NOAA Paleoclimatology Program, NCDC Paleoclimatology Branch

CA Streamflow Home - Background - Chronologies - Reconstructions - Case Study - Resources

Feather River inflow to Lake Oroville, CA

Principal Investigator	David Meko, University of Arizona - dmeko@ltrr.arizona.edu
Gage record source	California DWR
Calibration/validation	1906-1977
Full reconstruction	901-1977
Predictor chronologies	see note below
Explained variance	61% (901-1299); 67% (1300-1607); 79% (1608-1977)

The full Feather River reconstruction described here consists of three sub-period reconstruction models (901-1299; 1300-1607; 1608-1977), with each reconstruction calibrated against the gage record (1906-1977) using a different but overlapping set of tree-ring data. This was done because the number of tree-ring chronologies declines going back in time, and so each sub-period reconstruction takes advantage of all the tree-ring data available for that period. A total of 7 chronologies were used to develop the 901-1299 model, 13 chronologies (including those 7) were used to develop the 1300-1607 model, and 26 chronologies (including those 9 of those 13) were used to develop the 1608-1977 model.

Gaged and reconstructed flow for Feather River

Gaged (blue) and reconstructed (green) records over the calibration/validation period (1906-1977). The 1608-1977 sub-period reconstruction is shown; the other two sub-period reconstructions have slightly worse fits to the gage record, as indicated by the validation statistics.

Reconstructed flow for Feather River

Full reconstruction (901-1977), with annual values (green) and a 5-year running mean (black).

Sample depth for Feather River reconstruction

Total sample depth (typically, two measured series per tree) for the reconstruction. The steps at 1300 and 1608 reflect the use of additional chronologies in the reconstruction models that start in those years.

Validation Statistics

(See Blue River Case Study for detailed explanation)

901-1299 Sub-Period Reconstruction:

	R_cal	R_val
calibration model	0.78
PRESS validation		0.73

	R²_cal	RE_(val)
calibration model	0.61
PRESS validation		0.54

	standard error of estimate	RMSE
calibration model	0.137*
PRESS validation		0.145*

1300-1607 Sub-period Reconstruction

	R_cal	R_val
calibration model	0.82
PRESS validation		0.76

	R²_cal	RE_(val)
calibration model	0.67
PRESS validation		0.58

	standard error of estimate	RMSE
calibration model	0.128*
PRESS validation		0.137*

1608-1977 Sub-period Reconstruction

	R_cal	R_val
calibration model	0.89
PRESS validation		0.84

	R²_cal	RE_(val)
calibration model	0.79
PRESS validation		0.70

	standard error of estimate	RMSE
calibration model	0.103*
PRESS validation		0.116*

* The units for RMSE are log-10 million acre-feet, because of log-transformation of the gage record prior to calibration of the reconstruction model.


http://www.ncdc.noaa.gov/paleo/streamflow/ca/feather.html Downloaded Friday, 29-Aug-2008 15:02:09 EDT Last Updated Monday, 04-Dec-2006 13:30:30 EST by paleo@noaa.gov Please see the Paleoclimatology Contact Page or the NCDC Contact Page if you have questions or comments.

Feather River inflow to Lake Oroville, CA

Download data

Calibration/validation

Full Reconstruction

Sample Depth

Validation statistics

Gaged (blue) and reconstructed (green) records over the calibration/validation period (1906-1977). The 1608-1977 sub-period reconstruction is shown; the other two sub-period reconstructions have slightly worse fits to the gage record, as indicated by the validation statistics.

Full reconstruction (901-1977), with annual values (green) and a 5-year running mean (black).

Total sample depth (typically, two measured series per tree) for the reconstruction. The steps at 1300 and 1608 reflect the use of additional chronologies in the reconstruction models that start in those years.