Twenty wetland-related data products (including remote sensing datasets, compilation datasets and model simulation datasets) were collected to evaluate the characteristics (area and distribution) of the wetlands in the Qinghai–Tibet Plateau (QTP) during four stages (1980s, 1990s, 2000s, and 2010s). We conducted a statistical analysis of the wetland areas from different datasets and compared the pixel consistency regarding wetland spatial distribution. The results showed that high uncertainty exists in the wetland area and low consistency exists in the distribution among the different datasets. The wetland area in the QTP ranged from 1.5 × 10⁴ to 121.16 × 10⁴ km². In the remote sensing datasets, the wetland area in the QTP ranged from 3.25 × 10⁴ to 11.28 × 10⁴ km², the calculated area was between 1.50 × 10⁴ and 72.21 × 10⁴ km² in the compilation datasets, and the area simulated from model datasets was between 3.81 × 10⁴ and 121.16 × 10⁴ km². For the total wetland area in the QTP, the uncertainty in the measured datasets was lower than that in the model simulation datasets. However, for the distribution of wetlands, the measured datasets were more inconsistent than the model datasets. In the measured datasets, as the pixel consistency increased, the corresponding probability throughout the area decreased. The probability of achieving 75% consistency was less than 2%, and was 0.61%, 0.35%, 1%, and 1.43% in the four stages, respectively. In the model products, the probability of achieving 75% consistency was 40.39%. Our study will enrich the global wetland database and contribute to the establishment of a plateau wetland information system, which will be significant for the protection and management of wetlands.

中文翻译：

---

基于多源数据的青藏高原湿地分布的高度不确定性

收集了二十个与湿地有关的数据产品（包括遥感数据集，编辑数据集和模型模拟数据集），以评估青藏高原（QTP）四个阶段（1980年代，1990年代， 2000年代和2010年代）。我们对不同数据集的湿地面积进行了统计分析，并比较了有关湿地空间分布的像素一致性。结果表明，湿地地区存在高度不确定性，不同数据集之间的分布一致性较低。QTP的湿地面积为1.5×10 ⁴到121.16×10 ⁴ km ²。在遥感数据中，QTP的湿地面积为3.25×10 ^4。到11.28×10 ⁴ km ²，在编译数据集中计算的面积在1.50×10 ⁴和72.21×10 ⁴ km ²之间，从模型数据集模拟的面积在3.81×10 ⁴和121.16×10 ⁴ km ^2之间。对于QTP中的总湿地面积，实测数据集中的不确定性低于模型模拟数据集中的不确定性。但是，对于湿地的分布，测得的数据集比模型数据集更加不一致。在测量的数据集中，随着像素一致性的提高，整个区域的相应概率降低。在这四个阶段中，达到75％一致性的可能性小于2％，分别为0.61％，0.35％，1％和1.43％。在模型产品中，达到75％一致性的可能性为40.39％。我们的研究将丰富全球湿地数据库，并有助于建立高原湿地信息系统，这对于保护和管理湿地将具有重要意义。