Land surface modeling for real-time hydrologic prediction and drought forecasting Dennis P. Lettenmaier Department of Civil and Environmental Engineering University of Washington Meeting of the Americas 2008 Joint Assembly Fort Lauderdale, FL May 29, 2008

Introduction: LSM structure

Variable Infiltration Capacity (VIC) snow model schematic

Off-line testing and intercomparisons Participating models in PILPS-2e (from Nijssen et al, 2003)

PILPS-2e model intercomparison wrt reproduction of mean seasonal distribution of streamflow

Relative bias in streamflow for N-LDAS models (from Mitchell et al, 2004)

Mean snow cover days over CONUS for four N-LDAS models (from Sheffield et al, 2003)

University of Washington West-wide Seasonal Hydrologic Forecast System A real-time simulation test-bed for climate forecast use in hydrology and water resources satellite data assimilation multi-model approaches forecast verification LDAS-era land surface models

Initial Conditions: Hydrologic Simulations 1-2 years back start of month 0 end of mon 6-12 forecast ensemble(s) model spin-up initial conditions climatology ensemble NCDC met. station obs. up to 2-4 months from current 2000-3000 stations in west LDAS/other real-time met. forcings for remaining spin-up ~300-400 stations in west climate forecast information data sources Forecast Products streamflow soil moisture runoff snowpack derived products e.g., reservoir system forecasts important point(s): after bias correcting and downscaling the climate model forecasts, the procedure for producing hydrologic forecasts is as follows: we spin up the hydrologic model to the start of the forecast using observed met. data (from 2 sources: NCDC cooperator stations through 3-4 months before the start of the forecasts, then LDAS 1/8 degree gridded forcings thereafter). The GSM forecasts comprise 2 sets of ensembles, one for climatology and one for the forecast. The climatology ensemble yields a distribution of the conditions we’ve seen over the period 1979-1999, while the forecast ensemble yields the distribution of the conditions we might see for the next 6 months. Although the climatology ensemble is nominally unbiased against a simulated climatology based on observed met. data (rather than bias-corrected, downscaled GSM met. forcings), we compare the forecast and GSM climatology so that any unforeseen biases (resulting, perhaps, from the downscaling method) occur in both climatology and forecast. Eventually this cautionary step may be eliminated, and we’ll compare directly to the simulated observed climatology. at the end of the spin-up period and one month before month 1 (out of 6) of the forecasts, we save the hydrologic model state. The state is then used for initializing the forecast runs. Through the first month, the model runs on observed data to the last date possible, then switches to the forecast data. Usually, we process the observed forcings up through the 15th to 25th of this initialization month, then the forecast forcing data carries the run forward for the remaining days in the month, and throughout the following 6 month forecast period. Note, the state files used for the climatology runs correspond to the spin-up associated with the particular year (out of 1979-1999) from which the climatology ensemble member is drawn. the spin-up period captures the antecedent land surface hydrologic conditions for the forecast period: in the Columbia basin, the primary field of interest is snow water equivalent. forecast products are spatial (distributed soil moisture, runoff, snowpack (swe), etc.), and spatial runoff + baseflow is routed to produce streamflow at specific points, the inflow nodes for a management model, perhaps. obs snow state information (eg, SNOTEL)

Initial Conditions: estimating run-up conditions Problem: met. data availability in 3 months prior to forecast has only a tenth of long term stations used to calibrate model Solution: use interpolated monthly index station precip percentiles and temperature anomalies to extract values from higher quality retrospective forcing data -- then disaggregate using daily index station signal. sparse station network in real-time dense station network for model calibration

Initial Conditions: snow state assimilation Problem sparse station spin-up period incurs some systematic errors, but snow state estimation is critical Solution use SWE anomaly observations (from the 600+ station USDA/NRCS SNOTEL network and a dozen ASP stations in BC, Canada) to adjust snow state at the forecast start date

Initial Conditions: Initial snow state assimilation Assimilation Method weight station OBS’ influence over VIC cell based on distance and elevation difference number of stations influencing a given cell depends on specified influence distances spatial weighting function elevation weighting function SNOTEL/ASP VIC cell distances “fit”: OBS weighting increased throughout season OBS anomalies applied to VIC long term means, combined with VIC-simulated SWE adjustment specific to each VIC snow band important point(s): the approach attempts to make use of forecast skill from 2 sources: better understanding of synoptic scale teleconnections and the effects of persistence in SSTs on regional climate, as reproduced in coupled ocean-atmosphere models; the macroscale hydrologic model yields an improved ability to model the persistence in hydrologic states at the regional scale (more compatible with climate model scales than prior hydrologic modeling). Climate forecasts with monthly and seasonal horizons are now operationally available, and if they can be translated to streamflow, then they may be useful for water management.

Initial Conditions: Snow state assimilation SWE state differences due to assimilation of SNOTEL/ASP observations, Feb. 25, 2004

Initial Conditions: final product Snow Water Equivalent (SWE) and Soil Moisture

Comparison with RFC regression forecast for Columbia River at the Dalles UW forecasts made on 25th of each month RFC forecasts made several times monthly: 1st, mid-month, late (UW ESP unconditional forecasts shown) UW RFC

Applications to drought monitoring DM is widely used, but link to direct observations (e.g., of soil moisture) is weak – hence reliance on indirect methods, such as PDSI. Need for reproducible basis for identifying drought-affected regions. Land surface model representations of soil moisture (and runoff) offer an alternative means for estimating severity, frequency, duration, and variability of current droughts, and linking them to the climatology of observed droughts.

The challenge: Different land schemes have different soil moisture dynamics

Models VIC: Variable Infiltration Capacity Model (Liang et al. 1994) CLM3.5: Community Land Model version 3.5 (Oleson et al. 2007) NOAH LSM: NCEP, OSU, Air Force, Hydrol. research lab (Mitchell et al. 1994, Chen and Mitchell 1996) Catchment LSM: NASA NSIPP LSM (Koster et al. 2000; Ducharne et al. 2000) VIC-CLM: Hybrid of VIC and CLM 3.0 Grid-based Sacramento: Variation of NWS operational (Burnash et al, 1973; Koren et al, 2004

Averaged soil moisture percentiles 1932-38

Spatial distribution of average (monthly) between-model correlations (of soil moisture percentiles)

e-folding time of soil moisture autocorrelation (months)

Multimodel results from UW real-time surface water monitor, 11/07 – 1/08

UW SWM for 5/21/08 (updated 5/20/08)

Drought recovery – the concept Real-time applications! Drought recovery probability described by soil moisture percentiles: (a) Current drought area (based on August 1933); and for different lead times, maps showing the probability (in each grid cell experiencing drought) that soil moisture percentiles will recover. (b) The grid cell-specific recovery probabilities are derived from real-time soil moisture simulations up to the current date, after which simulations are driven by ensemble climate forecasts based on a variety of sources -- e.g., ESP, climate index-conditioned ESP, and the CPC seasonal climate outlooks

Some obstacles and opportunities in hydrological application of climate information The “one model” problem (multi-model ensemble is the solution; need to deal with weighting of ensembles) Calibration at the basin scale (need to consider regional parameter estimation approaches; post-processing as an alternative to calibration) The value of visualization Opportunities to utilize non-traditional data (e.g. remote sensing) So why does the U.S. not have a national hydrologic prediction strategy?