Modeling high-resolution precipitation by coupling a regional climate model with a machine learning model: an application to Sai Gon–Dong Nai Rivers Basin in Vietnam

Trinh, T.; Do, N.; Nguyen, Van Thinh; Carr, K. J.

doi:10.1007/s00382-021-05833-6

Cited by 8 publications

(3 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1, studies have indicated that including atmospheric covariates is helpful for estimating precipitation (e.g., Baño-Medina et al, 2020;Li et al, 2022;Rasp and Lerch, 2018). The other three scenarios also consider atmospheric covariates of P from MERRA2 as predictors, which include geopotential height, specific humidity, air temperature, eastward wind, and northward wind at three different vertical levels (250, 500, 850 hPa) (e.g., Baño-Medina et al, 2020;Rasp and Lerch, 2018) as well as vertical wind (e.g., Trinh et al, 2021) at 500 hPa (OMEGA500), sea level pressure and 2 m air temperature in a single level (e.g., Panda et al, 2022;Rasp and Lerch, 2018) (see Table 2). We chose these variables based on precipitation formation theory (cloud mass movements and thermodynamics) as well as findings from previous studies as already indicated.…”

Section: Experimental Designmentioning

confidence: 99%

Customized deep learning for precipitation bias correction and downscaling

2023

View full text Add to dashboard Cite

Abstract. Systematic biases and coarse resolutions are major limitations of current precipitation datasets. Many deep learning (DL)-based studies have been conducted for precipitation bias correction and downscaling. However, it is still challenging for the current approaches to handle complex features of hourly precipitation, resulting in the incapability of reproducing small-scale features, such as extreme events. This study developed a customized DL model by incorporating customized loss functions, multitask learning and physically relevant covariates to bias correct and downscale hourly precipitation data. We designed six scenarios to systematically evaluate the added values of weighted loss functions, multitask learning, and atmospheric covariates compared to the regular DL and statistical approaches. The models were trained and tested using the Modern-era Retrospective Analysis for Research and Applications version 2 (MERRA2) reanalysis and the Stage IV radar observations over the northern coastal region of the Gulf of Mexico on an hourly time scale. We found that all the scenarios with weighted loss functions performed notably better than the other scenarios with conventional loss functions and a quantile mapping-based approach at hourly, daily, and monthly time scales as well as extremes. Multitask learning showed improved performance on capturing fine features of extreme events and accounting for atmospheric covariates highly improved model performance at hourly and aggregated time scales, while the improvement is not as large as from weighted loss functions. We show that the customized DL model can better downscale and bias correct hourly precipitation datasets and provide improved precipitation estimates at fine spatial and temporal resolutions where regular DL and statistical methods experience challenges.

show abstract

Section: Experimental Designmentioning

confidence: 99%

Customized deep learning for precipitation bias correction and downscaling

2023

View full text Add to dashboard Cite

show abstract

“…This method primarily involves nesting regional climate models within global models to predict regional future climate change scenarios. Currently, widely used RCMs include models such as Weather Research and Forecasting (WRF), Regional Climate Mode (RegCM), and PRECIS [8,9]. These RCMs, to some extent, can effectively simulate the geographical distribution and seasonal variation characteristics of temperature and precipitation at the regional scale.…”

Section: Introductionmentioning

confidence: 99%

Dynamical Downscaling of Daily Extreme Temperatures over China Using PRECIS Model

Guo,

Jia,

Wang

et al. 2024

Sustainability

View full text Add to dashboard Cite

As global warming intensifies and the frequency of extreme weather events rises, posing a major threat to the world’s economy and sustainable development, accurate forecasting of future extreme events is of great significance to mankind’s response to extreme weather events and to the sustainable development of society. Global Climate Models (GCMs) have limitations in their applicability at regional scales due to their coarse resolution. Utilizing dynamical downscaling methods based on regional climate models (RCMs) is an essential approach to obtaining high-resolution climate simulation information in future. This study represents an attempt to extend the use of the Providing REgional Climates for Impacts Studies (PRECIS) regional climate model by employing the BCC-CSM2-MR model from the Beijing Climate Center to drive it, conducting downscaling experiments over China at a spatial resolution of 0.22° (25 km). The simulation and prediction of daily maximum and minimum temperatures across the China region are conducted, marking a significant effort to expand the usage of PRECIS with data from alternative GCMs. The results indicate that PRECIS performs well in simulating the daily maximum and minimum temperatures over the China region, accurately capturing their spatial distribution and demonstrating notable simulation capabilities for both cold and warm regions. In the annual cycle, the simulation performance of PRECIS is superior to its driving GCM, particularly during cold months (i.e., December and from January to May). Regarding future changes, the daily extreme temperatures in most regions are projected to increase gradually over time. In the early 21st century, the warming magnitude is approximately 1.5 °C, reaching around 3 °C by the end of the century, with even higher warming magnitudes exceeding 4.5 °C under the SSP585 scenario. Northern regions will experience greater warming magnitudes than southern regions, suggesting faster increases in extreme temperatures in higher latitudes. This paper provides forecasts of extreme temperatures in China, which will be useful for studying extreme events and for the government to make decisions in response to extreme events.

show abstract

“…Climate models are developed using well-established earth physical principles ( 17) and applying those principles allows the modelling of past, current and future climates ( 17,18 ). The IPCC global model has been widely recognised and used for a variety of climate change modelling ( [19][20][21][22] ).…”

Section: Introductionmentioning

confidence: 99%

Flood modelling shows increased impacts from flooding in southern Thailand as a result of climate change

Curoy

Ward

Wang

et al. 2023

Preprint

View full text Add to dashboard Cite

Flooding is one of the most commonly occurring natural disasters across the world. Its occurrence is predicted to become more frequent with climate change and associated rainfall increases. This study used a bespoke software Flowroute-i, developed by Ambiental, UK specialists in flood risk assessment and modelling, utlising meteorological and spatial data to produce flood maps. The study was conducted in 6 catchments in southern Thailand modelling flood depth and extent associated with high rainfall events with return periods of 20, 50 and 100 years. Both a present-day scenario and a future scenario (RCP 6.0) with projections to 2100 were modelled. The models suggest that there could be an increase of up to 37.5% in flood extent, particularly in the middle of the catchment. This was particularly evident on the eastern side of the Thai peninsula, Nakhon Si Thammarat, in part as a result of the large flat coastal plain adjacent to steep basin geomorphology. These results should allow appropriate agencies to initiate flood mitigation measures, as the impacts of present-day flood events in the studied areas have been noted to be particularly devastating to life, livelihoods, and infrastructure and this looks set to worsen in a warming world.

show abstract

Modeling high-resolution precipitation by coupling a regional climate model with a machine learning model: an application to Sai Gon–Dong Nai Rivers Basin in Vietnam

Cited by 8 publications

References 51 publications

Customized deep learning for precipitation bias correction and downscaling

Customized deep learning for precipitation bias correction and downscaling

Dynamical Downscaling of Daily Extreme Temperatures over China Using PRECIS Model

Flood modelling shows increased impacts from flooding in southern Thailand as a result of climate change

Contact Info

Product

Resources

About