Presentation is loading. Please wait.

Presentation is loading. Please wait.

Assessment of high-resolution simulations of precipitation and temperature characteristics over western Canada using WRF model Asong. Z.E (elvis.asong@usask.ca),

Similar presentations


Presentation on theme: "Assessment of high-resolution simulations of precipitation and temperature characteristics over western Canada using WRF model Asong. Z.E (elvis.asong@usask.ca),"— Presentation transcript:

1 Assessment of high-resolution simulations of precipitation and temperature characteristics over western Canada using WRF model Asong. Z.E Y. Li, H. S. Wheater, S. Kurkute, L. Chen Global Institute for Water Security, University of Saskatchewan, 11 Innovation Blvd, Saskatoon, SK, Canada ABSTRACT RESULTS SUMMARY Lack of accurate estimates of precipitation (P) and temperature (T) are an important limitation for hydrological and earth systems modelling in Canada. Ground-based measurements are inevitably limited, given the large land area and small population density, fail to capture the effects of mountain topography in important runoff-producing areas and suffer from gross inaccuracies associated with cold climate P and T processes. The capability of the current generation of atmospheric models to represent P and T is therefore of major interest for hydrological practice. The skill of a high-resolution 4-km convection resolving Regional Climate Model (RCM)―Weather Research and Forecasting (WRF) in capturing the statistics of daily-scale P and T over western Canada within the period 2000 – 2013 , using observational data sets for comparison is evaluated in this study. We analyze not only the mean pattern of P and T distributions, but also the inter-annual variability and trends in higher order climate statistics such as wet-dry day frequency, 95th percentile daily maximum T, 5th percentile daily minimum T, and 95th percentile daily P are evaluated against ground observations. This preliminary assessment should enable more informed application of high-resolution RCMs for the investigation of current and future changes in socio-economic and environmentally relevant hydro-climatic characteristics over this topographically complex region of western Canada. ANU WRF DJF MAM JJA SON The capability of a baseline high-resolution convection-permitting WRF simulation over western Canada is evaluated in terms of its ability to capture observed daily-scale P and T characteristics (e.g. means, extremes, temporal and spatial structures). In terms of capturing P and T means, WRF reproduces very well the observed spatial structure and magnitudes of both variables but with a tendency to over-estimate P intensity (cf. Fig.2). For annual cycles, WRF agrees well with EC-S (Fig.7) compared to ANU (Fig.6). For inter-annual anomalies (Fig.8), very close correspondence is found with anomaly correlations (r) ranging between 0.7 – 0.98 except for P in DJF over MRB where very low r (0.06) is found. In terms of extremes, WRF tends to over-estimate their seasonal intensities especially when compared against ANU unlike EC-S (cf. Fig.12). Basin-averaged root mean squared errors (RMSE) of T and P variables for mean, standard deviation, and linear trend with respect to ANU fields are shown in Table 1. Fig.2: Spatial patterns of mean seasonal daily P (mm/day) Fig.3: Spatial patterns of mean seasonal daily Tmin Fig.4: Spatial patterns of mean seasonal daily Tmax Fig.6: Annual cycle: WRF compared to ANU Fig.7: Annual cycle: WRF compared to EC-S Fig.8: Annual P and T anomalies Table 1: Basin-averaged RMSE of temperatures and precipitation variables for mean, standard deviation, and linear trend with respect to ANU. Temperature Extremes OBJECTIVES Asses the ability of a high-resolution 4-km convection resolving WRF simulation to capture the statistics of observed daily-scale P and T in terms of: Mean and Extremes CONCLUSIONS EXPERIMENTAL DESIGN Based on the analyses presented in this study, the following main conclusions can be drawn: The WRF 4-km simulation captures satisfactorily the observed spatial and temporal structures of P and T events over western Canada. Although with a tendency to better reproduce the long-term mean of P and T extremes, WRF generally captures the inter-annual variability and linear trends of these variables. Regionally, for T05, noticeable cold biases are found over the Rocky Mountains while warm biases are mostly in the eastern regions. For T95, there is an overall tendency towards cold biases over most of the domain. For P variables, WRF tends to produce generally wet biases over the entire domain for both mean, standard deviation and linear trends. However, further work is recommended to investigate how WRF simulates wind and moisture fields which could induce dynamical biases. This kind of assessment has important implications for the broader study of climate variability and change. Also, it can enable more informed application of high-resolution RCMs for the investigation of current and future changes in socio-economic and environmentally relevant hydro-climatic characteristics over this topographically complex region of western Canada. Limitations: (1) there are important drawbacks in the ground-based measurements of P that should be born in mind when making these comparisons. (2) As ANUSPLIN is a gridded product, it could suffer from over-smoothing of high P intensities compared with gauge data. Therefore, caution should be exercised when interpreting the comparisons presented in this study. Computational Domain d01 Fig.9: Pattern statistics (a—SRB: left, MRB: right) and violin plots (b) describing the performance of WRF in simulating the observed (ANU) monthly maxima of daily Tmax (JJA) and minima of Tmin (DJF) during 2000 – The circle in the x-axis of (a) is the reference (ANU) standard deviation Fig.10: Spatial pattern of annual T95 statistics: 13-year mean, standard deviation (Std), and linear trend for WRF and ANU. The associated bias (WRF – ANU is also shown) Fig.11: Spatial pattern of annual T05 statistics: 13-year mean, standard deviation (Std), and linear trend for WRF and ANU. The associated bias (WRF–ANU is also shown) Precipitation Extremes Fig.1: Study domain (d01). Red polygon—Saskatchewan River Basin (SRB), blue polygon—Mckenzie River Basin (MRB) Model Configuration Fig.12: Distributions of monthly maxima of daily P for all grid points in SRB and MRB. The cumulative densities of P are shown for WRF vs ANU (right) and WRF vs EC-S (left) Fig.13: Spatial distribution of maximum number of consecutive dry days (CDD). Threshold used is 1 mm Fig.14: Spatial distribution of maximum number of consecutive wet days (CWD). Threshold used is 1 mm MODEL EVALUATION Ground-based Observations: Gridded observation data set (ANUSPLIN―10 km) Non-gridded station data (EC-S) Analysis period: Oct 2000 – Oct 2013 Evaluation Statistics: Mean state: Long-term mean seasonal daily P and T Annual cycles and anomalies of P and T Extremes: Monthly maxima (minima) of daily Tmax (Tmin) Annual statistics of 95th percentile daily Tmax (T95) Annual statistics of 5th percentile daily Tmin (T05) Annual statistics of 95th percentile daily P (P95) The mean, inter-annual standard deviation and linear trend of these variables analyzed at each grid point ACKNOWLEDGEMENTS The financial support from the Canada Excellence Research Chair in Water Security is gratefully acknowledged. Thanks are due to Eva Mekis from Environment and Climate Change Canada for providing access to homogenized (EC-S) precipitation data used in this study. We also thank Dan McKenney and his team at Natural Resources Canada for making available the gridded ANUSPLIN data set. The 6-hourly ERA-interim data set was provided by ECMWF. Fig.15: Mean of P variables (# of P days: NRD; sum annual P: SAP; P event average: PEA=SAP/NRD) Fig.16: Standard deviation of P variables. Remaining convention same as in Fig.15 Fig.17: Linear trend of P variables. Remaining convention same as in Fig.15


Download ppt "Assessment of high-resolution simulations of precipitation and temperature characteristics over western Canada using WRF model Asong. Z.E (elvis.asong@usask.ca),"

Similar presentations


Ads by Google