This is an experimental tool for calculating future water levels for Lake Lanier based on the current water level height and the daily fall rate in feet. The dropoff rate is calculated from the slope change of a line based on the lake fill data and adjusted by seasonal average inflow percentages. See the paragraphs below the chart for further explanation of how the slope data was derived and what the values for Seasonal Variation mean.
Disclaimer: To quote the stock price mantra: Past performance is no indication of future performance :-) The results returned by this are an educated guess at best and are very likely to be wrong. Many factors cannot be accounted for by this model that effect daily water levels. The results returned by this should be for entertainment value only.
To use this tool, enter the current level and the drop (in feet (i.e. .20)
and click on "Calculate"
The chart is then adjusted by the rainfall recorded in Atlanta. The position
of the rain gauge is relatively far from the lake (but the closest available).
The data is at:
http://climate.engr.uga.edu/state/ga_precip.html
To help smooth out localized showers I first totaled the rainfall during the entire fill period, then calculated the daily rain average. I then adjusted the fill days on a per-month basis by how any given month compared to the expected average. For example, if it rained 20% more in a month, I made that month take 20% less days. This is better than trying to adjust the elevation since the elevations are relevant to the shape of the lake.
The results of this were then run through a smoothing spline function using "R". This should approximate the overall change in rate.
The reason for running the spline curve rather than the rain adjusted daily water levels directly is due to the original data containing two regions where the curve becomes concave instead of convex. For this to hold true with constant water input, the lake would have to be smaller at that level than below that level (i.e. a flask shape) which we know is not true. Without knowing what the actual release volumes were at the time it cannot be directly accounted for, however a smooth spline curve will average the slope change as the width increases for an overall more accurate slope change. There could be some localized regions where the change in slope is actually more pronounced than the resulting spline curve reflects, however these errors should average to the final cumulative slope over time.
The Seasonal Variation field is an attempt to adjust the daily lake levels based based on recorded seasonal inflow fluctuations. For example if we are seeing a .01 feet drop per day in January, that would not equate to a .01 feet drop per day in October. The inflows data shows a very solid trend how average inflows vary on a month by month basis. The Seasonal Variation value is the average daily difference in lake levels from the peak of the rainy season (end of January/early February) and the driest months (September/October). The default value (.112) is from the past year. This is then adjusted by a table of median values built off recorded inflow values averaged over multiple years. The adjustment is done by adding the difference between the projected day minus the starting day. You can see the shape of this adjustment table by setting the Current Level to 1020, a Fall Rate of 0 and the Seasonal Variation to 1.0 The start of the graph is the current day. You can disable this added calculation with a value of 0.
The above chart was made much easier due to the use of the wz_jsgraphics.js package from Walter Zorn and the NumberFormat javacript package from Keith Jenci
Lawrence Freil
lef@freil.com