Improved prediction of severe thunderstorms over the Indian Monsoon region using high-resolution soil moisture and temperature initialization

Abstract

The hypothesis that realistic land conditions such as soil moisture/soil temperature (SM/ST) can significantly improve the modeling of mesoscale deep convection is tested over the Indian monsoon region (IMR). A high resolution (3 km foot print) SM/ST dataset prepared from a land data assimilation system, as part of a national monsoon mission project, showed close agreement with observations. Experiments are conducted with (LDAS) and without (CNTL) initialization of SM/ST dataset. Results highlight the significance of realistic land surface conditions on numerical prediction of initiation, movement and timing of severe thunderstorms as compared to that currently being initialized by climatological fields in CNTL run. Realistic land conditions improved mass flux, convective updrafts and diabatic heating in the boundary layer that contributed to low level positive potential vorticity. The LDAS run reproduced reflectivity echoes and associated rainfall bands more efficiently. Improper representation of surface conditions in CNTL run limit the evolution boundary layer processes and thereby failed to simulate convection at right time and place. These findings thus provide strong support to the role land conditions play in impacting the deep convection over the IMR. These findings also have direct implications for improving heavy rain forecasting over the IMR, by developing realistic land conditions.

Assessment of the Weather Research and Forecasting (WRF) model for simulation of extreme rainfall events in the upper Ganga Basin

“…In the case of Noah, soil moisture varies in response to the rainfall and is found to vary between 0.25 and 0.45 m 3 m −3 . Higher surface moisture conditions improve mass flux, convective updrafts and diabatic heating in the boundary layer that contributes to low-level positive potential vorticity or convective potential, which leads to enhanced rainfall potential (Osuri et al, 2017a). The importance of representing soil moisture variability over India for extreme weather conditions is also highlighted through this work.…”

“…However, the underestimation of rainfall is less in Domain 2b (G2C scale) compared to others, indicating the necessity of finer grid spacing as the first-order requirement for simulating the magnitudes of the extremely heavy rainfall events. The bias in the WRF simulations is typically due to a number of interactive factors: (i) scale feedback between mesoscale convection and large-scale processes within the model (Bohra et al, 2006), (ii) lack of local observations that can add mesoscale features Osuri et al, 2015), (iii) lack of proper land surface processes (Niyogi et al, 2006;Chang et al, 2009;Osuri et al, 2017a), and (iv) the inability of the model to fully resolve the complex topography (Argüeso et al, 2011;Cardoso et al, 2013;Chevuturi et al, 2015). To assess the performance of the WRF simulations, quantitative scores (MAE and RMSE) with respect to the observed data are computed for daily rainfall data, which is then averaged over the 4-day period.…”

“…The advanced research version of the Weather Research and Forecasting model (hereafter referred to as the WRF model) is a regional popular community model that is widely used for both studying as well as forecasting a variety of high-impact meteorological events, such as rainfall (Vaidya and Kulkarni, 2007;Deb et al, 2008;Kumar et al, 2008;Chang et al, 2009;Routray et al, 2010;Mohanty et al, 2012), tropical cyclones (Raju et al, 2011;Routray et al, 2016;Osuri et al, 2017b) and thunderstorms (Madala et al, 2014;Osuri et al, 2017a). Several works are reported in the literature which have considered the WRF model over the Himalayan region.…”

Evaluation of Machine Learning Classifiers for Predicting Deep Convection

“…Interestingly, the most important feature for this domain is reported as soil temperature, which also came second in a boosted tree model. Whether this result is physical cannot be confirmed, but the significance of land surface conditions on the initiation of deep convection over the Indian Monsoon region has been noted previously in Osuri et al (2017).…”

Lessons Learned From Modeling Irrigation From Field to Regional Scales

“…Journal of Advances in Modeling Earth Systems parallelized LSM driver that can be executed for a single point (e.g., field-scale simulations) and for continental scales. It has evolved from a soil-state initialization tool to an efficient framework to test and evaluate the LSM (e.g., Gao et al, 2015;Li et al, 2018;Osuri et al, 2017;Xin et al, 2018;Zhang et al, 2016).…”