L-band brightness temperature (TB) has been shown to provide the best sensitivity to soil moisture (SM) although C- and X-band based products offer a longer time-series from satellite-based measurements. Currently, global coverage SM is routinely produced from spaceborne measurements using all three frequency bands, but despite continued validation efforts of the products, the relative characteristics and performance of these observations have not been fully established. Therefore, this study focused on the parametrization of SM retrieval algorithms at L-, C- and X-bands using TB observations from the L-band radiometer on NASA's SM Active Passive (SMAP) mission and the C- and X-band channels of JAXA's Advanced Microwave Scanning Radiometer 2 (AMSR2) onboard the GCOM-W satellite. These can be applied in global SM retrieval algorithms using either one of the frequencies or a combination of them. The reference in situ SM data was obtained from 12 core validation sites across various land cover types around the world. The investigation highlighted the known challenges of retrieving SM from C- and X-band data compared to the higher sensitivity of the L-band data. Even with a site-specific retrieval algorithm parameterization, the mean correlation of the C- and X-band retrievals for the core validation site SM measurement were much lower than that for L-band, being 0.52 (0.54) and 0.45 (0.47) for vertical (horizontal) polarization, respectively, while for the L-band retrieval the corresponding values were 0.81 (0.77). The parameterization exercise showed that matching the C- and X-band TB measurements with an emission model was not difficult; the problem was relating the observations to SM under the influence of large roughness and vegetation effects. As a result, parameter optimization produced values for some sites that were not realistic or did not allow any practical sensitivity to SM at C- and X-band. Considering the L-band observations, the parameter optimization resulted in superior bias performance as compared to the operational SMAP product parameterization, but the sensitivity to SM changes (R and unbiased root mean square difference) did not improve markedly, or in some cases degraded at the expense of a smaller bias.

L 波段亮温 (TB) 已被证明可提供对土壤湿度 (SM) 的最佳灵敏度，尽管基于 C 波段和 X 波段的产品从基于卫星的测量中提供了更长的时间序列。目前，全球覆盖 SM 通常是使用所有三个频段从星载测量中产生的，但尽管对产品进行了持续的验证工作，但这些观测的相对特征和性能尚未完全确定。因此，本研究的重点是使用来自 NASA 的 SM 主动无源 (SMAP) 任务的 L 波段辐射计的 TB 观测数据以及美国国家航空航天局的 C 和 X 波段信道，对 L、C 和 X 波段的 SM 反演算法进行参数化。 GCOM-W 卫星上 JAXA 的高级微波扫描辐射计 2 (AMSR2)。这些可以使用频率之一或它们的组合应用于全局 SM 检索算法。参考原位 SM 数据来自全球各种土地覆盖类型的 12 个核心验证站点。与 L 波段数据的更高灵敏度相比，该调查强调了从 C 波段和 X 波段数据中检索 SM 的已知挑战。即使使用特定站点的检索算法参数化，核心验证站点 SM 测量的 C 和 X 波段检索的平均相关性也远低于 L 波段，分别为 0.52 (0.54) 和 0.45 (0.47)垂直（水平）极化，而对于 L 波段检索，相应的值为 0.81（0.77）。参数化练习表明，将 C 波段和 X 波段 TB 测量值与发射模型相匹配并不困难；问题是在大粗糙度和植被效应的影响下，将观测结果与 SM 联系起来。结果，参数优化为一些不现实或不允许对 C 和 X 波段的 SM 有任何实际敏感性的站点产生的值。考虑到 L 波段的观察结果，与可操作的 SMAP 产品参数化相比，参数优化产生了优越的偏差性能，但对 SM 变化（R 和无偏均方根差）的敏感性没有显着提高，或者在某些情况下下降到以较小的偏差为代价。结果，参数优化为一些不现实或不允许对 C 和 X 波段的 SM 有任何实际敏感性的站点产生的值。考虑到 L 波段的观察结果，与可操作的 SMAP 产品参数化相比，参数优化产生了优越的偏差性能，但对 SM 变化（R 和无偏均方根差）的敏感性没有显着提高，或者在某些情况下下降到以较小的偏差为代价。结果，参数优化为一些不现实或不允许对 C 和 X 波段的 SM 有任何实际敏感性的站点产生的值。考虑到 L 波段的观察结果，与可操作的 SMAP 产品参数化相比，参数优化产生了优越的偏差性能，但对 SM 变化（R 和无偏均方根差）的敏感性没有显着提高，或者在某些情况下下降到以较小的偏差为代价。