• 3-D Radiation-Topography Interactions (3DRT) incorporated in TaiESM1

• How does this parameterization scheme improve climate models?

It is well-established that strong land-atmosphere interactions over the Tibetan Plateau (TP) significantly impact the Indian Summer Monsoon (ISM). To understand the thermal deviations caused by TP's complex topography and their effects on ISM onset, we compared TaiESM models with (3D_on) and without (3D_off) the 3-D topography effect (3DRT). The 3DRT is a parameterization of solar fluxes over mountain surfaces utilizing the multiple linear regression analysis associated with topography developed by Lee et al. (2011) based on 3‐D Monte Carlo photon tracing simulations. Our findings indicate that the 3D_on tends to cause earlier ISM onsets, reducing model biases. A related paper is currently being written and is planned for submission later this year.

Recently, we found that the ENSO amplitude and standard deviation in 3D_on performed better than in 3D_off, as its results are closer to observations. We look forward to further exploring the mechanisms linking 3DRT and ENSO within the model.

The Falling Ice Radiative Effects (FIREs) parameterization scheme estimates the contribution of precipitating ice to radiative fluxes. Many climate models fail to take FIREs into account, which may lead to underestimations of shortwave albedo and downward longwave radiation. A few studies have analyzed the impact of including FIREs, finding that these effects contribute to improved simulations of Antarctic sea ice (Li et al., 2017) and reducing SST biases (Li et al., 2022).

• Cloudless optical images are often too slow to acquire in emergencies!

• SAR images are challenging to interpret, particularly when detecting small landslides.

In satellite remote sensing, the multi-band information of optical images is widely used to interpret landslide areas. However, it usually takes several weeks or even months to acquire cloudless optical images. Data from SAR images—including phase, intensity, and polarization—are used in related landslide detection research, but these methods are limited by poor interpretation when detecting landslides smaller than 5 hectares. To detect small landslides, we apply statistical-based anomaly detection methods, including statistical thresholding and the Gaussian model, to identify abnormal signals from landslides.