Abstract: Background Joshimath, a town situated on the southern slope of the Himalayas, is built on a paleo-landslide deposit overlying fractured metamorphic rocks near the Main Central Thrust (MCT) and Munsiari Thrust (MT). Its fragile geological setting, compounded by rapid urban expansion and increasingly frequent cryospheric dis-asters, has led to recurrent slope instabilities. The large-scale subsidence that occurred between 2022 and 2023 damaged over 860 buildings, drawing international concern. However, most previous studies provided only qualitative descriptions and lacked systematic quantitative analyses. Methods This study processed 239 Sentinel-1 as-cending-track images acquired from March 20, 2017, to June 18, 2025, using the Small Baseline Subset InSAR (SBAS-InSAR) technique to derive long-term surface deformation. The PELT change-point detection algorithm was applied to identify abrupt accelerations associated with the 2021 Ronti ice-rock avalanche. Singular Spectrum Analysis (SSA) was then used to decompose deformation sequences into trend and periodic components. Coupling analysis between SBAS-derived defor-mation and CRU-MSN climate datasets quantified the lagged correlations with tem-perature and precipitation. Finally, LSTM and SARIMAX models were established to predict deformation trends and assess potential risks across nine deformation zones. Results 1) The landslide in Joshimath exhibits pronounced spatial heterogeneity, with Zones Ⅲ and Ⅶ identified as the most active and high-risk areas. 2) multiple zones experienced abrupt deformation changes before and after the 2021 avalanche, marking it as a key external trigger. 3) both subsidence and uplift zones display clear seasonal patterns, with high temperatures and intense rainfall during June–September acting as dominant drivers, while several zones exhibit a 1–2-month lagged response to climatic forcing. 4) compared with SARIMAX, the LSTM model achieved higher accuracy in capturing nonlinear trends and early-warning signals, and the resulting risk heat map effectively delineated highly sensitive regions. Conclusions The 2021 Ronti ice-rock avalanche substantially intensified slope deformation in Joshimath. Temperature-induced thawing and rainfall-driven infiltration jointly govern the on-going instability. LSTM-based forecasting identifies Zones Ⅲ and Ⅶ as future high-risk areas. These results provide scientific support for early-warning systems and slope management in Himalayan mountain towns under a warming climate.