Validating MODIS land surface temperature products using long-term nighttime ground measurements
Validating MODIS land surface temperature products using long-term nighttime ground measurements
复制标题
DOI：
10.1016/j.rse.2007.05.024
复制
发表时间：
2008-03
影响因子：
13.5
通讯作者：
Wenhui Wang;S. Liang;T. Meyers
中科院分区：
文献类型：
--
作者：
Wenhui Wang;S. Liang;T. Meyers
NASA Terra和Aqua地球观测系统卫星上的中分辨率成像光谱仪（MODIS）每天提供多种陆地表面温度（LST）产品。然而，这些产品还没有得到充分的验证。本文介绍了利用FLUXNET和碳欧洲综合项目（CarboEurope-IP）在8个植被站点上的长期地面测量，对MODIS Terra的两个日地表温度产品MOD11_L2（版本4）和MOD07_L2（版本4）进行验证的初步结果。由于地面测量的地表温度仅在每个验证站点的一个固定点上可用，因此该研究经过精心设计，通过使用1800多个MODIS Terra立交桥的夜间地面测量数据来缓解尺度不匹配问题。初步结果表明，在大多数情况下，MOD11_L2 LSTs的绝对偏差和均方根误差（RMSE）都小于MOD07_L2 LSTs。MOD11_L2 lst与布鲁金斯、奥杜邦、迦南谷和布莱克山等地的地面测量结果匹配良好，绝对偏差小于0.8°C，均方根误差小于1.7°C。在Fort Peck、Hainich、Tharandt和Bondville站点，MOD11_L2 LSTs被低估了2-3°C。MOD11_L2 lst的偏倚与MOD07_L2 lst的偏倚相关。由于MOD07_L2 LST积是MOD11_L2 LST算法的输入参数之一，MOD11_L2 LST的偏差可能会受到MOD07_L2 LST偏差的影响。这两种产品的误差与传感器视图天顶角的关系不大，但与地面空气温度、湿度、风速和土壤湿度无关。
The Moderate Resolution Imaging Spectroradiometer (MODIS), onboard the NASA Terra and Aqua Earth Observing System satellites, provides multiple land surface temperature (LST) products on a daily basis. However, these products have not been adequately validated. This paper presents preliminary results of validating two MODIS Terra daily LST products, MOD11_L2 (version 4) and MOD07_L2 (version 4), using the FLUXNET and Carbon Europe Integrated Project (CarboEurope-IP) long-term ground measurements over eight vegetated sites. Since ground-measured LSTs were only available over one fixed point in each validation site, the study was carefully designed to mitigate the scale mismatch issue by using nighttime ground measurements concurrent to more than 1800 MODIS Terra overpasses. The preliminary results show that MOD11_L2 LSTs have smaller absolute biases and root mean squared errors (RMSE) than those of MOD07_L2 LSTs in most cases. The match of MOD11_L2 LSTs with ground measurements in the Brookings, Audubon, Canaan Valley, and Black Hills sites is good, yielding absolute biases less than 0.8 °C and RMSEs less than 1.7 °C. In the Fort Peck, Hainich, Tharandt, and Bondville sites, MOD11_L2 LSTs were underestimated by 2–3 °C. Biases in MOD11_L2 LSTs correlate to those in MOD07_L2 LSTs. Since the MOD07_L2 LST product is one of the input parameters to the MOD11_L2 LST algorithm, biases in MOD11_L2 LSTs may be influenced by biases in MOD07_L2 LSTs. The errors in both products depend weakly on sensor view zenith angle but are independent of surface air temperature, humidity, wind speed, and soil moisture.