Data Availability Statement
Supporting data for this study are openly available in MOD11A2 MODIS/Terra Land Surface Temperature and the Emissivity 8-Day L3 Global 1km SIN Grid athttp://doi.org/10.5067/MODIS/MOD11A2.006, reference number (NASA LP DAAC, 2019) and MYD11B2 MODIS/Aqua Land Surface Temperature and the Emissivity 8-day L3 Global 6km SIN Grid athttp://doi.org/10.5067/MODIS/MYD11B2.006, reference number (NASA LP DAAC, 2019)
1. Introduction
Water temperature of inland water bodies is a critical parameter that governs an array of aquatic ecosystem functions that involve physical, chemical, and biological processes (Li, Chen, & Zhang, 2015; Ma et al., 2016). Inland water temperature is regulated by multiple interacting factors, and excessive lake water warming has severe ecological consequences (Winslow, Read, Hansen, Rose, & Robertson, 2017; Kraemer, Mehner, & Adrian, 2017). Adrian et al. reported that some of the largest lakes across the world have shown notable temperature increases due to climate change (Adrian et al., 2009; Schmidt et al., 2018). For example, the temperature of Lake Baikal (Russia) has increased by approximately 1.2 °C since 1946 (Dörnhöfer & Oppelt, 2016). Lake temperature reflects its morphology, watershed conditions and hydrological dynamics, and variably influences the biology of resident aquatic organisms (Wetzel, 2001; Parastatidis, Mitraka, Chrysoulakis, & Abrams, 2017; Song et al., 2016). Conventional approaches for measuring lake water temperature using in situ sensors and data loggers provide information that is temporally continuous but spatially limited. These limitations therefore hinders widespread application of conventional approaches for capturing temperature variation at large spatial scales (Parastatidis et al., 2017). Yet, synoptic information is needed to capture thermal heterogeneity in large lakes, examine patterns of thermal variation, and explain fundamental biophysical and chemical processes in these water bodies (Hook, Vaughan, Tonooka, & Schladow, 2007; Ke & Song, 2014).
Water temperature of lakes is primarily influenced by their absorption of solar energy in a process affected by an array of physical, chemical and hydro-morphologic properties of lake ecosystems (Li et al., 2015; Woolway et al., 2018). In this respect, the surface temperature of lake water is highly dynamic, as it changes seasonally and diurnally due to variation in air temperature and the insulation effect of snow/ice (Li et al., 2015; Parastatidis et al., 2017; Song et al., 2016; Zhang et al., 2014). The amount of sunlight absorbed by water increases exponentially with the distance it travels through the water column, particularly for radiation wavelengths shorter than 750 nm (Schmidt et al., 2018; Song et al., 2016). The high specific heat of water enables the dissipation of light energy and its accumulation as heat in the water column. However, retention of that energy depends on multiple factors (wind speed, currents and other water movements, watershed geomorphology) influencing its distribution within a lake system, and the change rate between water input and discharge through the tributaries (Gorham, 1964). Thus, the pattern of thermal evolution and stratification influences fundamentally the cycling of physical and chemical components of lakes, which in turn drives primary productivity and decomposition processes (Wetzel, 2001).
Thermal remote sensing methods for monitoring lake water surface temperature (LWST) circumvent problems of accessibility to lakes in remote areas and so provide a synoptic context for evaluating relationships between landscape features and water thermal characteristics(Kraemer et al., 2017; Moukomla & Blanken., 2016; Zhong, Notaro, & Vavrus, 2018). Thermal infrared (TIR) remote sensing can reliably map LWST and its circulation patterns in lakes using various satellite sensors (Ke & Song, 2014). Of those, the Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectroradiometer (MODIS) on board Terra/Aqua have been widely used for monitoring lake surface temperatures (Zhang et al., 2014; Phillips, Saylor, Kaye, & Gibert, 2016). Landsat-Thematic Mapper(TM)/ Enhanced Thematic Mapper Plus (ETM+)/Operational Land Imager (OLI) and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data are commonly used to retrieve thermal features of inland and coastal waters (Kraemer et al., 2017; Trumpickas, Shuter, Minns, & Cyr, 2015).
Although LWST studies have been conducted at various scales around the world, the distribution pattern of LWST for Eurasia lakes has yet to be analyzed. Investigations about the relationship between LWST and factors regulating its spatiotemporal variations have been reported (Kraemer et al., 2017; Parastatidis et al., 2017; Trumpickas et al., 2015). The overall distribution pattern of LWST for Eurasia lakes deserves investigation (Zhang, Xie, Kang, Yi, & Ackley, 2011; Yuan et al., 2015).
In this study, 3,840 composites of MODIS imagery data covering the terrestrial area of Eurasia were processed and analyzed. The characteristics of LWST for 1,098 lakes that are at least 25 km2 in size were examined to determine spatial associations with climatic, landscape and hydrologic conditions. The objectives of this study were to (1) examine LWST variation of lakes at the Eurasian continental scale based on time-series MODIS LST data for the period 2001-2015; (2) examine intra-annual rates of temperature change of some typical large lakes in Eurasia; and (3) evaluate potential factors that contribute to variations in LWST.
  1. Materials and Methods
  2. Criteria of lakes selection
In the pre-satellite remote sensing stage, three types of independent approaches were generally used to obtain a global lake census: (1) a lake-type approach based on the origin of lakes, (2) an extrapolation of known regional censuses, and (3) a climatic approach based on lake distribution in homogeneous temperature and runoff environments (Meybeck, 2011). With respect to the last approach, it links lake distribution with five major climatic biomes, i.e., deglaciated regions (52%), temperate regions (13%), dry and arid regions (25%), deserts (1%), and wet tropics (9%). This approach has clear merits in combining temperature and surface-water runoff (Wetzel, 2001).
With the advent of satellite remote sensing, a global census of lake distribution and thermal characterization has become possible (Downing et al., 2006). Using MODIS LST products at a nominal spatial resolution of 1 km2, this study extracted lakes and water reservoirs larger than 25 km2 in Eurasia from the Global Lakes and Reservoirs Database (Lehner & Döll, 2004); this yielded a total of 1,098 water bodies (Figure 1). Among those, 861 lakes and 237 water reservoirs were in Asia and Europe, respectively. It should be noted that some lakes or reservoirs greater than 25 km2 might not have been selected if one of their dimension is smaller than 1-km, the spatial resolution of the MODIS LST product. Such elongated water bodies would likely to be located along shorelines. The selected 1,098 lakes cover 0.63 million km2, which is about 44.3% of the world’s total lake areas (Lehner & Döll, 2004).