The direct impact of high resolution land surface initialization on the forecast bias in a regional climatemodel in recent years over Indian summer monsoon region is investigated. Two sets of regional climatemodel simulations are performed, one with a coarse resolution land surface initial conditions and secondone used a high resolution land surface data for initial condition. The results show that all monsoonyears respond differently to the high resolution land surface initialization. The drought monsoon year2009 and extended break periods were more sensitive to the high resolution land surface initialization.These results suggest that the drought monsoon year predictions can be improved with high resolutionland surface initialization. Result also shows that there are differences in the response to the land surfaceinitialization within the monsoon season. Case studies of heat wave and a monsoon depression simulationshow that, the model biases were also improved with high resolution land surface initialization. Theseresults show the need for a better land surface initialization strategy in high resolution regional modelsfor monsoon forecasting.

Surface level soil moisture from two gridded datasets over India are evaluated in this study. The firstone is the UK Met Office (UKMO) soil moisture analysis produced by a land data assimilation systembased on Extended Kalman Filter method (EKF), which make use of satellite observation of AdvancedScatterometer (ASCAT) soil wetness index as well as the screen level meteorological observations. Seconddataset is a satellite soil moisture product, produced by National Remote Sensing Centre (NRSC) usingpassive microwave Advanced Microwave Scanning Radiometer 2 measurements. In-situ observations ofsoil moisture from India Meteorological Department (IMD) are used for the validation of the gridded soilmoisture products. The difference between these datasets over India is minimum in the non-monsoonmonths and over agricultural regions. It is seen that the NRSC data is slightly drier (0.05%) and UKMOsoil moisture analysis is relatively wet during southwest monsoon season. Standard AMSR-2 satellitesoil moisture product is used to compare the NRSC and UKMO products. The standard AMSR-2 andUKMO values are closer in monsoon season and AMSR-2 soil moisture is higher than UKMO in allseasons. NRSC and AMSR-2 showed a correlation of 0.83 (significant at 0.01 level). The probabilitydistribution of IMD soil moisture observation peaks at 0.25 m^3/m^3, NRSC at 0.15 m^3/m^3, AMSR-2 at0.25 m3/m3 and UKMO at 0.35 m^3/m^3 during June–September period. Validation results show UKMOanalysis has better correlation with in-situ observations compared to the NRSC and AMSR-2 datasets.The seasonal variation in soil moisture is better represented in UKMO analysis. Underestimation of soilmoisture during monsoon season over India in NRSC data suggests the necessity of incorporating theactual vegetation for a better soil moisture retrieval using passive microwave sensors. Both productshave good agreement over bare soil, shrubs and grassland compared to needle leaf tree, broad leaf treeand urban land cover types.