Eric Jones Senior Hydrologist Lower Mississippi River Forecast Center 89 th AMS Meeting
July Began development and implementation Used NWS OHD Research Distributed Hydrologic Model V2.0. SACSMA used for rainfall-runoff No snow model used Rating curve method of routing Initially used basins 3, expanded to 7 Calibrated before implemented Verification done on initial three basins Black Creek at Brooklyn, MS (BKN) Leaf River at Collins, MS (CLS) Emory River at Oakdale, TN (OAK)
BKN CLS OAK
++= Basin Model % bias Daily RMSE Corr Coef (CMSD) CLS Distrib Lumped BKN Distrib Lumped OAK Distrib Lumped
January October 2008 Simulated Lumped vs Distributed models Not adjusted to stage observations No runtime modifications included in lumped model Statistics on forecasts +24 hours Used quality controlled multi-sensor precipitation est. Mean average PE used for distributed model Mostly computed PE used for lumped model
BKN Distrib Lumped CLS Distrib Lumped OAK Distrib Lumped
BKN Distrib Lumped CLS Distrib Lumped OAK Distrib Lumped
BKN Dist Lump CLS OAK Dist Lump
One square = 4 km 2 Distrib Lumped Obs
Distrib Lumped Obs One square = 4 km 2
Distrib Lumped Obs One square = 4 km 2
Distrib Lumped Obs
Distrib Lumped Obs One square = 4 km 2
Distrib Lumped Obs
One square = 4 km 2 Distrib Lumped Obs
Distrib Lumped Obs
Distrib Lumped Obs
Distrib Lumped Obs
Unadjusted routing Adjusted routing Observed Forecast Flood stage Observed Forecast Flood stage
Distributed model vs lumped model Distributed model does better in some cases Shape of distributed hydrograph appears to be better Suggestions for hydrologic forecasters Should base forecasts using data from both models