Some outdoor mixed reality (MR) applications, such as exploratory visualization of smoothly varying environmental fields, tolerate meter-scale global placement error but still require spatially extended overlays to conform to the local terrain. However, standalone MR headsets often lack Global Navigation Satellite System (GNSS) receivers and magnetometers, and flat-ground assumptions can cause overlays to float above or intersect uneven terrain. We present TeAPOT (Terrain-Aligned Positioning and Orientation Toolkit), which transfers smartphone GNSS position and true heading to a headset through fiducial alignment, then reconstructs a local 2.5D terrain patch on the headset from stereo environment-depth observations for terrain-aligned rendering. Across 150 outdoor trials at three locations and standing distances up to 10m, TeAPOT achieved a median end-to-end horizontal placement error of 2.04m (P95 = 5.52m), with end-to-end error strongly correlated with the corresponding standing-position error. A separate 60-session characterization yielded a median absolute true-heading error of 4.08° (P95 = 8.19°), consistent with the observed increase in added error at larger standing distances. For local terrain reconstruction, 30 reconstructions of 1×1m patches across three terrain types yielded a median absolute height error of 0.6cm (P95 = 1.1cm) relative to reference scans of the same patches. The median time to stability was 1.1s, and the median GPU time was 5.6ms. These results demonstrate meter-tolerant outdoor MR on commodity hardware, combining meter-scale absolute placement with centimeter-level local surface-shape accuracy.