Historical reconstruction of aquatic vegetation of typical lakes in Northeast China based on an improved CA-Markov model

Aquatic vegetation is an important marker of the change in the lake ecosystem. It plays an important supporting role in the lake ecosystem, and its abundance and cover changes affect the ecosystem balance. Collecting accurate long-term distribution data on aquatic vegetation can help monitor the change in the lake ecosystem, thereby providing scientific support for efforts to maintain the balance of the ecosystem. This work aimed to establish an improved CA-Markov model to reconstruct historical potential distribution of aquatic vegetation in the two typical lakes (Xingkai Lake and Hulun Lake) in Northeast China during 1950s to 1960s. We firstly analyzed remote sensing data on the spatial distribution of aquatic vegetation data in two lakes in six periods from the 1970 to 2015. Then, we built a transfer probability matrix for changes in hydrothermal conditions (temperature and precipitation) based on similar periods, and we designed suitability images using the spatial frequency and temporal continuity of the constraints. Finally, we established an improved CA-Markov model based on the transfer probability matrix and suitability images to reconstruct the potential distributions of aquatic vegetation in the two northeastern lakes during the 1950s and 1960s. The results showed the areas of aquatic vegetation in the 1950s and 1960s were 102.37 km(2) and 100.7 km(2) for Xingkai Lake and 90.81 km(2) and 88.15 km(2) for Hulun Lake, respectively. Compared with the traditional CA-Markov model, the overall accuracy of the improved model increased by more than 50%, which proved the improved CA-Markov model can be used to effectively reconstruct the historical potential distribution of aquatic vegetation. This study provides an accurate methodology for simulating the potential historical distributions of aquatic vegetation to enrich the study of the historical evolution of lake ecosystem.