Land surface temperature (LST) is a critical parameter at the land-atmosphere interface, playing a key role in hydrology, climate, and ecological studies. The widely used generalized split-window (GSW) algorithm assumes constant aerosol influence, yet aerosol variability introduces significant errors in LST retrieval. To address this limitation, we propose an improved algorithm (GSW_AOD) that explicitly incorporates an aerosol optical depth (AOD) correction term into the GSW framework. Evaluation using an independent simulation dataset demonstrated substantial improvement. Under low-aerosol conditions (AOD < 0.3), the root mean squared error (RMSE) decreased from approximately 1.7 K for GSW to 0.8 K for GSW_AOD. Under high-aerosol conditions (AOD ≥ 0.3), RMSE decreased more dramatically from approximately 4.0 K for GSW to 1.7 K for GSW_AOD. Validation with in situ measurements from eight ground sites showed that while both algorithms performed similarly at low AOD (AOD < 0.3; RMSEs between 2.0 K and 3.3 K, differences < 0.3 K), GSW_AOD significantly outperformed GSW under high aerosol loading (AOD ≥ 0.3), where GSW RMSEs reached 2.8 K to 4.0 K. Compared to the standard MODIS LST product, GSW_AOD achieved lower RMSEs at most sites for AOD < 0.3 and reduced RMSEs at all sites for AOD ≥ 0.3. These results confirm that explicitly correcting for variable AOD significantly enhances the accuracy and robustness of thermal infrared LST retrievals, particularly under moderate to high aerosol conditions.

This paper reports our five-year lessons of developing and using MobileInsight, an open-source community tool to enable software-defined full-stack, runtime mobile network analytics inside our phones. We present how MobileInsight evolves from a simple monitor to a community toolset with cross-layer analytics, energy-efficient real-time user-plane analytics, and extensible user-friendly analytics at the control and user planes. These features are enabled by various novel techniques, including cross-layer state machine tracking, missing data inference, and domain-specific cross-layer sampling. Their powerfulness is exemplified with a 5-year longitudinal study of operational mobile network latency using a 6.4TB dataset with 6.1 billion over-the-air messages. We further share lessons and insights of using MobileInsight by the community, as well as our visions of MobileInsight's past, present, and future.