README:  SW Monitor Forecast Products

Forecast maps for of soil moisture and cumulative runoff fields are a new 
addition to the SW Monitor.  These are based on the climatological Ensemble 
Streamflow Prediction technique, in which historical weather traces (for 
same the calendar period of the forecast, but from past years) are used as 
meteorological forecast inputs for a hydrology model.  The forecasts are 
initialized using current hydrologic states (primarily soil moisture and 
snow water equivalent), hence the results merge persistence in the anomalies 
of these states with climatological uncertainty about future weather.  

The ESP traces are taken from the years 1960-1999, and the forecasts will be 
updated at least every week.  Currently, I run traces that are 6 months in length, 
but provide products only for the first 3 months, consistent with the lead time of 
the CPC Drought Outlook.

The current forecast products are shown for lead times of 1, 2 and 3 months.

- soil moisture percentiles (medians of the forecast ensemble) for the day falling 
  at each of these lead times (these are linked to 'lead 1m', 'lead 2m', 'lead3m' 
  in the soil moisture column)

- cumulative runoff percentiles (median of the forecast ensemble) for the accumulated 
  runoff during the 3 month period leading up to the day falling at these lead times. 
    (these are linked to '3mo lead 1m', '3mo lead 2m', '3mo lead3m' in the runoff column)

  (I also calculate the quartiles and the min and max of the forecast distributions for 
   these variables, and can provide them if there is interest.)

- the probability of percentiles of the soil moisture and accumulated runoff being in the 
  lowest quintile (i.e., lower than 20th percentile) of the historical range for the forecast 
  date (current date + lead time).  

An ENSO-determined subset of the ESP forecasts is also evaluated and plotted for comparison
with the climatological ESP.  This subset includes only the years falling into the current 
nominal ENSO category. The ENSO years are taken from the period 1950-2002 to enlarge the sample size.

Lastly, current conditions soil moisture plots based one two additional land surface models,
Noah and the grid-based Sacramento model, have recently been added.  These models share the 
forcings used to drive the VIC model, but were implemented by Ted Bohn under the supervision
of Dr. Dennis Lettenmaier.  Forecasts are not available for the multi-model ensemble at this time.

-A. Wood, 20071227