Modeling Results.

This appeal is based on legal deficiencies in the FEIS and ROD, but underlying each of those legal deficiencies is a serious error or omission or arbitrary decision regarding an issue of substance. The seriousness of the errors and omissions, and the arbitrary nature of key intermediate decisions and the overall final decision, are well characterized by the tables and graphs that summarize the analytical modeling done for the ID Team and available to the decision maker. Some of these results were published as tables and graphs in the FEIS, and others are only in the planning record, not readily available to the public. Thirty pages of these tables and graphs from the planning record are attached as Appeal Appendix B. Some of these graphs were also published in the FEIS.

In the following discussions we will refer to pages in Appeal Appendix B by numbers we added to the first 13 pages for convenient reference. Please note that the lines for one or more alternatives (Alt 2 and perhaps others) may not be visible because the color of that line in the original was too light to show in a black and white printout. If a line doesn't show up on the graph, please refer to the tabulated values above the graph for comparisons.

In these comparisons, we will ignore the entries for "MLV" (no management) and "Alt 1" (no-action alternative). These comparisons are in terms of the preferred alternative 8-mod versus the other action alternatives 2 through 8.

All Wildfire. (Appeal Appendix B, pg 1) Right from the beginning, the action alternatives diverge and their relative ranking persists for the entire analysis period, with three worse than 8-mod and four better than 8-mod. The difference between 8-mod and the group of four better alternatives settles down to about 10 to 15 thousand acres per year.

Lethal Wildfire. (pg 2) Again there is a quick divergence and continuing difference. Alts 2 and 5 are worse than 8-mod, Alt 8 tracks right with 8-mod, and the four better alternatives average five to ten thousand fewer acres per year lethally burned through the analysis period. (As we note in an earlier section of this appeal, this graph is for forest lands only, and is thus very significantly different from the similar-looking graph in the FEIS that typically added 20 thousand acres per year of brushfields to the lethally burned forest lands.)

Owl Nesting Habitat. (pg 3) Differences show up in the second decade and grow wider over time. By the fourth decade every other action alternative does better than 8-mod, and on average the gap continues to widen for the rest of the analysis period.

Old Growth. (pgs 4 and 5) These are somewhat different measurements of the same thing, whether called "late seral stage" (pg 4) or "LSOG ranks 4 and 5" (high ranking Late Seral Old Growth, pg 5). By the second decade every other action alternative does better than 8-mod, and on average the gap continues to widen.

Since spotted owl nesting habitat and old growth forest are given over-riding priority in this FEIS and ROD, it is startling and disturbing, not to mention unlawful, to find Alternative 8-mod right at the bottom of these rankings.

Large Trees. (pg 6) Differences start to appear in the second decade, then persist and increase over time. Alts 2, 5, and 8 stay close to 8-mod, and the other four action alternatives produce about 19 percent more large trees by the end of the analysis period.

Very Large Trees. (pg 7) Since it takes a longer time to affect the numbers of very large trees (greater than 50 inches diameter), differences don't start to show until about the fifth decade. After that, Alts 2, 5, and 8 stay close to 8-mod, and the other four produce about 15 percent more very large trees by the end of the analysis period.

Large Hardwoods. (pg 8) Differences begin to show in the second decade, when two alternatives are worse than 8-mod and the rest somewhat better. Later in the analysis period the familiar pattern emerges, where only 2, 5, and 8 are with or worse than 8-mod, and the others become 5 to 10 percent better.

These long term results of the analysis are sometimes discounted on grounds that accuracy can't be assured for such long term projections. While due caution is required when interpreting any projection, it must also be recognized that (1) differences in the projected values are more reliable than the values themselves, and (2) the only way you can get increased old growth forest and large trees is to start with the correct methods and stick with them for a long time. You can't go fifty years on the wrong path, then expect magically to be on the right path toward different and better hundred-year results.

Large Snags. (greater than 15 inches diameter, pg 9) In the second decade four alternatives start to drop below 8-mod while 2, 5, and 8 are equal or better, establishing a pattern that persists through the analysis period. The differences peak at about 75 years, then the worst group tends to level off and the best group drop back a bit. Note that by the fourth decade every alternative has five or more large snags per acre. When is enough enough?

The turnaround of 2, 5, 8, and 8-mod from worst on owl habitat and other old forest characteristics to best on large snags needs examination. What accounts for the switch? Perhaps it's because the 8-mod group is better at making large snags by killing large old green trees in lethal wildfires.

Crown Cover. (pg 10) There is no significant difference among alternatives right across the whole analysis period. Every alternative tracks very close to 50 percent average crown cover across the whole forest.

Potential Biomass Tonnage. This is a measure of how much usable biomass other than saw timber would be available for removal during fuel reduction and silvicultural treatments. It varies widely, peaking in the early decades, when the initial fuel reduction treatments are done, and thereafter falling back to more or less level amounts with consistent differences. Alts 2, 5, and 8 are again with or below 8-mod, while the other four alternatives are usually well above 8-mod.

Note the sawtooth pattern of some outputs on pages 10 to 12. These are explained in a document in the planning record as artifacts of the timing of treatments that happened to occur in the modeling. That is, outputs near the borderline year from one decade to the next sometimes fell in one decade, sometimes in another, so the output by decade jumped up and down quite a bit for some alternatives.

Timber Harvest. (pg 12) The highest value for any alternative is in the first decade when fuel reductions would tend to peak, and thereafter production falls back to relatively stable outputs and rankings. Alts 2, 5, and 8 tend to remain below 8-mod, Alt 3 stays pretty much with 8-mod, Alt 6 yields about twice 8-mod, Alt 7 yields 5 to 6 times as much as 8-mod, and Alt 4 yields more than 7 times as much as 8-mod.

Net Public Benefit. (pg 13) The only alternatives than manage to average worse than 8-mod are Alts 3 and 8. Otherwise every other alternative produces more net public benefit from the treatments. Alternative 4 averages one billion dollars more per decade, 100 million dollars per year greater net benefit. Alt 7 averages 600 million dollars per decade more than 8-mod.

By any reasonable interpretation of these graphs, or any combination of other graphs in the FEIS and Appeal Appendix B, the decision to adopt Alternative 8-mod cannot be said to rest logically on the outcome of such modeling and analysis as was done and presented to the Regional Forester by the ID Team.

These potentials to produce biomass and timber harvest during fuel reduction and forest health treatments, and the net revenue benefit gained from conducting treatments under those alternatives, are necessary considerations when we are trying to figure out how to get the work done and pay for it within the budget likely to be available. Whether your motive is to protect the forests and adjacent human population from lethal wildfire, to preserve and increase old growth forest, or just to gain the maximum economic benefit with ecologically sound forest management, the obvious advantage is in the group of alternatives that consistently contains at least Alternatives 4 and 7, perhaps others, but does not contain Alternative 8-mod.

But on the other hand, if commercial removal of the material is a huge embarrassment while you are kow-towing to the U.S. Fish and Wildlife Service or environmentalist lobbies, then 8-mod looks just great.

We leave it to the reader to judge which motives were strongest when Alternative 8-mod was chosen.

It might be argued that the results of modeling are not valid because the analysis period is so long compared to the expected life of this particular management plan, and long term projections are not very reliable anyway. These are reasonable points that deserve consideration.

It is true that some of the measurements of projected outcome show no significant difference during the expected lifetime of the plan amendments. Long term projections of ecosystem response regarding old forest spotted owl habitat and large trees show significant differences, but nothing of statistical significance shows up in much less than 20 years for most measures, so it might seem there is no legitimate basis in the analysis for a choice among the alternatives.

First, we think long term differences in modeling outcomes are significant when making even short term choices among management actions. If you don't make the correct choice of roads at the first intersection, you're unlikely to get to your destination.

Second, modeling is more accurate in finding the difference among long term results than in finding the absolute results themselves. All we need for a valid choice is to judge the differences among alternatives, not their absolute outcomes to the third decimal.

Third, these modeling methods have been in use for quite a time already in Forest Service management, and the ones in use now are better than when first introduced. There is no justification for the Forest Service to make a major management decision so directly and completely contrary to these modeling outcomes. Not liking these particular outcomes is no excuse for abandoning or selectively discrediting the models that produced them.

If you consider the modeling and analysis results for the issues said to be of greatest significance (i.e. fire protection, old forest, and owl habitat), then grade "on the curve," Alternative 8-mod can do no better than C minus, possibly D plus.

Why is Alt 8-mod so far from best among the alternatives considered? There are only two possible answers:

Either (1) The range of alternatives considered truly encompasses the full range that is possible, so Alt 8-mod really is the dull-normal result-challenged alternative indicated by the modeling results. In this case the choice of Alt 8-mod fails to be consistent with the best information available to the decision maker, which is one legal definition of "arbitrary and capricious."

Or (2) The range of alternatives is so narrow that it makes no difference which alternative is chosen, so why not alt 8-mod? First that would be a direct violation of 40 CFR 1502.14(a) "Rigorously explore and objectively evaluate all reasonable alternatives..." and second it would be a text-book example of an "arbitrary and capricious" choice, with or without a legalistic interpretation.

The Decision is revealed to be arbitrary and capricious at its root and along all its branches

Failure to consider an alternative consistent with the HFQLG Pilot Project.

In a particularly egregious violation of both the NFMA and NEPA requirements for inclusion and evaluation of all reasonable alternatives, the DEIS and ROD treated the HFQLG Pilot Project as an unwelcome intruder that had to be handled as an exception, not as an opportunity to consider longer term implementation of an obviously reasonable alternative. The ID Team's decision not to include such an alternative is particularly troubling, since the HFQLG Pilot Project Final EIS had been completed not long before the SNFPA Draft EIS was developed and issued, so it would have been relatively easy and very efficient to base a QLG-like SNFPA alternative on an extension of the HFQLG concepts and analyses.