Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Challenges of Data Scarcity in Statistical Downscaling of Rainfall Using Large-Scale GCM Models
In climate change impact studies, statistical downscaling is considered to be quite effective while predicting future data of any meteorological parameter. However, complications arise when available data is of short duration. In the northeastern states of India, researchers have been facing this difficulty of scarcity of data for a long time. Short duration of observed data makes it difficult to choose suitable datasets for calibration and validation. The future dataset obtained from downscaling these datasets will be acceptable, if the calibrated model is precisely validated before using in future prediction. Therefore, we need a new method to find out a suitable dataset for calibrating the model. In the present study, hence, a novel approach is being tried to determine the best fit dataset for calibrating a statistical downscaling model. Five IMD stations, situated in the northeastern region are considered as focused area in this study. Six different combinations of calibration–validation sets are compared to find out their effectiveness and related challenges. To find out the correlation of the calibration–validation datasets, statistical parameters are evaluated for all the six combinations, and for all the stations. The results suggest that, the best combination is the one with higher R2 value and lower errors in RMSE, RE_μ, RE_ϭ and percentage change in maximum. Hence, we may say that lower the errors, better is the combination and better is the model calibrated.
Challenges of Data Scarcity in Statistical Downscaling of Rainfall Using Large-Scale GCM Models
In climate change impact studies, statistical downscaling is considered to be quite effective while predicting future data of any meteorological parameter. However, complications arise when available data is of short duration. In the northeastern states of India, researchers have been facing this difficulty of scarcity of data for a long time. Short duration of observed data makes it difficult to choose suitable datasets for calibration and validation. The future dataset obtained from downscaling these datasets will be acceptable, if the calibrated model is precisely validated before using in future prediction. Therefore, we need a new method to find out a suitable dataset for calibrating the model. In the present study, hence, a novel approach is being tried to determine the best fit dataset for calibrating a statistical downscaling model. Five IMD stations, situated in the northeastern region are considered as focused area in this study. Six different combinations of calibration–validation sets are compared to find out their effectiveness and related challenges. To find out the correlation of the calibration–validation datasets, statistical parameters are evaluated for all the six combinations, and for all the stations. The results suggest that, the best combination is the one with higher R2 value and lower errors in RMSE, RE_μ, RE_ϭ and percentage change in maximum. Hence, we may say that lower the errors, better is the combination and better is the model calibrated.
Challenges of Data Scarcity in Statistical Downscaling of Rainfall Using Large-Scale GCM Models
Advances in Sustainability sci. & technol.
Bhattacharjya, Rajib Kumar (Herausgeber:in) / Talukdar, Bipul (Herausgeber:in) / Katsifarakis, Konstantinos L. (Herausgeber:in) / Hazarika, Jayshree (Autor:in) / Sarma, Arup Kumar (Autor:in)
18.06.2022
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
| British Library Conference Proceedings | 2010
| Taylor & Francis Verlag | 2016
RAINFALL DOWNSCALING AND STOCHASTIC GENERATION
| British Library Conference Proceedings | 2007
Statistical Downscaling of Daily Rainfall Processes for Climate-Related Impact Assessment Studies
| British Library Conference Proceedings | 2011