{"success":true,"data":{"author":"AODN","bucketAccessType":"public","bucketListable":true,"bucketName":"aodn-cloud-optimised","bucketPrefix":"satellite_austemp_heatwave_14day.zarr","category":"Earth Observation & Geospatial","cloudProvider":"aws","createdAt":"2026-10-04T00:00:00.000Z","description":"This dataset contains the marine heatwave (MHW) and marine cold spell (MCS) categories from 1 January 2012. The SST Atlas of Australian Regional Seas (SSTAARS) daily climatology (Wijffels et al., 2018) is the baseline for the heat-stress metrics: SSTAARS daily 90th, 50th and 10th percentile thresholds classify MHW and MCS following the Hobday et al. (2018) definitions. The dataset also carries the SST anomaly (SSTa) against the SSTAARS daily climatology and the Degree Heating Days (DHD) metric, the accumulation of daily positive SSTa. Where SST cannot be retrieved because of cloud the cell is blank. AusTemp follows on from the Bureau of Meteorology's decommissioned ReefTemp Next Generation (2012-2024), using the night-time-only IMOS MultiSensor L3S SST product; a GeoPolar MultiSensor L3S adding Himawari-8/9 improves coverage by about 20 percent on average. This Zarr store holds the daily 0.02-degree grids (96-174E, 48-8S) from 2012-01-01 to the present, with a daily field and a 14-day gap-filled mosaic for SST, SSTa, MHW/MCS category and DHD, plus mosaic_age and the source L2P flags.\n\nMarine heatwave conditions in these AusTemp metrics are the driver the South Australian Government linked to the 2025-26 Karenia harmful algal bloom in the South Australian gulfs.","displayTitle":"Satellite - Sea surface temperature - AusTemp - Marine heatwave and cold spell categories, degree heating days and SST anomaly - 14 day mosaic","fuseMountBrowsable":true,"id":"40c8ddfe-22f0-5f2b-8161-eb0a24f8b1d1","license":"http://creativecommons.org/licenses/by/4.0/","objectCountEstimate":19,"probeStatus":"ok","region":"ap-southeast-2","schemaCard":{"accessHints":[{"snippet":"# Task: list the store root (NOT recursive) and print every array's dtype/shape/chunks from the consolidated metadata\naws s3 ls --no-sign-request --region ap-southeast-2 s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr/\naws s3 cp --no-sign-request --region ap-southeast-2 s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr/.zmetadata - | python3 -c \"import json,sys; m=json.load(sys.stdin)['metadata']; [print(k[:-8], m[k]['dtype'], m[k]['shape'], m[k]['chunks']) for k in sorted(m) if k.endswith('/.zarray')]\"","language":"bash"},{"language":"python-anonymous","snippet":"# Task: open the store anonymously (works outside the pod too) and decode the MHW category flags.\n# The S3 settings are not optional: the time axis is 180 small objects and s3fs 2026.9 sends a HEAD\n# before every whole-object GET unless max_concurrency=1; a default open took ~50-60 s, this form ~10 s.\nimport xarray as xr, zarr\nzarr.config.set({'async.concurrency': 64})\nSO = {'anon': True, 'client_kwargs': {'region_name': 'ap-southeast-2'},\n      'config_kwargs': {'max_pool_connections': 64}, 'max_concurrency': 1}\nds = xr.open_zarr('s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr',\n                  consolidated=True, storage_options=SO, drop_variables=['filename'])\nds = ds.sortby('time')  # 29-30 Aug 2026 are stored before 24 Aug 2026; .sel(time=slice(...)) needs a monotonic index\nprint(dict(ds.sizes), str(ds.time.values[0])[:10], '->', str(ds.time.values[-1])[:10])\nprint('lat', float(ds.lat[0]), '->', float(ds.lat[-1]), '| lon', float(ds.lon[0]), '->', float(ds.lon[-1]))\na = ds.MHW_category.attrs\nprint(dict(zip(a['flag_values'], a['flag_meanings'].split())))"},{"snippet":"# Task: open from the FUSE mount inside the pod (same dataset; the mount is read-only, no credentials needed)\nimport xarray as xr\nds = xr.open_zarr('/mnt/archive', consolidated=True, drop_variables=['filename']).sortby('time')\nprint(dict(ds.sizes), list(ds.data_vars))","language":"python-local"},{"language":"python-anonymous","snippet":"# Task: daily 14-day-mosaic SST anomaly at one point (Gulf St Vincent) since 2025-03-01.\n# A point read still fetches whole [5,1000,1300] chunks: one object per 5 days per variable, ~2 MB for the category\n# fields but ~17 MB for float64 ssta_mosaic (measured: 117 objects, 2,060 MB for 580 days). Prefer short ranges here.\nimport time, xarray as xr, zarr\nzarr.config.set({'async.concurrency': 64})\nSO = {'anon': True, 'client_kwargs': {'region_name': 'ap-southeast-2'},\n      'config_kwargs': {'max_pool_connections': 64}, 'max_concurrency': 1}\nds = xr.open_zarr('s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr',\n                  consolidated=True, storage_options=SO, drop_variables=['filename']).sortby('time')\npt = ds.ssta_mosaic.sel(lat=-34.9, lon=138.3, method='nearest').sel(time=slice('2025-03-01', None))\nt0 = time.time(); s = pt.to_series(); dt = time.time() - t0\nprint(f'{len(s)} days, {pt.data.npartitions} chunk objects, {dt:.1f}s, cell lat={float(pt.lat):.3f} lon={float(pt.lon):.3f}')\nprint('monthly mean 14-day-mosaic SST anomaly at Gulf St Vincent (deg C):')\nprint(s.resample('MS').mean().round(2).to_string())"},{"language":"python-anonymous","snippet":"# Task: what marine-heatwave category did the South Australian gulfs show from Sep 2024 to Mar 2025 (the run-up to\n# the Karenia mikimotoi bloom)? Uses the gap-filled 14-day-mosaic category; the bbox is one spatial chunk.\nimport time, xarray as xr, zarr\nzarr.config.set({'async.concurrency': 64})\nSO = {'anon': True, 'client_kwargs': {'region_name': 'ap-southeast-2'},\n      'config_kwargs': {'max_pool_connections': 64}, 'max_concurrency': 1}\nds = xr.open_zarr('s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr',\n                  consolidated=True, storage_options=SO, drop_variables=['filename']).sortby('time')\ncat = ds.MHW_category_mosaic.sel(lat=slice(-36.5, -32.0), lon=slice(135.0, 139.5), time=slice('2024-09-01', '2025-03-31'))\nt0 = time.time(); cat = cat.compute(); dt = time.time() - t0\nprint(f'{dict(cat.sizes)} {cat.nbytes/1e6:.0f} MB in memory, {dt:.1f}s; NaN share {float(cat.isnull().mean()):.3f}')\nsea = cat.notnull().sum(('lat', 'lon'))\nfor k, name in enumerate(['none', 'moderate', 'strong', 'severe', 'extreme']):\n    share = ((cat == k).sum(('lat', 'lon')) / sea).resample(time='1MS').mean().round(2)\n    print(f'{name:9s}', share.to_pandas().to_dict())\npeak = ((cat >= 2).sum(('lat', 'lon')) / sea).to_series()\nprint('days with the largest share of cells at strong or worse:'); print(peak.sort_values(ascending=False).head(5).round(2).to_string())"},{"snippet":"# Task: one full-domain map (2000 x 3900 cells = 6 objects per variable) and a freshness mask from mosaic_age.\n# mosaic_age is NOT days: it holds the low 32 bits of a nanosecond timedelta (age_days * 86400e9 mod 2**32) as int32,\n# so decode it through a lookup before comparing. 0 also means \"no mosaic value\", so mask with the category NaNs.\nimport time, numpy as np, xarray as xr, zarr\nzarr.config.set({'async.concurrency': 64})\nSO = {'anon': True, 'client_kwargs': {'region_name': 'ap-southeast-2'},\n      'config_kwargs': {'max_pool_connections': 64}, 'max_concurrency': 1}\nds = xr.open_zarr('s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr',\n                  consolidated=True, storage_options=SO, drop_variables=['filename']).sortby('time')\nday = ds[['MHW_category_mosaic', 'mosaic_age', 'l2p_flags']].sel(time='2025-03-15')\nt0 = time.time(); day = day.compute(); dt = time.time() - t0\nprint(f'{dict(day.sizes)} {day.nbytes/1e6:.0f} MB in memory, {dt:.1f}s')\ncode = np.array([(d * 86400 * 10**9) % 2**32 for d in range(15)], dtype=np.int64); code[code >= 2**31] -= 2**32\norder = np.argsort(code); raw = day.mosaic_age.values.astype(np.int64)\nidx = np.clip(np.searchsorted(code[order], raw), 0, 14)\nage = xr.DataArray(np.where(code[order][idx] == raw, order[idx], -1), coords=day.mosaic_age.coords).where(day.MHW_category_mosaic.notnull())\nprint('decoded mosaic age (days) value counts:', {int(k): int(v) for k, v in zip(*np.unique(age.values[~np.isnan(age.values)], return_counts=True))})\nfresh = day.MHW_category_mosaic.where(age <= 3)              # cells whose mosaic value was observed within 3 nights\nprint('cells per category (all mosaic cells / observed <=3 nights ago):',\n      {int(k): (int((day.MHW_category_mosaic == k).sum()), int((fresh == k).sum())) for k in range(5)})\nprint('land cells (l2p_flags bit value 2):', int(((day.l2p_flags.fillna(0).astype(int) & 2) > 0).sum()))","language":"python-anonymous"}],"caveats":["Scale: chunks are [5, 1000, 1300] for every 3-D variable, so a point read fetches whole 52 MB-raw chunks. Measured 2026-10-04: 580 days at Gulf St Vincent of ssta_mosaic = 117 objects, 2,060 MB compressed (723 s on a ~3 MB/s link); the same days of MHW_category_mosaic over the entire SA-gulfs box = 117 objects, 159 MB (135 s). Pick categorical/int variables for long series, or cut the time range; the full-domain map is cheap (6 objects per variable, 49 MB for three).","Quirk (image): the geospatial notebook image built before 2026-10-05 ships s3fs 0.4.2 (a 2020 synchronous release) next to zarr 3.1.6, so EVERY s3:// open through xarray or zarr there fails with \"Session.__init__() got an unexpected keyword argument 'asynchronous'\" (measured 2026-10-04 in a real pod). On that image read the FUSE mount instead: xr.open_zarr('/mnt/archive', consolidated=True, ...) works but the open took 39-240 s for these stores in the probe. `pip install -U s3fs` in the pod also fixes it. The starterCell tries s3:// first and falls back to the mount on TypeError.","Scale: the open itself costs ~180 small GETs for the time coordinate. With s3fs defaults xr.open_zarr took 49-58 s (zarr 3.1.6, measured 17:40 UTC); with storage_options max_concurrency=1 (s3fs 2026.9 otherwise sends a HEAD before every whole-object GET), config_kwargs max_pool_connections=64 and zarr.config async.concurrency=64 it took 10 s (18:02) and 91 s on a congested link (18:51). Open once per session.","Quirk: the time index is NOT monotonic: 2026-08-29 and 2026-08-30 are stored before 2026-08-24 (indices 5349-5350). ds.sel(time=slice(...)) raises KeyError on a non-monotonic index; call ds.sortby('time') right after opening (lazy, no data read). No calendar day is missing.","Quirk: mosaic_age (int32, no fill) does NOT hold days. It holds the low 32 bits of a nanosecond timedelta: value = (age_days * 86400e9) mod 2**32 as signed int32, e.g. 1 day = -1857093632, 2 = 580780032, 13 = 1627586560; 0 is both 'observed tonight' and 'no mosaic value'. Decode with a lookup over range(0, 15) (all 5.9 million mosaic cells on 2025-03-15 decoded cleanly: 38 percent age 0, 21 percent age 1, 0.1 percent age 13) and mask with the NaNs of a mosaic variable. The product PDF says 'days since valid observation used per pixel'; the sibling *_australia stores store their *_mosaic_age as float.","DO NOT: compare mosaic_age <= N on the raw values; and DO NOT read the *_mosaic fields as instantaneous: each pixel is the most recent valid night within the previous 14 days, so a 14-day mosaic on a given date can mix observations up to 13 nights apart (ages 0-13 seen on 2025-03-15).","Quirk: the non-mosaic daily fields (sst, ssta, MHW_category, MCS_category, dhd, dhdc) are NaN wherever that night was cloudy; the *_mosaic twins are the gap-filled versions. Over the SA-gulfs box Sep 2024-Mar 2025, 61.8 percent of (time, lat, lon) cells of MHW_category_mosaic were NaN, mostly land (Yorke and Eyre peninsulas): mask land with (l2p_flags & 2) > 0 (1,877,618 land cells on 2025-03-15 of 7.8 million).","Departure from canonical: categories follow Hobday et al. 2018 but are computed per pixel per day as floor(1 + (sst - p90)/(p90 - p50)) from SSTAARS daily percentiles, 0 when sst <= p90, clipped to 4, with NO minimum-duration criterion (Hobday's 5-day rule is not applied), so a single warm night is category >= 1. Night-time-only skin SST with +0.17 C bulk offset. Stored as float64 with flag_values [0,1,2,3,4] = none moderate strong severe extreme; NaN under cloud.","Quirk: dhd, dhdc and their mosaics are summer-only (computed 1 Dec..30 Apr, reset on 1 Dec, dhd_t = max(ssta,0) + dhd_{t-1}); outside those months the arrays exist but carry no accumulation. Their chunks are the heaviest in the store (dhd 16.7 MB, dhd_mosaic 16.5 MB per SA-gulfs chunk).","Quirk: the store grows daily. .zmetadata was rewritten at 16:20 UTC on 2026-10-04 (time 5,385 -> 5,388, last day 2026-09-28 -> 2026-10-01); a Dataset opened before the update keeps the old shape, and slice(..., None) silently ends at the open-time last day. date_created and start_time/stop_time in .zattrs describe the latest source NetCDF, not the store.","Quirk: filename is an object (vlen-utf8) array [time]; it opened and read fine under zarr 3.1.6 and 2.18, but drop_variables=['filename'] costs nothing and protects against stricter zarr versions. crs is an int32 per-time-step array (EPSG:4326), not a scalar; ignore it.","Quirk: lat is ASCENDING (-48 -> -8) in this store, the opposite of the four sibling *_australia and *_14day SST/DHD stores (which run -8.21 -> -45.99); here use sel(lat=slice(-36.5, -32.0)) and do not reuse slices across the family. This grid is also wider (96-174E, 48-8S, 2000 x 3900) than the siblings' 110-163.7E, 45.99-8.21S, 1890 x 2685.","Quirk: time decodes to 00:00 UTC of the analysis date here (the siblings decode to 15:20 UTC, the nominal L3S night-time file time); align by date, not timestamp, when joining the family."],"columns":[{"type":"int32 [time], chunks [30], 'seconds since 1981-01-01', gregorian; decodes to midnight UTC of the analysis date; 5385 values 2012-01-01..2026-09-28, one out-of-order pair (29-30 Aug 2026 stored before 24 Aug 2026)","name":"time"},{"name":"lat","type":"float32 [lat], 2000 values -48.0..-8.0 ASCENDING, step 0.02 (valid_min -48, valid_max -8)"},{"type":"float32 [lon], 3900 values 96.0..174.0 ascending, step 0.02","name":"lon"},{"name":"crs","type":"int32 [time], chunks [5]; EPSG:4326 latitude_longitude grid mapping, carried per time step (ignore)"},{"type":"object (vlen-utf8) [time], chunks [30]; source NetCDF name e.g. 20250315_IMOS_AusTemp-marine-heatwave_AUS_fv02.nc; reads fine under zarr 3.1.6, drop it if your zarr complains","name":"filename"},{"name":"sst","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; Sea Surface Sub-skin Temperature (L3S-01day), degrees Celsius, valid -50..50; SSES-bias-adjusted +0.17 C offset (bulk), NIGHT-TIME only, NaN under cloud"},{"name":"sst_mosaic","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; 14-day SST mosaic: per pixel the most recent valid night within the previous 14 days, degrees Celsius"},{"type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; daily SST anomaly vs SSTAARS daily climatology (1992-2016) for the analysis day-of-year, degrees Celsius","name":"ssta"},{"type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; anomaly of the 14-day mosaic SST vs SSTAARS daily climatology, degrees Celsius","name":"ssta_mosaic"},{"type":"int32 [time, lat, lon], chunks [5,1000,1300], no fill value (0 where no data; mask with NaN of sst_mosaic); days since the valid observation used per pixel, valid 0..14 (0 = observed tonight)","name":"mosaic_age"},{"name":"MHW_category","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; Marine Heat Wave Category from the DAILY sst: flag_values [0,1,2,3,4] = none moderate strong severe extreme (Hobday et al. 2018, cat = floor(1 + (sst - p90)/(p90 - p50)) from SSTAARS daily percentiles, 0 when sst <= p90, clipped to 4); NaN under cloud; stored float64, has least_significant_digit=3"},{"type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; same categories computed from sst_mosaic (gap-filled; the field to use for coverage)","name":"MHW_category_mosaic"},{"name":"MCS_category","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; Marine Cold Spell Category from daily sst, flag_values [0,1,2,3,4] = none moderate strong severe extreme (mirror of MHW using p10/p50)"},{"name":"MCS_category_mosaic","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; Marine Cold Spell Category from sst_mosaic"},{"name":"dhd","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; Degree-Heating Days = running sum of max(ssta,0) since 1 December, degrees Celsius days, valid 0..600; computed Dec-Apr only, reset on 1 Dec"},{"name":"dhd_mosaic","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; Degree-Heating Days from the 14-day mosaic SST anomaly, degrees Celsius days"},{"name":"dhdc","type":"int32 [time, lat, lon], chunks [5,1000,1300], no fill value (0 where no data; mask with NaN of sst_mosaic); number of days that contributed to dhd (days with positive anomaly), valid 0..600"},{"name":"dhdc_mosaic","type":"int32 [time, lat, lon], chunks [5,1000,1300], no fill value (0 where no data; mask with NaN of sst_mosaic); number of days that contributed to dhd_mosaic"},{"name":"l2p_flags","type":"float64 [time, lat, lon], chunks [5,1000,1300], fill NaN; GHRSST L2P flags copied from the L3S source, stored as float64 (NaN where no flag), flag_masks [1,2,4,...,16384] = microwave land ice lake river reserved aerosol analysis lowwind highwind edge terminator reflector swath delta_dn; bit value 2 = land"}],"format":"v2-hybrid","generatedAt":"2026-10-04T00:00:00Z","layout":{"pattern":"Flat Zarr v2 directory: .zmetadata (27 kB, consolidated), .zattrs, .zgroup, then one directory per array. 3-D arrays use chunk keys t.y.x with t = 0..1077 (5-day blocks), y = 0..1 (1000-row blocks), x = 0..2 (1300-col blocks): 6 objects per 5 days per variable, 6,462 objects per variable (MHW_category directory listed 2026-10-04: 6,462 chunk objects, 13.13 GB, median 2.09 MB, max 3.98 MB). Compressed object size depends on the variable and the region: for the 2025-03-15 block the SA-gulfs chunk was 2.2 MB (MHW_category_mosaic), 1.7 (MHW_category), 1.4 (MCS_category), 2.0 (l2p_flags), 2.8 (mosaic_age), 6.6 (sst), 7.5 (ssta), 14.4 (sst_mosaic), 16.7 (dhd), 17.6 MB (ssta_mosaic). time is 180 objects of 30 steps, crs 1078 objects of 5, filename 180 objects; lat and lon one object each. blosc/lz4 level 5, byte shuffle, C order."},"mountPath":"/mnt/archive","overview":"IMOS/BoM AusTemp marine-heatwave product: Hobday et al. 2018 MHW/MCS categories from night-time IMOS MultiSensor L3S GHRSST skin SST (SSES-adjusted, +0.17 C) against the SSTAARS daily climatology (1992-2016) and its daily 90/50/10th percentiles, plus SSTa, Degree Heating Days (Dec-Apr) and 14-day most-recent-valid-pixel mosaics. One Zarr v2 store on a daily 0.02-degree grid, 2000 lat (ASCENDING -48..-8) x 3900 lon (96..174), 5,385 days 2012-01-01..2026-09-28 at 17:14 UTC on 2026-10-04, 5,388 (..2026-10-01) by 18:51 UTC: appended daily. Every 3-D variable is chunked [5 days, 1000, 1300] (52 MB raw), so the store wants MAPS and SMALL BOXES, not long point series: one full-domain day of three variables was 18 objects, 49 MB, 26 s; the SA-gulfs box (225 x 225 cells, one spatial chunk) for 580 days since 2025-03-01 was 117 objects, 159 MB, 135 s; the same 580 days at ONE point of float64 ssta_mosaic was also 117 objects but 2,060 MB, 723 s (categorical chunks ~2 MB, float64 chunks 8-18 MB). Timings are from a ~3 MB/s link; object and byte counts are the durable facts. In the SA gulfs the 14-day-mosaic category rose from 95 percent 'none' in Sep 2024 to 56-58 percent 'moderate' in Dec 2024-Jan 2025 and 69 percent moderate plus 13 percent strong in Mar 2025, peaking 2025-03-15 with 54 percent of sea cells at strong; no severe or extreme cells Sep 2024-Mar 2025 (read 2026-10-04).","rowCount":"time 5,388 x lat 2,000 x lon 3,900 cells per variable (42.0 billion cells; 14 3-D variables) when read 2026-10-04 18:51 UTC; 5,385 x 2,000 x 3,900 at 17:14 UTC the same day","sampleHeaders":[{"truncatedAt":512,"bytes":"{\n    \"chunks\": [\n        5,\n        1000,\n        1300\n    ],\n    \"compressor\": {\n        \"blocksize\": 0,\n        \"clevel\": 5,\n        \"cname\": \"lz4\",\n        \"id\": \"blosc\",\n        \"shuffle\": 1\n    },\n    \"dimension_separator\": \".\",\n    \"dtype\": \"<f8\",\n    \"fill_value\": \"NaN\",\n    \"filters\": null,\n    \"order\": \"C\",\n    \"shape\": [\n        5388,\n        2000,\n        3900\n    ],\n    \"zarr_format\": 2\n}","encoding":"utf8","key":"satellite_austemp_heatwave_14day.zarr/MCS_category/.zarray"},{"truncatedAt":512,"bytes":"{\n    \"_ARRAY_DIMENSIONS\": [\n        \"time\",\n        \"lat\",\n        \"lon\"\n    ],\n    \"comment\": \"Marine cold spell category defined as per Hobday et al., 2018\",\n    \"flag_meanings\": \"none moderate strong severe extreme\",\n    \"flag_values\": [\n        0,\n        1,\n        2,\n        3,\n        4\n    ],\n    \"long_name\": \"Marine Cold Spell Category\",\n    \"source\": \"20120101152000-ABOM-L3S_GHRSST-SSTskin-MultiSensor-1d_night.nc, SSTAARS_daily_fit_001.nc\",\n    \"valid_max\": 4,\n    \"valid_min\": 0\n}","encoding":"utf8","key":"satellite_austemp_heatwave_14day.zarr/MCS_category/.zattrs"}],"sampleObjects":[".zmetadata","MHW_category_mosaic/.zarray","MHW_category_mosaic/964.0.1","ssta_mosaic/964.0.1","mosaic_age/964.0.1","time/160","filename/160"],"sampleRows":[{"ssta_mosaic":2.141,"lon":138.29,"where":"Gulf St Vincent","MHW_category_mosaic":2,"time":"2025-03-15T00:00","lat":-34.89},{"where":"Spencer Gulf","MHW_category_mosaic":1,"time":"2025-03-15T00:00","lat":-33.49,"lon":137.59,"ssta_mosaic":1.933},{"ssta_mosaic":0.796,"lat":-34.89,"lon":138.29,"MHW_category_mosaic":0,"where":"Gulf St Vincent","time":"2025-01-10T00:00"},{"time":"2025-01-10T00:00","where":"Spencer Gulf","MHW_category_mosaic":0,"lat":-33.49,"lon":137.59,"ssta_mosaic":0.655},{"ssta_mosaic":0.147,"lat":-34.89,"lon":138.29,"where":"Gulf St Vincent","MHW_category_mosaic":0,"time":"2024-09-15T00:00"},{"lon":137.59,"where":"Spencer Gulf","MHW_category_mosaic":0,"lat":-33.49,"ssta_mosaic":0.358,"time":"2024-09-15T00:00"}],"sourceObjectKey":"satellite_austemp_heatwave_14day.zarr/MCS_category/.zarray","starterCell":"import time, xarray as xr, zarr\nzarr.config.set({'async.concurrency': 64})\nSO = {'anon': True, 'client_kwargs': {'region_name': 'ap-southeast-2'},\n      'config_kwargs': {'max_pool_connections': 64}, 'max_concurrency': 1}\ntry:\n    ds = xr.open_zarr('s3://aodn-cloud-optimised/satellite_austemp_heatwave_14day.zarr', consolidated=True, storage_options=SO, drop_variables=['filename'])\nexcept TypeError:  # geospatial image before 2026-10-05: s3fs 0.4.2 cannot back zarr 3; the FUSE mount works, slower open\n    ds = xr.open_zarr('/mnt/archive', consolidated=True, drop_variables=['filename'])\nds = ds.sortby('time')\n# South Australian gulfs since the start of the 2025 Karenia bloom; the 14-day mosaic is the gap-filled category.\n# The bbox sits in one spatial chunk, so this costs one ~2 MB object per 5 days (about 115 objects).\ncat = ds.MHW_category_mosaic.sel(lat=slice(-36.5, -32.0), lon=slice(135.0, 139.5), time=slice('2025-03-01', None))\nt0 = time.time(); cat = cat.compute(); print(f'loaded {dict(cat.sizes)} in {time.time()-t0:.1f}s')\nsea = cat.notnull().sum(('lat', 'lon'))\nmonthly = ((cat >= 1).sum(('lat', 'lon')) / sea).resample(time='1MS').mean().round(2)\nprint('share of SA-gulf sea cells in any marine-heatwave category, by month:'); print(monthly.to_pandas().to_string())\ngsv = cat.sel(lat=-34.9, lon=138.3, method='nearest').to_series()\nprint('Gulf St Vincent, days per category since 2025-03-01:', gsv.value_counts(dropna=False).sort_index().to_dict())","starterSamplePath":".","totalObjects":"84,232 objects, ~590 GB (orchestrator summarize, early 2026-10-04); .zmetadata math: 14 3-D variables x 6,462 chunks = 90,468 chunk objects once all are written plus 1,440 time/crs/filename objects; one listed variable directory (MHW_category) = 6,462 objects, 13.13 GB","verifiedRead":{"path":".","facts":["the mount root is a Zarr store: open it with xarray.open_zarr on the root, not the chunk files","under 'satellite_austemp_heatwave_14day.zarr/': 19 directories (e.g. MCS_category, MCS_category_mosaic, MHW_category, MHW_category_mosaic); 3 objects named like .zattrs, 24 B to 27 KB","under 'satellite_austemp_heatwave_14day.zarr/time/': 182 objects named like ##, 136 B to 348 B"],"starterCell":{"seconds":31.41,"ok":true},"summary":"xarray Dataset dims [time=5388, lat=2000, lon=3900], 16 variables e.g. ['MCS_category', 'MCS_category_mosaic', 'MHW_category', 'MHW_category_mosaic', 'crs', 'dhd', 'dhd_mosaic', 'dhdc']","library":"xarray/zarr","snippet":"import os\nimport xarray as xr\nds = xr.open_zarr(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], '.'))\nprint(ds)","family":"zarr","rssDeltaMb":0,"image":"531775725781.dkr.ecr.us-east-1.amazonaws.com/scigantic-notebook-geospatial:latest","attempts":[{"seconds":1.11,"error":"ValueError: binary content, not text","ok":false,"family":"text-head","path":"time/19"},{"path":"time/19","ok":true,"seconds":0.88,"family":"bytes-head"},{"path":"time/19","family":"netcdf-xarray","seconds":7.85,"ok":false,"error":"ValueError: did not find a match in any of xarray's currently installed IO backends ['netcdf4', 'h5netcdf', 'scipy', 'cfradial1', 'datamet', 'furuno', 'gamic', 'gini', 'hpl', 'iris', 'metek', 'nexradlevel2', 'odim', 'rainbow', 'rasterio', 'uf', 'zarr']. Consider explicitly selecting one of the installed engines"},{"path":".","family":"zarr","ok":true,"seconds":45.15}],"verifiedAt":"2026-10-05T12:59:26.003Z","seconds":45.15},"verifiedReadAttempt":{"attempts":[{"ok":false,"family":"text-head","seconds":1.12,"path":"time/19","error":"ValueError: binary content, not text"}],"outcome":"no-reader","at":"2026-10-04T21:15:57.031Z","image":"531775725781.dkr.ecr.us-east-1.amazonaws.com/scigantic-notebook-geospatial:latest","facts":["the mount root is a Zarr store: open it with xarray.open_zarr on the root, not the chunk files","under 'satellite_austemp_heatwave_14day.zarr/': 19 directories (e.g. MCS_category, MCS_category_mosaic, MHW_category, MHW_category_mosaic); 3 objects named like .zattrs, 24 B to 27 KB","under 'satellite_austemp_heatwave_14day.zarr/time/': 182 objects named like ##, 136 B to 348 B"],"starterCell":{"ok":true}}},"summary":"This dataset contains the marine heatwave (MHW) and marine cold spell (MCS) categories from 1 January 2012.","title":"Satellite - Sea surface temperature - AusTemp - Marine heatwave and cold spell categories, degree heating days and SST anomaly - 14 day mosaic","updatedAt":"2026-10-06T22:38:08.183Z","visibility":"public","websiteUrl":"https://catalogue-imos.aodn.org.au/geonetwork/srv/eng/catalog.search#/metadata/2ffccdad-1197-4e41-b412-a9033517cfb2"}}