If you're already springing for the geothermal, why aren't you just running the garage as a separate zone? The heat load of the garage is well under a single-ton of uptick on the geothermal specifications- smaller than the granularity between sizes within a model line. It would add a small amount to the ground heat exchange loop- that's about it. If you're drilling through granite for the geo even a small upsizing can be pretty expensive, but if it's trenched in slinkys or a pond loop it's an inconsequential number.
Also note, iun the main house your heat load numbers might be in within the output range of the Daikin Altherma, air-to-water high-efficiency modulating heat pump system if you're doing radiant or low-temp panel radiators for heating there too, which is usually $20K or less. If your geo proposals are all air-output and your floor plan is fairly open you'd likely be able to heat the place with $7-12K of high-efficiency ductless, 2 or 3 heats. In your climate the seasonal average COP of the -FHxxNA series mini-splits will run about 3.5, which is comparable to "typical" geothermal systems at your groundwater temps when the all-in power consumption of pumps & air handlers are factored in.
I have nothing against geo, but understand that it comes with both higher costs & higher risks. Every geo system is a custom system, and the net efficiency is determined by the quality of the design(er), and it's substantially more expensive per output-BTU than ductless air source solutions. Ductless air source heat pumps have very low design risk- it's apre-engineered "system in a can"- air is air, and behaves the same everywhere, with the primary determinant of performance being the seasonal temperatures. A decade ago ductless systems weren't up to snuff for heating in your climate, but they have made HUGE strides since then. As recently 4-5 years ago the "FExxnA" Mitsubishi were seasonally averaging a COP of about 3.0 in your climate zone, but their direct decendants the "-FHxxNA" deliver more than 20% more heat per kwh. At the difference in upfront cost for say 3 tons of state-of-the art geo delivering a COP of 4 and 3-4 tons of ductless delivering a COP of 3.5+ you can more than make up the annual kwh delta with a slightly larger PV array and still have money to party with.
For verification of efficiency & output performance of the Mitsubishi cold temp ductless systems in a comparably cool climate see the discussion about the Eastern Idaho cluster in Idaho falls in the addendum beginning on page 119 (PDF pagination) of this document, which consisted of ten MSZ-FE12NA units installed & monitored in-situ in occupied houses. The recently released "FH" versions are even considerably more efficient than those (now 5 year old) systems.
The primary down side for ductless is heat distribution to doored off rooms, but if you're putting in U0.20 triple-panes and decently high-R in the main house too that's not much of a problem. A couple years ago I was involved in a deep energy retrofit on 3-story ~3000' house in central MA (outside design temp= +5F), and with ~U0.025 walls (R40-ish whole wall, after thermal bridging) and U0.18 Paradigm double-hungs the individual room loads were low enough that it could be heated with a single ductless head per floor, with at worst a 5F temperature difference @ 0F outdoor temps in the doored off rooms with the doors closed. It would have made it just fine with three 1-tonners, but the owner got nervous (having no experience with high-R houses) and three 1.5 tonners (Mitsubishi MSZ-FE18NA) were installed. Due to the oversizing they're running slightly less efficiently than if the one-ton units had been installed, not that you'd notice in the power billing. With the oversizing that place is good down to about -30F to 35F, even though the units do not have specified output at those temps.
I have also been corresponding with guy in Quebec (who also posts on a few blogs) who is heating his place with three Fujitsu XLTH series mini-splits who sailed through just fine at -32F this winter. The nameplate efficiency of that series is a bit behind the -FH Mitsubishis, but the capacity is clearly there even though the output is only fully specified down to -15F. (These units are good enough to qualify for subsidy across the lake from you in VT.)
This technology has arrived- totally ready for prime time. It's low risk, and dirt-cheap relative to drilled geothermal- pond loops or trenched slinkys can sometimes be competitive if you needed more than 3 heads to heat the place, but sometimes not. Ductless heat pumps have become the most common HVAC solution for "Net Zero Energy" houses and certified PassiveHouse homes in New England, due to the low cost, high efficiency,high reliability, and high comfort due to the modulating output. (They run higher exit-air temp than air-delivery geothermal too.)
Bottom line, whatever you do, make sure you get the heat load calculations dialed & checked three ways- the LAST thing you want to do is buy too MUCH geothermal for your house! If you want to cheap out on the garage radiant as a separate even the smallest electric boiler like say, the EMB-S-1 (1.1kw) would have you covered, and at an up-front cost of well under a grand for the boiler. With no fuel lines or venting to run and a much cheapr boiler it's tiny fraction of the cost of installing a propane boiler. Unless propane prices crash (not too likely, given the trends) the difference in operational cost between condensing propane and an electric boiler would literally never pay off the difference in installed within the lifecycle of the ridiculously oversized propane boiler. If you would go wild with worry that the tiny 1.1kilowatt puppy isn't enough you could bump up to the EMB-S-2 (2.5kw) and be ~3x oversized for the load with a $700 boiler. If you're not going to serve that zone with a heat pump (either geo or ductless), that's a far more appropriate solution than any propane burner.
But if you did a real Manual-J heat load on the "after" picture rather than the down & dirty I=B=R approach I sketched out for you, the load numbers would probably come in even lower. Look up the U-factors on your exact windows & doors (add insulated-glass storm doors if the doors are uninsulated.) There isn't a published U-factor for your wall construction, but it can be calculated pretty easily on first-principles. I rounded up, and used a presumed framing fraction of 25%. But with so few windows & doors the framing-fraction is probably less than 25%, and I probably should have rounded down if accuracy was the goal.
Also note, iun the main house your heat load numbers might be in within the output range of the Daikin Altherma, air-to-water high-efficiency modulating heat pump system if you're doing radiant or low-temp panel radiators for heating there too, which is usually $20K or less. If your geo proposals are all air-output and your floor plan is fairly open you'd likely be able to heat the place with $7-12K of high-efficiency ductless, 2 or 3 heats. In your climate the seasonal average COP of the -FHxxNA series mini-splits will run about 3.5, which is comparable to "typical" geothermal systems at your groundwater temps when the all-in power consumption of pumps & air handlers are factored in.
I have nothing against geo, but understand that it comes with both higher costs & higher risks. Every geo system is a custom system, and the net efficiency is determined by the quality of the design(er), and it's substantially more expensive per output-BTU than ductless air source solutions. Ductless air source heat pumps have very low design risk- it's apre-engineered "system in a can"- air is air, and behaves the same everywhere, with the primary determinant of performance being the seasonal temperatures. A decade ago ductless systems weren't up to snuff for heating in your climate, but they have made HUGE strides since then. As recently 4-5 years ago the "FExxnA" Mitsubishi were seasonally averaging a COP of about 3.0 in your climate zone, but their direct decendants the "-FHxxNA" deliver more than 20% more heat per kwh. At the difference in upfront cost for say 3 tons of state-of-the art geo delivering a COP of 4 and 3-4 tons of ductless delivering a COP of 3.5+ you can more than make up the annual kwh delta with a slightly larger PV array and still have money to party with.
For verification of efficiency & output performance of the Mitsubishi cold temp ductless systems in a comparably cool climate see the discussion about the Eastern Idaho cluster in Idaho falls in the addendum beginning on page 119 (PDF pagination) of this document, which consisted of ten MSZ-FE12NA units installed & monitored in-situ in occupied houses. The recently released "FH" versions are even considerably more efficient than those (now 5 year old) systems.
The primary down side for ductless is heat distribution to doored off rooms, but if you're putting in U0.20 triple-panes and decently high-R in the main house too that's not much of a problem. A couple years ago I was involved in a deep energy retrofit on 3-story ~3000' house in central MA (outside design temp= +5F), and with ~U0.025 walls (R40-ish whole wall, after thermal bridging) and U0.18 Paradigm double-hungs the individual room loads were low enough that it could be heated with a single ductless head per floor, with at worst a 5F temperature difference @ 0F outdoor temps in the doored off rooms with the doors closed. It would have made it just fine with three 1-tonners, but the owner got nervous (having no experience with high-R houses) and three 1.5 tonners (Mitsubishi MSZ-FE18NA) were installed. Due to the oversizing they're running slightly less efficiently than if the one-ton units had been installed, not that you'd notice in the power billing. With the oversizing that place is good down to about -30F to 35F, even though the units do not have specified output at those temps.
I have also been corresponding with guy in Quebec (who also posts on a few blogs) who is heating his place with three Fujitsu XLTH series mini-splits who sailed through just fine at -32F this winter. The nameplate efficiency of that series is a bit behind the -FH Mitsubishis, but the capacity is clearly there even though the output is only fully specified down to -15F. (These units are good enough to qualify for subsidy across the lake from you in VT.)
This technology has arrived- totally ready for prime time. It's low risk, and dirt-cheap relative to drilled geothermal- pond loops or trenched slinkys can sometimes be competitive if you needed more than 3 heads to heat the place, but sometimes not. Ductless heat pumps have become the most common HVAC solution for "Net Zero Energy" houses and certified PassiveHouse homes in New England, due to the low cost, high efficiency,high reliability, and high comfort due to the modulating output. (They run higher exit-air temp than air-delivery geothermal too.)
Bottom line, whatever you do, make sure you get the heat load calculations dialed & checked three ways- the LAST thing you want to do is buy too MUCH geothermal for your house! If you want to cheap out on the garage radiant as a separate even the smallest electric boiler like say, the EMB-S-1 (1.1kw) would have you covered, and at an up-front cost of well under a grand for the boiler. With no fuel lines or venting to run and a much cheapr boiler it's tiny fraction of the cost of installing a propane boiler. Unless propane prices crash (not too likely, given the trends) the difference in operational cost between condensing propane and an electric boiler would literally never pay off the difference in installed within the lifecycle of the ridiculously oversized propane boiler. If you would go wild with worry that the tiny 1.1kilowatt puppy isn't enough you could bump up to the EMB-S-2 (2.5kw) and be ~3x oversized for the load with a $700 boiler. If you're not going to serve that zone with a heat pump (either geo or ductless), that's a far more appropriate solution than any propane burner.
But if you did a real Manual-J heat load on the "after" picture rather than the down & dirty I=B=R approach I sketched out for you, the load numbers would probably come in even lower. Look up the U-factors on your exact windows & doors (add insulated-glass storm doors if the doors are uninsulated.) There isn't a published U-factor for your wall construction, but it can be calculated pretty easily on first-principles. I rounded up, and used a presumed framing fraction of 25%. But with so few windows & doors the framing-fraction is probably less than 25%, and I probably should have rounded down if accuracy was the goal.