The Main Vernal Pool on the Mesa de Colorado at the Santa Rosa Plateau begins to fill with water only after certain rainfall requirements are met. I present here a determination of those rainfall requirements from an analysis of rainfall data from the last four years.
The display board at the Pool says that it takes 8" of rain to fill the pool, but that is only a rough rule-of-thumb. In fact, the amount of year-to-date total rainfall does not by itself determine whether a pool forms. As any gardener knows, the moisture content of soil at any given time depends little on the rainfall prior to two months ago. Thus whether a pool forms depends on the exact rainfall history in approximately the last two months, as well as the weather during this period.
I have collected rainfall data in Fallbrook, as well as the approximate date on which the Pool first contains water, beginning in late 1996. The rainfall was measured at my house in Fallbrook, 11 miles almost directly south of the Pool, at an elevation of 690'. Although rainfall amounts can vary significantly over much shorter distances, total rainfall over several storms is highly correlated even between much greater distances. Thus the amount of rainfall at Fallbrook should be a good proxy for the amount of rainfall at the Pool, with the two rainfall amounts differing only by a constant multiplicative factor. For this analysis, the value of that factor does not matter.
Total rainfall is clearly not the key factor. It required somewhat less than 6.5" to form the pool in 1996-1997 and 1999-2000, but it took over 7.9" to form the pool in 1997-1998, and the pool never formed in 1998-1999 despite 9.1" of rain by 12 April 1999. In fact, although 6.5" of rain had fallen by 10 December in both 1996 and 1997, the pool formed in 1996 at that point but another 1.4" of rain was needed in 1997 to form the pool. The difference was that in 1996 more rain had fallen within the last month at that time.
See Pool Presence or Absence vs. Total Rainfall To Date for the detailed rain history for each year. Filled blue circles represent the first day the pool was observed to be present and empty blue circles represent observations of "no pool". To resolve the two ambiguities about which filled blue circle goes with which year, note that in each case one of the intersecting lines has an empty blue circle later in the year. Hence the filled blue circle belongs with the other year.
I parameterize the rainfall data in terms of the amount of rainfall in the last month and the amount of rainfall in the last two months, for the following reasons. Soil moisture is the key determinant as to when the Pool fills, since rainfall initially wets the soil and causes the clay bottom of the Pool to seal. As mentioned above, rainfall occurring over two months ago is largely irrelevant to current soil moisture, especially at the beginning of the rainy season when significant periods of warm weather evaporate that rainfall. Thus one key variable is roughly the amount of rainfall in the last two months. Because some of that rainfall from two months ago will have evaporated, another measure of more recent rainfall is needed. I chose simply and arbitrarily to halve the interval of two months for this analysis to provide an independent measure of recent rainfall. (I could have chosen a different interval, such as a two week interval, as the second independent measure of recent rainfall, but as will be seen, the interval of one month produced good results. Hence there was no need to explore the use of other intervals.)
An example should make clear how these two measures are calculated. The following table gives the rainfall (in inches) in 1996-1997:
| Date | Rainfall on that Date | Cumulative Rainfall | Rainfall in last month | Rainfall in last two months | Comments |
|---|---|---|---|---|---|
| 10/30/96 | 0.94 | 0.94 | 0.94 | 0.94 | first rainfall |
| 11/21/96 | 1.11 | 2.05 | 2.05 | 2.05 | |
| 11/22/96 | 2.87 | 4.92 | 4.92 | 4.92 | |
| 11/23/96 | 0.32 | 5.24 | 5.24 | 5.24 | |
| 11/24/96 | 0.00 | 5.24 | 5.24 | 5.24 | last observation with no pool |
| 11/30/96 | 0.00 | 5.24 | 4.30 | 5.24 | the one month total no longer includes 10/30/96 |
| 12/06/96 | 0.31 | 5.55 | 4.61 | 5.55 | |
| 12/09/96 | 0.16 | 5.71 | 4.77 | 5.71 | |
| 12/10/96 | 0.83 | 6.54 | 5.60 | 6.54 | pool established |
Note that both the "rainfall in the last month", and the "rainfall in the last two months", are identical to the cumulative rainfall until 11/30/96. On that date, the rainfall on 10/30/96 drops out of the "rainfall in the last month" since the rainfall on 10/30/96 occurred a month and a day earlier. Hence the "rainfall in the last month" dropped from 5.24" to 4.30" on that date. It increased to 4.61" on 12/6/96, since the 0.31" on that day was added to the "last month total", with no rainfall lost in that total since no rain fell from 10/30/96 to 11/6/96.
The "rainfall in the last two months" will continue to equal the total rainfall until 12/30/96, at which point the 10/30/96 rainfall will drop out of the two month total.
The figure below summarizes the rainfall data observations for all four years between the date of my last observation of no water in the Pool and the date of my first observation that the Pool contained water in each year. This plot shows clearly the amount of rainfall needed to fill the Pool, but it is a bit complicated, so read the explanatory text below the plot to interpret the plot.
Consider first the red line at the middle right of the plot, which represents the 1996-1997 data presented above. This line begins at (5.24,5.24), which has a blue unfilled circle around it, moves left to (4.30, 5.24) and then ends at (5.60, 6.54), a blue filled circle.
- These points give first the one month rainfall, followed by the two month rainfall. Thus at the time of the first point, 11/24/96, all the rainfall, 5.24", fell in the last month, so both the one month and two month rainfall amounts equal 5.24". This point has a blue unfilled circle around it, since I observed that there was no pool on that day. (Not all points from all years are plotted, just the points near the transition from "no pool" to "pool with water".)
- The next point, (4.30, 5.24), occurred at 11/30/96, when the one month rainfall no longer included rainfall on 10/30/96, and hence the one month rainfall dropped from 5.24" to 4.30". The two month rainfall of course stayed the same.
- The last point before the pool was observed to be filled, (4.77, 5.71) occurred on 12/9/96. Rainfall on 12/10/96 made the rainfall totals (5.60, 6.54), and the Pool had water in it after that rainfall.
- The first point has a light blue empty circle around the point, representing an observation of an empty Pool. The last point has a filled blue circle, representing an observation of a filled Pool. The other points do not have any circles around them, since I did not observe the Pool on those days.
The other two lines ending in filled blue circles similarly represent the 1997-1998 data (zigzag aqua line ending in the filled blue circle at the middle top of the plot) and the 1999-2000 data (the nearly straight pink line ending in the lowest filled blue circle at middle right).
The 1998-1999 data, the "year without a Pool", is represented by the four red "rectangles with a cross inside" connected by red lines, with each point surrounded by a light blue empty circle. Although much more data exists, I have plotted only the largest one and two month rainfall totals from this season.
The long black line just below the diagonal of the plot represents the limit where the rainfall in one month equals the rainfall in two months; i.e. all the rainfall occurred in the last month.
Note the well-defined region of the plot occupied by the filled blue circles, as well as the narrow region that separates them from the empty blue circles. The green straight line beginning at upper left in the plot is the simplest line that defines that narrow region. It therefore represents the simplest model for the amount of rainfall needed in both a one and two month period for the Pool to begin to fill.
It is actually somewhat surprising that the transition between "no Pool" and "Pool with water" is so well defined. There are many more variables for this problem than just the one and two month rainfall totals. In particular, the distribution of rainfall within the two months, the precise dates on which the rainfall occurs, and the weather during the two months all must play a role in determining how much of the rainfall in the last two months survives in the ground moisture. Weather factors include the deviation from normal temperatures, normal wind speeds, and normal humidities. Hence it would not be unexpected if future observations show a wider transition region.
Other, more complicated models can of course fit these limited number of observations as well. In particular, note that it is possible that the Pool could have begun to fill at any on the points between the filled and empty circles. Note, however, that in both 1996-1997 and 1997-1998, there is only one other possible point, the one immediately prior to the filled blue circle point, that could plausibly represent a time when the Pool began to fill. Unfortunately, in 1999-2000, there were many closely-spaced rainy days between my observations of "Pool empty" and "Pool filled", so that year by itself doesn't constrain such models very well.
However, all such models must give roughly the same numbers for the one and two month rainfall amounts, due to the small distance between the filled and empty circles. Those numbers are given in this table:
| Rainfall (") Needed to Form the Pool In | |
|---|---|
| One Month | Two Months |
| 5.8 | 5.8 |
| 5.5 | 6.0 |
| 4.0 | 7.1 |
| 3.0 | 7.8 |
| 2.5 | 8.2 |
| 2.0 | 8.5 |
Again, it is unlikely that these numbers are accurate to the tenths place, due to all the other variables which must affect whether the Pool forms. The numbers probably are accurate to about an inch of rainfall.
Apparently 6-8" of rain, as measured in Fallbrook, is needed in a two month period, before the Pool can form. In addition, a well-defined amount of that rain has to fall in the last month of that period to begin to fill the Pool.
Given this tight correlation observed so far, one can predict at which point the Pool might form in the future. For example, the Fallbrook rainfall totals as of 1/13/00 are (2.7", 2.7"). It is likely that ~3.1" of rain is needed within the next month after this date for the Pool to begin to fill, and more than ~3.1" of rain if that amount does not fall within the next month.
Prior to the formation of the Pool each year, the prediction from this model for the estimated amount of rain needed to form the Pool will be given in Current Status of Flowering Plants and the Vernal Pool.