing and government classification regimes of agricultural land may differ.

Our conclusion that there is more agricultural land than reported does not eliminate concern about the loss of agricultural land associated with rapid rates of economic development. Indeed, our results indicate that more agricultural land is being converted than reported.

Our results do, however, indicate that analysts must exercise caution when they use official statistics of agricultural land area and rates of loss to assess future rates of agricultural production in China.