基于MODIS的新疆地表温度时空变化特征分析

地表温度（LST）作为陆地环境相互作用过程研究中的重要地学参量,其时空分异与下垫面属性、地形地貌格局等息息相关。本文基于2000—2020年MOD11C3数据,采用GIS空间分析、空间重心模型及改进的半径法等手段,对新疆地区月、季、年及昼夜等序列时空尺度LST的分布特征及变化趋势进行研究,并借助LUCC、DEM等多源遥感数据进一步探讨各类下垫面、不同地貌单元及部分典型区域的LST垂直递减率、增减温及热（冷）岛效应等地理特征。结果表明：① 研究期间新疆年均LST为9.45 ℃,呈减温和增温趋势的区域分别占研究区面积的13.3%和86.7%,增温区域占绝大部分,故近20年研究区整体呈升温趋势（0.024 ℃·a^-1）,较全国高出约0.01 ℃·a^-1,其中春季增幅最大,冬季次之;夜间LST变化明显强于白天,春、夏两季昼夜温差均大于23.7 ℃,而冬季最小（约15 ℃）;研究区年际高、低温重心迁移轨迹大多分布于LST变幅较大区域,各月LST空间分布及高低温重心转移轨迹均表现出以7月为中点的年内强对称性规律。② 不同下垫面地表冷热环境空间分布差异较显著。其中以荒漠和裸土为主的未利用地年均LST最高（13.42 ℃）且昼夜温差最大（23.6 ℃）,高寒区冰川年均LST最低（-7 ℃）且昼夜温差（14.3 ℃）最小,其他下垫面年均LST在2.6~11 ℃之间,昼夜温差较小且较一致（平均16.75 ℃）。③ 东西走向的“三山夹两盆”地貌结构,使新疆LST的纬度地带规律被大幅削弱。研究区各山区（群）LST垂直递减率不尽相同,地处较高纬的阿尔泰山递变（0.63 ℃/100 m）最明显,山体高大的帕喀昆阿山群递变（0.57 ℃/100 m）次之,天山山脉递变（0.54 ℃/100 m）最小。另外各山区（群）山麓处均出现了不同程度的逆温现象,而山腰处LST垂直递减率线性拟合效果最佳。④ 本研究所选取的部分典型城市大都表现出不同程度的热岛效应。其中乌鲁木齐市热岛效应最为强烈,伊宁市其次,哈密市最微弱,而阿克苏市则表现出绿洲城市冷岛效应,但各城市热（冷）岛效应均存在一定环数（8环）范围的冷、热岛足迹。

Spatio-temporal variation characteristics of land surface temperature in Xinjiang based on MODIS

Land surface temperature (LST) is an extremely important geological parameter in the study of the process of terrestrial environmental interaction, and its spatio-temporal differentiation is closely related to the attributes of the underlying surface and the topography and geomorphology pattern. Based on MOD11C3 data from 2000 to 2020, this paper uses GIS spatial analysis, spatial center of gravity model and improved radius method to study the distribution characteristics and changing trends of LST on the spatial and temporal scales of the month, season, year, day and night sequences in Xinjiang, and with the help of multi-source remote sensing data such as LUCC and DEM, we further explore the geographical characteristics of various underlying surfaces, different geomorphic units and some typical areas of LST vertical decline rate, temperature increase and decrease, and urban heat (cold) island effect. The results show that: (1) during the study period, the average annual LST in Xinjiang was 9.45 ℃. The areas with decreasing temperature and increasing temperature accounted for 13.3% and 86.7% of the study area, respectively. So in the past 20 years, the study area showed an overall warming trend (0.024℃·a^-1), which was about 0.01 ℃·a^-1 higher than that of the whole country. The largest increase was observed in spring, followed by winter; the change of LST at night was significantly stronger than that in the daytime. The temperature differences between day and night in both spring and summer are greater than 23.7 ℃, and the smallest temperature difference is in winter (about 15 ℃); the interannual high and low temperature centers of gravity migration trajectory in the study area is mostly distributed in the area with large LST variation, and the spatial distribution of LST and the centers of high and low temperature in each month The transfer trajectories all showed strong symmetry during the year with July as the midpoint. (2) The spatial distribution of the hot and cold environments on different underlying surfaces is significantly different. Among them, the unused land dominated by deserts and bare soil has the highest average annual LST (13.42 ℃) and the largest day-night temperature difference (23.6 ℃). The glaciers in alpine regions have the lowest annual average LST (-7 ℃) and the lowest day-night temperature difference (14.3 ℃). The average annual LST of other underlying surfaces is between 2.6-11 ℃, and the temperature difference between day and night is small and consistent (average 16.75 ℃). (3) The east-west trending of “three mountains and two basins” geomorphic structure greatly makes the latitude zone of Xinjiang LST less stable. The vertical decline rates of LST in the mountainous areas (groups) of the study area are not the same. The LSTs of the Altay Mountains at higher latitudes vary (0.63 ℃/100 m), followed by the Pakakun's Group with high mountains (0.57 ℃/100 m), and the Tianshan Mountains (0.54 ℃/100 m). In addition, various degrees of temperature inversion occurred in the foothills of the mountainous areas (groups), and the linear fitting effect of the LST vertical decline rate at the mountainside was the largest. (4) Most of the typical cities selected in this study show different degrees of urban heat island effect. Among them, Urumqi city has the strongest urban heat island effect, followed by Yining city, while Hami city is the weakest; Aksu city exhibits the oasis urban cold island effect, but the urban heat (cold) island effect of each city has a certain number of rings (8 rings) about urban heat (cold) island footprints. The research results can provide theoretical support for soil development and protection, regional agricultural production and planting, urban engineering construction, as well as production and life in Xinjiang.