From f5907c7e60cab3f98378d67396a9cf404d062793 Mon Sep 17 00:00:00 2001
From: Xylar Asay-Davis <xylarstorm@gmail.com>
Date: Mon, 5 Feb 2024 11:09:22 -0600
Subject: [PATCH] Divide sea ice area/volume data into 10 year chunks

This prevents out-of-memory issues when computing total area and
volume per hemisphere for very long runs.
---
 mpas_analysis/default.cfg            |  9 ++++++---
 mpas_analysis/sea_ice/time_series.py | 10 +++++++++-
 2 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/mpas_analysis/default.cfg b/mpas_analysis/default.cfg
index 8173ccdd0..3903a90f2 100755
--- a/mpas_analysis/default.cfg
+++ b/mpas_analysis/default.cfg
@@ -2175,7 +2175,7 @@ seasons =  ['ANN', 'JFM', 'JAS']
 fieldList = ['significantWaveHeight', 'peakWavePeriod']
 
 era5ObsStartYear = 1959
-era5ObsEndYear = 2021 
+era5ObsEndYear = 2021
 sscciObsStartYear = 1991
 sscciObsEndYear = 2018
 
@@ -2183,7 +2183,7 @@ sscciObsEndYear = 2018
 ## options related to plotting climatology maps of significant wave height
 
 # colormap for model/observations
-colormapNameResult = viridis 
+colormapNameResult = viridis
 # whether the colormap is indexed or continuous
 colormapTypeResult = continuous
 # the type of norm used in the colormap
@@ -2208,7 +2208,7 @@ normArgsDifference = {'vmin': -5., 'vmax': 5.}
 ## options related to plotting climatology maps of peak wave frequency
 
 # colormap for model/observations
-colormapNameResult = plasma 
+colormapNameResult = plasma
 # whether the colormap is indexed or continuous
 colormapTypeResult = continuous
 # the type of norm used in the colormap
@@ -3949,6 +3949,9 @@ titleFontSize = 18
 # plot on polar plot
 polarPlot = False
 
+# the number of years to chunk the data over to prevent running out of memory
+chunkYears = 10
+
 # An optional first year for the tick marks on the x axis. Leave commented out
 # to start at the beginning of the time series.
 # firstYearXTicks = 1
diff --git a/mpas_analysis/sea_ice/time_series.py b/mpas_analysis/sea_ice/time_series.py
index 709327b1f..46f83963e 100644
--- a/mpas_analysis/sea_ice/time_series.py
+++ b/mpas_analysis/sea_ice/time_series.py
@@ -587,10 +587,13 @@ def _compute_area_vol(self):
         indices to process.
         """
 
+        config = self.config
+        chunkYears = config.get('timeSeriesSeaIceAreaVol', 'chunkYears')
+
         outFileNames = {}
         for hemisphere in ['NH', 'SH']:
             baseDirectory = build_config_full_path(
-                self.config, 'output', 'timeSeriesSubdirectory')
+                config, 'output', 'timeSeriesSubdirectory')
 
             make_directories(baseDirectory)
 
@@ -609,6 +612,11 @@ def _compute_area_vol(self):
             startDate=self.startDate,
             endDate=self.endDate)
 
+        nTime = ds.sizes['Time']
+        # chunk into 10-year seguments so we don't run out of memory
+        if nTime > 12 * chunkYears:
+            ds = ds.chunk({'Time': 12 * chunkYears})
+
         for hemisphere in ['NH', 'SH']:
 
             if hemisphere == 'NH':