Interesting suggestion, but it's not immediately obvious WHY meristem types would differ in sensitivity to temperature or other factors.  Extension growth is highly regulated and under strong selection pressures: for this reason a mechanism involving any simple response to abiotic environment factors seems to me unlikely.

Wind exposure has large and predictable effects on height-diameter allometry (e.g., Thomas et al. 2015 - see attached paper).  However, somewhat counter-intuitively, high altitude locations are NOT predictably more "windy", at least in terms of average wind speeds.  As put by C. Körner, "Contrary to popular belief, mountains, as a whole, are among the less windy places on Earth, owing to the topographic shelter between or within mountain ranges, except for exposed crests, summits and isolated mountains on islands" (Körner 2007). 

As for places to start, it's critical to note that altitudinal gradients involve a complex of factors including temperature, moisture regimes, and CO2 partial pressures, among others.  Both genetic and phenotypic responses are involved, where this has been examined in detail.

Article Körner C.. The use of ‘altitude’ in ecological research. Tre...

Article Genetic vs. phenotypic responses of trees to altitude

Article Tropical trees in a wind-exposed island ecosystem: Height-di...

Let me add to that debate a humble microclimatic rather than physiological explanation. Stem cambial tissue in high elevation trees  is likely to profit from solar stem warming (particularly at high latitude) or morning and evening sun, more so in open than closed stands. Stems although buffer thermal oscillations and may store some heat. In contrast tips of twigs i.e. their apical meristems are very closely coupled to atmospheric circulation, and thus are likely to be cooler. I am discussing this in Alpine Treelines (Springer 2013), and associated that with the taper issue. Trees get more stunted at high elevation, that is, their growth in height is more restricted than their growth in thickness. There may be other reasons related to stress and hormonal gradients, but in general the main stem is likely to be a more cosy place for cells to divide and differentiate than an exposed apical meristem. I refer you to page 80 and 97ff. 

Christians answer is interesting. While nobody knows the truth about it,  I would object that length growth of many treeline conifers is predetermined and that final length depends on the growth conditions of the previous year. I think that this would make a direct regulation by microclimatic conditions unlikely.

Thank you so much for all the answers. I read the papers Sean recommend and Alpine treelines (7.2.2 Apical growth dynamics) by Christians.

Beside this, I found one more paper help to understand this question, Antonucci (2015) compared the relationships between apical and lateral meristems (bud development and xylem phenology), they found that degree days sum (DDS) in the exposed shoot phase were higher than that in the first mature cell phase for both pine species(fir and spruce). Moreover, the mean thermal threshold was estimated at lower level for the first enlarging cell phase compared to open bud phase in spruce, and similar level for fir. 

So, for treeline trees, their apical and lateral meristems at least have the difference of the following: location determine they can benefit from shelter or not; and they did have but species-specific thermal threshold difference. 

Further question about co-existence of shrubby and normal trees of one particular species at treeline, I know that tree ring studies had some answers at the facet of climate condition (Devi et al., 2008; Hagedorn et al., 2014). But at the facet of  underlying process, if vulnerable apical bud endure frequent harm(cold or wind abrasion) be the reason why shrubby trees always appear and finally become the dominant form at upper treeline(Our study site at Changbai Moutain, China with treeline speices of birches)? 

Article Expanding forests and changing growth forms of Siberian larc...

Article Treeline advances along the Urals mountain range – driven by...

Article Synchronisms and correlations of spring phenology between ap...

We actually found different reactions to manipulations for diameter and height growth at the treeline.

Article Testing the growth limitation hypothesis for subarctic Scots pine

Just a note on Frank's idea: Why should the bud vers trunk T difference not exist in the previous year? In any year? And shoot length control goes also by internode length, not just by primordial number that is defined in August of the previous year (so there could be more needles per cm of terminal shoot length, or the same number of needles per shorter total shoot length. Such data would be interesting to collect. Fact is that trunk diameter declines slower with elevation than height, and this is most likely phenotypic, not genotypic, so does reflect some response to the concurrent environment.

And a note to Xiaou:

(1) sun exposed branches with needles may warm up due to aerodynamic boundary layer. I am not sure if these authors reported terminal (!) bud temperature for the critical period in spring. 

(2) Chambial activity of the stem always starts much earlier then terminal bud activity. I think this is true for any tree in the temperate-boreal zone and is under hormonal/genetic control. The cambium may start dividing while the ground is still snow covered. See the referenz to Lenz in Alpine Treelines (ATL).

(3) I think abrasion has been oversold in the past. It is very rare, confined to very exposed sites, and is not generally associated with the treeline phenomenon. 

(4) The shrub stature is an aerodynamic trap, permitting heat to accumulate (sort of air conditioning), hence inside such cripples it is significantly warmer than at the top of a 2 m upright tree next to it. I provide info on that in ATL see the attached Ambio paper, I can send a pdf on request.  As soon as shoots emerge from such shruby trees the aerodynamic shelter is gone, and the cold air prevents further growth. Climate warming could allow such krummholz shrub to grow into upright trees, as was shown for the Canadian boreal by S Payette (see ATL)

Article Treelines Will be Understood Once the Functional Difference ...

Actually, I must admit that it seems that Christian is right. I remembered some results of the paper I mentioned below wrongly.

I think that actually for shoot growth there is a quite nice publication of Olavi Junttila 30 years ago. Olavi Junttila  Effects of temperature on shoot growth in northern provenances of Pinus sylvestris L. Tree Physiology I, 185-192 (1986). 

http://treephys.oxfordjournals.org/content/1/2/185.full.pdf

This idea of studying primordia density is actually quite nice. There is a methods published by somebody called Risto Jalkanen that allows actually the analysis of primordia density retrospectively on a tree. 

I think taper  (form) and diametric growth is more has influence of air pressure as suggested in Metzers theory though other hormonal and nutrient theory may also play role. Role / function of apical and lateral meristem in for height and diameter is well known in trees. Moreover secondary growth is unique phenonmemnon for diametric growth in trees which can not be equated with other plants. Treeline characters of Rohodenderon arboreum is best example in Himalayan region.

Thanks again for all the answers. My primary objective to detect the difference of height-diameter is to understand treeline tree's architecture, if low stature ones or multi-stemmed ones have more advantages for heat capture. Exactly, Alpine treeline by Körner gives enough answers. As stated "cripples it is significantly warmer than at the top of a 2 m upright tree next to it", a thermal imaging of both shrubby and erect trees gives intuitive explanation. Further combined with the evidences that shrubby-erect form change under warming condition, different heat capture capacity seems the most reasonable explanation of form change(erect-shrubby tree) at treeline.