Evapotranspiration Estimates over Canada based on Observed, GR2 and NARR forcings Korolevich, V., Fernandes, R., Wang, S., Simic, A., Gong, F. Natural Resources Canada, Earth Sciences Sector, 588 Booth St, Ottawa, ON K1A 0Y7 Canada ABSTRACT Evapotranspiration (ET) represents a major loss of water from the Canadian land surface. Model based estimates of ET are required for water resources assessments, especially in remote ungauged basins and for climate or land use impact studies. Validated 1km resolution maps produced through surface and remote sensing survey are now available to parameterize process based land surface models to estimate ET for current conditions. However, in situ observations of required climate observations (forcings) are poorly distributed in space; especially for models requiring long ter, sub-daily data. New forcing data sets are now available from both statistical and model based reanalysis. Objectives: 1.How do gridded observed forcings (CRU TS 2.0, Mitchell et al. 2003), GR2 Reanalysis and NARR compared to forcings observed at climate stations? (forcing anomalies) 2.How does ET differ between model result based on observed versus reanalysis forcings? (ET anomalies) 3.What are the contributions to forcing anomalies to modelled ET anomalies as a function of type of forcing data and region in Canada? APPROACH Validated numerical 1-D land surface model of coupled carbon, water, energy and nutrient cycles (EALCO; Wang et al., 2002) driven by hourly forcings and observed land parameters. Assess differences in long term annual averages of both point and gridded forcings and modelled ET based on these forcings. Numerically compute partial contributions of errors in forcings to errors in ET at each observation point and aggregate over long term runs. CWEEDS DATA GAPS (%hours missed) Mean Annual Precipitation ( ) at CWEEDS stations [mm/year] Canadian Weather Energy & Engineering Data Sets (CWEEDS) Reanalysis Model Forcings Anomalies CONCLUSIONS Leaf Area Index Surface parameters Mean Annual ET ( ) at CWEEDS stations [mm/year] : Mean Annual anomalies for NARR-driven ET related to CWEEDS Climate Research Unit Gridded Observations Hourly Rain, Snow, Insolation, Air Temp, Specific Humidity, Surface Pressure for _______________________ 138 continuous long term station records based on minimal gap filling. Hourly precipitation estimated from daily totals and hourly rain and snow flag. Monthly 0.5 degree gridded fields. Hourly time series synthesized by scaling each grid cell with nearest CWEEDS station using ratio of monthly totals 10-Day 1km resolution LAI from SPOT_VEGTATION satellite. Observed Forcings RESULTS NCEP NARR, Precip., 85 March 10, 06:00 NCEP/DOE GR2, 2m Temp., 85 March 10, 06:00 NCEP/DOE GR2, Precip., 85 March 10, 06:00 NCEP NARR, 2m Temp., 85 March 10, 06:00 1km resolution 53 classes from SPOT_VEGTATION satellite. Land Cover 1:1,000,000 scale maps from Canadian Soil Information System (texture, carbon, water capacity) % water cover for each 1km cell From Canadian National Topographic Database 1:50,000 maps Soil InformationWater Cover Typical annual value applied at each point climate stations corresponding to modal land cover class in 50 km radius of station. LAI for July Average Annual Anomalies in Forcings Relative to CWEEDS Co-incident temperature and pressure anomalies suggest they are due to differences between grid cell elevation and CWEEDS station elevation. GR2 and NARR show substantial downwelling radiation anomaly (as in Liu et al for NCEP/NCAR GR1 release also) Precipitation anomaly high over Rocky Mountains and around Eastern Canada. GR2 and NARR anomalies anti- correlated in Eastern Canada. Partial Contribution of Forcing Anomalies to ET Anomaly Modelled ET Anomalies NCEP NARR (North American Regional Reanalysis, Mesinger et al. 2004) Data from Gridded at 30km resolution, linear interploation To hourly time step. NCEP/DOE Global Reanalysis (G2, Kanamitsu et al. 2002) Data from Gridded at 1.85 degree resolution, linear interploation To hourly time step. Larger anomalies observed near Rocky Mountains and East Coast. NARR AND GR2 driven ET anomalies anti-correlated. Anomalies relative to total annual ET are large in the Arctic and may be of significance for modeling fluxes in Arctic basin. Humidity and temperature anomalies contribute most to ET anomaly on west coast due to excess water availability and cloudy conditions. Precipitation errors contribute most to ET anomaly in Prairies due to water limitation. GR2 shows greater contribution of radiation anomalies to ET anomalies in East Coast likely due to averaging of cloudiness. Great Lakes region shows no dominant contribution to ET anomalies although precipitation is generally of low impact. Large radiation anomaly in input forcings does not always result in dominant contribution to ET anomalies. Possibly due to lack of coupling of model with forcings (e.g. radiation errors may be causing sensible heat or long wave emission anomalies as well). Station observations together with satellite and ground survey data offer a valuable data set for climate model assessments. GR2 and NARR have similar biases to CWEEDS. Downwelling radiation bias is substantial and needs more investigation as to timing and causes. GR2 and NARR provide reasonable ET estimates compared to CWEEDS based ET except over Rockies. The lack of a “PRISM” type precip downscaling in NARR over Canada may explain why NARR does not provide substantially lower ET anomalies in comparison to GR2. Gridded fields based on observations tended to outperform reanalysis forcings although this may not be the case in remote areas where gridded fields are very uncertain. Local calibration of either reanalysis forcings or reanalysis driven ET may be desirable if possible [cm/month] Station longitude (deg E) GR2 NARR