Are China's water bodies (lakes) underestimated?

In their study of surface water bodies (WBs) in China, Feng et al. (1) used the global surface water dataset (GSWD) by Pekel et al. (2) to report that the abundance of WBs (>1 km2) and their total area were underestimated greatly by previous studies (3, 4). However, the incorrect definitions of WBs, data coverage, and the temporal windows in Feng et al. (1) could have resulted in large errors, and thus discrepancies for the number of WBs detected and their area changes.

As Earth’s surface water is driven by changing climate and anthropogenic factors, it is important that an epoch date be defined at which the WB calculations are referenced. Feng et al. (1) used the GSWD (2), excluding rivers and streams, to calculate the WB occurrences in China during 1984 to 2015 and found there is a total of 6,821 WBs (>1 km2). They argued that previous studies (e.g., refs. 3 and 4) have underestimated the number and sizes of WBs in China. Indeed, a smaller number (5,535) of lakes and reservoirs has been derived from Landsat images during 2005 to 2008 (3), and 2,693 lakes were mapped during 2005 to 2006 (4). Zhang et al. (5) demonstrated that China had 2,554 lakes (>1 km2) in 2015 and that lake number and area in China including Tibetan Plateau (TP) have been increasing between the 1970s and 2015 (Fig. 1). When compiling a lake inventory, reservoirs and dams should be excluded, and a relatively stable season for WBs should be chosen to avoid seasonal, interannual, or longer variability (6). This is the reason why ±2-y or longer data intervals should be employed (5). These considerations are not addressed by Feng et al. (1), leading to the feasibility that the maximum WBs number and area were estimated. For the TP, Feng et al. (1) claim that WB number in their study is close to the value (1,291) determined by Mao et al. (7) from 2014, as lake number (>1 km2) has been increasing in recent decades, and there are few reservoirs on the TP (Fig. 1). It is unsurprising that Feng et al. (1) mapped more WBs than previous studies (3, 4), as a longer period (1984 to 2015) was covered, the historical maximum water area was used, and more WB types were included. The differences in total area found between Feng et al. (1) and other studies (3, 8) are due to similar reasons.

Fig. 1.

Lake number and area in China and their changes. (A) Lake distribution including 2,554 lakes (>1 km2) in 2015 and 17,167 glacial lakes with size in the range 0.0036 to 1 km2 in 2015. Glacial lakes are derived from http://data.tpdc.ac.cn/en/. (B) Reservoir and dam distributions in China. The boundary of the Tibetan Plateau (TP) is outlined. Reservoir and dam inventories are from a global dataset at http://globaldamwatch.org/data/. (C) Lake number and area changes in China between the 1970s and 2015. (D) Lake number and area changes in the TP between the 1970s and 2015. Lake number and area changes are from Zhang et al. (5).

In addition, Feng et al. (1) indicate that the WBs in China follow the size–abundance relationships. Although these relationships have been widely used, this hypothesis is reported to be inconsistent (e.g., refs. 9 and 10). Finally, a comparison of changes was presented in China’s WBs between 1985−1999 and 2000−2015 (1). For the first of these time periods, there was a limited availability of Landsat images (2, 5), which could result in large uncertainties in the comparison of WB changes. We suggest that a rigid criteria of data selections and fully justified comparison and analysis are important for discovering surface water features and their variations.