Evaluating high-resolution global NWP model day-1 precipitation forecasts for the Indian monsoon: Potential to use in real-time applications.

This study aimed to investigate whether the current state of high-resolution operational numerical weather prediction (NWP) model forecasts has advanced to a level where they could supplant satellite-based precipitation estimates for real-time applications. Recent advancements in data assimilation, model physics, and computing power have enabled short-term NWP model forecasts to compete with satellite-based rainfall estimates, offering the added benefit of a high spatial and temporal resolution. This study evaluates the 24 hr rainfall forecast of three global NWP models: (i) the NCUM (National Centre for Medium Range Weather Forecasting (NCMRWF) Unified Model), (ii) the UKMO (United Kingdom (UK) Met Office (UKMO) model), and (iii) the IMD-GFS (Indian Meteorological Department (IMD)-Global Forecast System (GFS)) model along with the MEPS (multi-model ensemble predicted system) and two satellite precipitation products (IMERG: Integrated multi-satellite retrievals for global precipitation measurement and GSMaP: Global satellite mapping of precipitation) against the IMD gauge gridded observations for the 2016–2023 summer monsoon seasons at a daily scale. The findings indicate that 24 hr model forecasts are comparable to those of satellite-based products. In certain scenarios, these forecasts exhibited a better performance in capturing spatiotemporal variations and detecting precipitation, albeit with a slightly higher rate of false alarms at lower thresholds. The NCUM and UKMO models exhibit approximately similar forecast skills, yet are marginally better than the GFS model. However, MEPS shows a good prediction of spatiotemporal variation but struggles with weak precipitation overestimation and high-intensity precipitation underestimation. This indicates that a simple average of the multi-model product does not enhance rainfall frequency or magnitude prediction. GSMaP shows a limited ability to detect monsoon rainfall, whereas IMERG performs moderately better but consistently overestimates at all rainfall thresholds. These findings provide valuable insights for incorporating forecasted precipitation estimates into day-to-day weather monitoring over the Indian monsoon areas, serving as a useful reference for precipitation retrieval algorithms for future generations. Research highlights: This study, for the first time, comprehensively evaluates the long-term (2016–2023) rainfall from three different operational models forecast and two different satellite products against the observed station-based rainfall over India. This study basically tries to establish whether the NWP short forecasts have the potential to replace the satellite-based precipitation estimates. These results are useful for researchers, weather information centres, and developers of operational models and satellite retrieval algorithms for the better usage of short forecasts for day-to-day weather monitoring. This study also has the potential future scope to improve short forecasts from NWP models by various post-processing methods. [ABSTRACT FROM AUTHOR]