Ocean Surface Temperature and Pelagic pH Example from 1968

I’ve been recently sampling within the year of 1968 from NOAA’s WOD.  A past sampling is featured at an earlier post  In this example figure set, it can be seen that although they share a similar sampling location pattern over the selected calendar year, more measurements were published in the WOD for pelagic ocean pH (15,713 points after elimination of obvious outliers and redundant entries), than for dry bulb temperature at the surface (10,880 points).  In producing these first two contours, I used the same limits and other settings, as well as the same kriging approach as in the earlier post.

oceanPH1968oceanDBT1968

The following figures are additonal examples of gridding and contouring of the same 1968 pH and ocean surface temperature  data sets.  In this case, I utilized the default SURFER settings within the method of minimum curvature for gridding of the raw data.  This method and the associated settings, including those regarding data value ranges, in these examples, succeed roughly at avoiding data-poor regions.

oceanPH1968_MinCurvMethodMWA

oceanSurfaceTemp1968_MinCurvMethodMWA

I continue to evaluate geostatistical parameters associated with each contour plot featured in this series of posts.  One example includes excerpts from a variogram analysis for the 1968 GOPpH data set.

GOPpH1968MWAvariogram

Univariate Statistics for Z column (GOPpH1968)
Minimum: 6.56
25%-tile: 8.04
Median: 8.18
75%-tile: 8.27
Maximum: 8.91
Midrange: 7.73
Range: 2.35
Interquartile Range: 0.22
Median Abs. Deviation: 0.11
Mean: 8.12
Trim Mean (10%): 8.14
Standard Deviation: 0.25
Variance: 0.06
Coef. of Variation: 0.03
Coef. of Skewness: -1.68

It is encouraging to see that the reported mean value for 1968 of 8.12 aligns well with a previous plot.  Yet, as always for these contour plots, please keep in mind the impacts of the boundaries, locations and depths of data, seasons of data collection, coordinate transformations, and numerous other factors, upon the contouring results.   In addition, the continental sketches are only approximate.  Accordingly, it is premature to reliably interpret any image shown here, other than gross characteristics perhaps.

 

Global Ocean pH and the Rio Grande in the US SW

In past posts of 2014, I’ve compared time series of Global Ocean Pelagic pH (GOPpH) to the Pacific Decadal Oscillation (PDO), and I’ve compared time series of the PDO to the Otowi streamflow gage of the Upper Rio Grande in the US Southwest.

Here in the first figure, as a simple exercise, the GOPpH time series is plotted directly against the Otowi time series.

GOPpHvsOtowiGage10ytaThe second figure shows a possibly remarkable result, that the trimmed* 10 yta  GOPpH time series exhibits a higher correlation to the10 yta Otowi record than any other parameter studied in this set, except for the PDO.  In other words, the correlations of ENSO, the AMO, and a GHG index to the Otowi are all lower than the correlation of the GOPpH to the Otowi (all for 10 yta condition).

GOPpHvsOthersVsOtowiCorrels

 

*Unless otherwise noted, pairs of series are trimmed to the limiting full data set sufficient to represent the  spans of interest.

MW&A First Hydroclimatologic Forecast a Qualified Success.

In early May 2014, MW&A published a prototype forecast of a ten-year-trailing-average (tyta) flow range applied specifically to the end of the calendar year for the Otowi gage of the Rio Grande in Northern New Mexico.   Given typical winter patterns for this gage, it is not too early to propose that the forecast was a success.

December2014MW&AforecastEval_a

A final evaluation of the forecast will be published in early 2015, when the USGS updates its monthly statistics for that gage.  However, their preliminary daily values can be used, along with the historical seasonal patterns that are well known, to give confidence to a successful score for this first published forecast.

The first figure is an update of the original MW&A forecast graphic that was published on May 4.  The final Otowi gage tyta flow value for calendar year 2014 will fall within the central region of the MW&A forecast range.  Moreover, the relative difference between the MW&A forecast mean value and the anticipated flow value will likely be less than 3% by the final accounting.  Arguably, we hit the bull’s eye, although in this graphical example, the edge was grazed.

Additional complementary forecast products were made on two occasions this year by MW&A.  One product utilized our stochastic landscape calculations in June 2014 to estimate net flow past the Otowi gage for the calendar year.  That estimate of ~700,000 af (acre feet) is expected to be within 10% of the final value.

The last product was published in September for subscribers only, and covered additional streams, including sections along the Navajo, Pecos, and Gila rivers and other Rio Grande tributaries.

We will follow up soon with a new forecast for the tyta value at Otowi gage for the end of 2015.  One signature difference between our method and that of conventional approaches is that our process does not typically require spatial estimation of snowpack components for the year in which the forecast applies.  Therefore, typically there is no need to wait until the Spring for our basic forecast to be published.

MW&A is pleased to present these positive results for our very first published forecast.   With the support of an informed clientele, along with our new collaboration with LANL atmospheric scientist Dr. Petr Chylek, our ongoing continual process improvement is anticipated to enhance and accelerate our forecasting performance record for 2015 and beyond.

Hydroclimatology and Hydrogeology