Terrestrial evapotranspiration (ET) is an important control factor for water cycling and energy transport, serving as a crucial link between ecological and hydrological processes. Accurate estimation of ET is essential for enhancing efficient utilization of water resources, improving agricultural productivity, preserving ecosystems, and advancing climate change research. Despite its significance, high spatiotemporal resolution continuous ET datasets remain scarce. In ET estimation, machine learning methods have been widely adopted, with tree-based machine learning models gaining increasing attention due to their computational efficiency and reliability accuracy. However, research comparing the performance of these models remains relatively limited. In this study, we use data from flux observation sites and various remote sensing sources to explore the performance of four tree-based machine learning models in ET estimation across the Contiguous United States (CONUS). Our findings demonstrate the proficient performance of all four models in estimating terrestrial ET across CONUS. Particularly noteworthy is the outstanding performance of the extremely randomized trees (ERT) model, showing a high correlation (R2 = 0.84), low bias (BIAS = −0.0003 mm/d), and low root mean square error (RMSE = 0.72 mm/d) with flux observation site data. Using this model, we successfully obtained a seamless terrestrial ET dataset (ERT_ET) with a spatial resolution of 1 km and multiple temporal resolutions (daily, 8-day, monthly, and seasonal) from 2008 to 2018 across the CONUS. Compared to the MOD16 ET product, the ERT_ET outperforms with a higher R2 by 0.40 and lower RMSE by 5.31 mm/8d, providing better performance in capturing detailed features. Moreover, our ERT_ET product is comparable to other widely used ET products (MOD16, PML-V2, and ETMonitor), further highlighting its reliability. These findings will contribute to studies in various fields, including global climate change, hydrological cycles, and drought monitoring.
For the needs of snow cover monitoring using multi-source remote sensing data, in the present article, based on the spectrum analysis of different depth and area of snow, the effect of snow depth on the results of snow cover retrieval using normalized difference snow index (NDSI) is discussed. Meanwhile, taking the HJ-1B and MODIS remote sensing data as an example, the snow area effect on the snow cover monitoring is also studied. The results show that: the difference of snow depth does not contribute to the retrieval results, while the snow area affects the results of retrieval to some extents because of the constraints of spatial resolution.
Obtaining the land use change and ecological carrying capacity change accurately and understanding the specific impact of land use change on biocapacity is of great significance for sustainable development.This study analyzed the influence of land use change on ecological carrying capacity, aimed at providing technical support for regional development planning.Land use change models are adopted to evaluate the dynamic change of land use in Wuhan Urban Agglomeration from 2000 to 2015, whose land use information is extracted from Landsat images.Ecological carrying capacity is calculated with 4 driving factors.The obtained results indicate that the trend of land use degree change is consistent with that of ecological carrying capacity.The relative change rates of different land use types in different cities also affect their biocapacity.We can effectively use the ecological resources, but the demand of human's development for resources is also growing rapidly.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)