THE EFFECT OF VARIOUS PRECIPITATION DOWNSCALING METHODS ON THE SIMULATION OF STREAMFLOW IN THE YAKIMA RIVER. Eric P. Salathé Jr.

Transcription

1 THE EFFECT OF VARIOUS PRECIPITATION DOWNSCALING METHODS ON THE SIMULATION OF STREAMFLOW IN THE YAKIMA RIVER Eric P. Salathé Jr. Climate Impacts Group, Joint Institute for the Study of the Atmosphere and Ocean (JISAO) / School of Marine Affairs (SMA), University of Washington, Seattle May 2002

2 1. INTRODUCTION Streamflow models are an important tool in assessing climate impacts on water resources by simulating the streamflow associated with climate change scenarios. Essential climate forcings are temperature and precipitation, which are required at high spatial resolution (0.125 degrees latitude and longitude) to simulate flow in typical river basins. Climate models, however, are run at much coarser resolution (2 degrees) and do not resolve important mesoscale processes and surface features that control the regional precipitation. Various downscaling methods have been used to span the gap from climate models to regional scales. These methods vary greatly in the sophistication with which they account for various physical processes. At the simplest extreme are analog methods that assume that whenever certain large-scale features are observed, the mesoscale conditions will be repeated. Thus, the historical record may be searched for an analog to the simulated large-scale conditions to find the corresponding mesoscale conditions. At the most complex, a physical mesoscale model is nested within the global model in an attempt to account for the full range of atmospheric processes at fine scales. In this paper, various empirical methods that take into account increasing amounts of physical forcings will be compared in order to asses the importance of improvements in the downscaling for simulating streamflow. In the Pacific Northwest, the surface orography creates dramatically different precipitation zones over the horizontal distance smaller than a climate-model grid cell. The north-south Coastal and Cascade Mountain ranges create a powerful rainshadow with considerable precipitation falling to the western (up wind) side of the Cascades and arid conditions in the Columbia Basin and desert regions of Washington and Oregon to the east (Figure 1). Typical global models do not resolve this topography, neither explicitly nor in subgrid-scale parameterizations, and thus cannot simulate the characteristic regional

3 precipitation. In order to create precipitation fields appropriate to force a streamflow model, additional information must be added to the climate simulation to account for the mesoscale variations through a downscaling method. Various statistical methods for downscaling Pacific Northwest precipitation are proposed by Widmann, Bretherton, and Salathé (2002) to produce monthly-mean precipitation downscaled to a 50-km mesoscale grid. These empirical methods are based upon a daily 50-km gridded dataset of precipitation over Washington and Oregon (Widmann and Bretherton, 2000). While regional atmospheric models have great potential in simulating the authentic regional scale precipitation, they are limitied in their application due to computational demands. Furthermore, regional models are not without some of the problems facing global models, such as systematic bias (Busuioc, et al, 2001). Gershuov et al (2000) have compared dynamical scaling with a regional climate model with two simple empirical methods for forcasting ENSO impacts in California. Their results show that the regional model does not add sufficient skill over the empirical methods to justify its increased complexity and cost. Other studies, however, have shown promise in the use of regional models for climate-change studies (Rummukainen et al, Leung Hamlet Lettenmaier) Regional models are less likely to be limited in the simulation of future climates than empirical methods that neciusarily assume the connections between large and small scale variablity will remain the same as the climate changes. When the application of the downscaled precipitation is for hydrologic modeling, the skill of the method may not be accurately illustrated simply by comparing spacial patterns of precipitation. For example, the improved precipitation simulation of a more costly method may not be realized in the hyrologic simulation if the details are smoothed out over time and space by storage in snowpack. In many instances, the flow is simply a matter of total winter accumulation

4 with temperature controlling the timing of melt and thus changes in flow. In this paper, various empirical precipitation downscaling methods are evaluated against their skill in driving a hydrological model of the Yakima River in central Washington, USA. The Yakima River, which will be studied here, originates on the east, leeward, crest of the Cascade Range in southern Washington. It flows south east to join the Columbia River. The basin spans an altitude range of about 2400 m and drains about 16,000 km 2. The river is highly developed for hydropower and irrigation, with a substantial fraction of the flow diverted for these uses. The regional ecomomy of the basin is based on agriculture and associated industry. Leading products include fruits, vegitables, specialty crops (e.g. hops, mint), beef, and dairy. 2. Data and methods A. Downscaling Two downscaling methods developed in Widmann, Bretherton, and Salathé (2002) will be presented here. These are a local scaling and a dynamical scaling of the large-scale precipitation field. Each method represents monthly-mean precipitation as the product of the large-scale precipitation and a scaling factor that is resolved on the mesoscale grid. For the purposes of this study, large-scale precipitation is taken from the NCEP reanalyses, which represents a perfect GCM in the sense that the daily large-scale weather patterns conform to reality. The precipitation, however, is entirely model-generated, and simulates the precipitation that would occur with the actual large-scale conditions over much smoother boundary conditions that reality. In particular, the NCEP precipitation field is not forced by the true topography of the Cascade and Coastal Ranges, but caputures only a gradual east-west gradient representing the western flank of the Rocky Mountains. Thus, there are severe biases in the NCEP precipitation,

5 changing sign from too dry upwind to too wet downwind of the Cascades. Despite this bias, the NCEP precipitation is highly correlated in time with observations at each point (Widmann and Bretherton, 2000). The period is used to derive fitting parameters for the period , while the second period is used to derive parameters for the first, yielding downscaled precipitation for the full period In the local scaling method, the scaling factors are fixed for each season and are simply the ratios s = p / P, where p and P are observed and NCEP precipitation and L is the seasonal time-man over the fitting period at each grid point. Thus, the locally-scaled precipitation is constrained to have the same long-term seasonal mean as the observations at each 50-km gridpoint. The method relies on the fact that the NCEP precipitation is well correlated in time to the observations at any point even though it has quite large biases that vary from point to point (see Widmann and Bretherton, 2001). Figure 2 shows the monthly-mean NCEP large-scale (upper left) and observed mesoscale precipitation (upper right) fields for January The lower left panel shows the scaling factors for the December-January-February season (DJF). The lower right panel is the locally-scaled precipitation for January 1992, and is simply the product of the upper and lower left panels. The local scaling factor increases the simulated NCEP precipitation in the various mountainous regions and decreases it in the rainshadows. In the dynamical scaling, the effect of atmospheric circulation is taken into account and the scaling factor depends also on the monthly-mean 1000-hPa heights. On the East side of the Cascade range, the small amount of precipitation that does occur is highly dependent upon the atmospheric circulation, which modulates how much moisture can be carried past the mountains. The leading mode of variability in 1000 hpa heights, as revealed by EOF analysis, is a modulation of the mean southwesterly onshore flow between a more westerly and a more

6 southerly phase (Fig. 3). The first phase is associated with drier-than-average conditions east of the Cascades while the more southerly phase allows more precipitation to the rain shadow (Fig 4, top). The locally-scaled NCEP precipitation does not capture this effect (Fig 4, bottom). To account for this effect, the scaling factor is derived by constraining the covariance between downscaled precipitation and the leading three modes of the GCM 1000-hPa heights to be the same as the covariance between observed mesoscale precipitation and heights, in addition to preserving the long-term mean. In this case, since we are testing the results for a perfect GCM, the NCEP reanalysis heights are used for both the observations and GCM, but in general, the GCM heights would be different. For a complete description of this method, see Widmann, Bretherton, and Salathé (2002) Figure 5 shows the temporal correlation of the downscaled monthly-mean precipitation for the two methods with the observations during the period The upper left is for the local scaling, the upper right is for the dynamical scaling. The dynamical scaling produces a significant improvement in the result over the dry region in the lee of the Cascades, indicating a strong control of the precipitation distribution due to circulation patterns. The Yakima Basin falls along the edge of this region, indicating dynamics may exert an important control on the mesoscale precipitation. B. Hydrology simulations In order to explore the the effects of precipitation downscaling, streamflow simulations for the Yakima River were made with observed temperature and various precipitation data for the period using the Variable Infiltration Capacity (VIC) hydrology model (Liang et al., 1994) implemented at degree resolution. The VIC model, implemented at 0.25-degrees for the Columbia basin, is fully described fully in Matheussen et al. (1999). The simulation is for

7 virgin flow, in the absense of any water management. Results are presented for the period , allowing the first five years for model spin up Precipitation data were taken from the downscaled precipitation using the two methods described above. To generate daily precipitation from the downscaled monthly-means, the daily NCEP precipitation is interpolated to each grid point, normalized by its monthly mean, and multiplied by the downscaled monthly-mean. Thus, the daily variability comes from the NCEP analyses while the mean is taken from the downscaling. In general, correspondence between daily temperature and precipitation is essential to modelling the hydrology. Hence, taking temperature and precipitation from different sources is a potential source of error, but the NCEP reanalysis, as opposed to a free-running GCM, is sufficiently contrained to the daily weather patterns that any errors are minimal. For comparison, two additional precipitation datasets are used: 1) NCEP precipitation interpolated to the mesoscale grid and 2) climatological daily mean precipitation, which is applied cyclically. Differences of the downscaled methods from the first illustrates the information added by the downscalling to the raw NCEP precipitation. Differences from the second helps separate the effects of interannual temperature variability, since the climatological precipitation is the same for every year, so all variability arises only from temperature. 3. Results In mountainous regions, where there is considerable storage in snowpack, streamflow is determined both by temperature, which controls melting or storage in the snowpack, and precipitation, which affects the total available water and also directly controls streamflow when the precipitation falls as rain. Flow in the Yakima River has two principal seasonal maxima, one

8 due to melting of the high-altitude snowpack during early summer and a second due to rainwater in late fall. Figure 6 shows the total water-year flow during the simulation period. After the first few years where the hydrology model is spinning up, the downscaled precipitation datasets capture the main observed interannual features. The interpolated NCEP precipitation, in addition to yielding too little flow, does not capture much of the observed interannual variability missing, for example, the increased flows during Thus, downscaling is essential even in order to capture flow variability at interannual time scales. The two downscaling methods give indistinguishable results for annual flow. Table 1 shows the linear correlation and regression coeffidient (slope) of each simulation to the simulation with observed precipitation for the period The local and dynamical scaling methods perform equally well in capturing inter annual variability. The analog method, however, does not add significantly to the raw NCEP precipitation field. Table 1. Correlations with observed precipitation simulkation for total annual flow Model Correlation slope Local scaling Dynamical scaling Analog NCEP By examining the flow on a monthly scale, the details of the interannual variability are revealed. Figure 4 shows the period where there is strong interannual variability. Variability in the simulation with climatological precipitation is due to interannual variations only in temperature, thus changes relative to the dotted line are due to differences in the

9 precipitation datasets. During this period, the downscaling methods both capture the monthly variability quite well. The NCEP precipitation, in addition to a consistent dry bias, does not capture some of the significant seasonal features. For example, there is significant rain-driven flow in the fall of 1975 and of 1977, as compared the previous year where all the flow occurs during the spring melt. The NCEP precipitation does not capture this feature, which is well represented by simulations using precipitation from either downscaling method. Table 2 shows the correlation and regression coefficient for the various methods to the observed monthly precipitation for the full period, It is important to note that climatological precipication combined with observed temperatre yields a streamflow simulation 87% correlated to the simulatioon with observed precipitation. As with the yearly total flow, the two scaling methods perform equally well for the Yakima basin and yield near ideal correlation. The analog method does no better than climatological precipitation. The raw NCEP precipitation does significantly worse. Table 2. Correlations with observed precipitation simulation for monthly flow Model Correlation slope Local scaling Dynamical scaling Analog NCEP Climatology If the results for climatological precipitation are subtracted from the other results, to produce an anomaly relative to the climatological precipitation simulation, the information added by the downscaling of the NCEP precipitation analyses can be assessed. Table 3 shows results for correlation and regression of each anomaly against the anomaly for the observed precipitation

10 simulation. The local and dynamical scaling both retain a significant correlation to the observed simulation, showing considerable information about the mesoscale precipitation is captured. The analog method is marginally correlated while the raw NCEP analyses are essentially uncorrelated. Table 3. Correlations with observed precipitation simulation for monthly flow, after subtracting climatology simulation Model Correlation slope Local scaling Dynamical scaling Analog NCEP CONCLUSIONS These simulations illustrate how local scaling, a very simple and efficient statistical downscaling method, is able to capture all the essential precipitation features required for accurate simulation of flow in the Yakima River. The Yakima is in the dry rainshadow of the Cascades, and is subject to precipitation variability that is connected to the large-scale winds, as indicated by the results of Figure 2. Nevertheless, the quality of information needed to perform hydrologic simulations does not evidently require such detail given that the dynamical scaling, which considerably improves the downscaling in the rainshadow region, performs no better for driving the streamflow. The local scaling method is as efficient and simple to implement as the analog method yet the analog method does not add much useful information over climatological precipitation for this region.

11 Secondly, it is clear that statistical downscaling can accomplish more that remove a uniform bias in a large-scale model simulation. The redistribution of water, implied by the scaling, can profoundly effect the hydrograph. Without downscaling, such significant a feature as rain-driven fall flows cannot be simulated with the NCEP precipitation. The large-scale precipitation, however, does provide the interannual and interseasonal information needed to capture these features once it is scaled at high spatial resolution. In terms of application to climate change simulations, the scaling methods are not limited by past climate extremes and are easily calibrated to short runs of present-day climate. These methods do not rely on historical relationships between circulation patterns and precipitation, rather allowing the physical model to simulate that connection albiet poorly resolved. 5. REFERENCES Busuioc, A., D. Chen, and C. Hellström, 2001: Performance of statistical downscaling models in GCM validation and regional climate change estimates: application for swidish precipitation. Inernational Journal of Climatology, 21, Gershunov, A., T. P. Barnett, D. R. Cayan, T. Tubbs, and L. Goddard, 2000: Predicting and downscaling ENSO impacts on intraseasonal precipitation statistics in california: The 1997/98 event. J. Hydrometeorology, 1, Hamlet, A. and D. Lettenmaier, 1999: Effects of climate change on hydrology and water resources in the Columbia River Basin. J. Amer. Water Res. Assoc., 35, L.R. Leung, A.F. Hamlet, D.P. Lettenmaier, and A. Kumar, 1999: Simulations of the ENSO Hydroclimate Signals in the Pacific Northwest Columbia River Basin. Bulletin of the American Meteorological Society, 80,

12 Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges, 1994: A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophy. Res., 99, Rummukainen, M., J. R äisänen, B. Bringfelt, A. Ullerstig, A. Omstedt, U. Willénen, U. Hansson, C. Jones, 2001: A regional climate model for northern Europe: Model description and results from the downsdcaling of two GCM control simulations. Climate Dynamics, 17, Widmann, M and C. S. Bretherton, 2000: Validation of mesoscale precipitation in the NCEP reanalysis using a new grid-cell precipitation dataset for the Northwestern United States. J. Climate, 13, Widmann, M, C. S. Bretherton, and E. P. Salathé, 2002: Precipitation downscaling over the Northwestern United States using numerically simulated precipitation as a predictor. J. Climate, Submitted. Wood, A.W., Maurer, E.P., Kumar, A. and D.P. Lettenmaier, Long Range Experimental Hydrologic Forecasting for the Eastern U.S. J. Geophys. Res. (accepted). Additional refs need to cite: Daly, C., R. P. Neilson, and D. L. Phillips, 1994: A statistical-topographic model for mapping climatological precipitation over mountainous terrain. J. Appl. Met., 33, Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell, S. Saha, G. White, J. Woollen, Y. Zhu, M. Chelliah, W. Ebisuzaki, W. Higgins J. Janowiak, K. C. Mo, C. Ropelewski, J. Wang, A. Leetmaa, R. Reynolds, R. Jenne, and D.Joseph, 1996: The NCEP/NCAR reanalysis project. Bull. Amer. Meteor. Soc., 77,

13 Kidson, J. W. and C. S. Thompson, 1998: A comparison of statistical and model-based downscaling techniques for estimating local climate variations. J. Climate, 11, Murphy, J., 1999: An evaluation of statistical and dynamical techniques for downscaling local climate. J. Climate, 12,