Update to the latest spartacus

ChangeLog

@@ -1,4 +1,35 @@
-version 0.7.2
+version 0.8
+	- New versions of the RAMI-V test scene input files (see
+	  test/rami5/README), using explicit vegetation fractional standard
+	  deviation
+
+version 0.7.3 (27 October 2021)
+	- Redefined veg_contact_fraction: now the fraction of the building
+	  edge that is touching veg, not the fraction of the vegetation edge
+	  that is touching a building
+	- Default veg_contact_fraction (in
+	  spartacus_surface_read_input.F90) is equal to the fraction of the
+	  non-building area containing vegetation, equivalent to randomly
+	  placed vegetation
+	- Normalized perimeter of vegetation-air interface now properly
+	  takes into account the building fraction in the symmetric
+	  vegetation scale formula; this is now in radsurf_norm_perim.F90
+	- vegetation_isolation_factor treatment changed such that a value
+	  of 1 now indicates that thick and thin vegetation regions equally
+	  likely to be in contact with the clear-air region, rather than
+	  that only the thick region is in contact with clear-air
+	- Modified the test/rami5/scene_nc files to use the new definition
+	  of veg_contact_fraction and a consistent definition of veg_scale
+	  to the code
+	- Added surface types 4 (SimpleUrban) and 5 (InfiniteStreet): fast
+	  methods for single-layer unvegetated urban canopies
+	- test/single_layer case for testing the new surface types
+	- Namelist parameter radsurf_config:isurfacetype can be used to
+	  override the surface_type in the input file
+	- Namelist parameter radsurf_config:vegetation_extinction_scaling
+	  can scale the vegetation extinction in the input file
+
+version 0.7.2 (4 October 2021)
 	- Longwave urban modifications to give correct emission from
 	  walls, fixing error most noticeable with 2-streams - see
 	  radtool_legendre_gauss.F90 and radsurf_urban_lw.F90
@@ -20,6 +51,9 @@ version 0.7.2
 	  or no clear-air region): exchange coefficients to/from such
 	  regions are now zero, and eigenvalues only computed for regions of
 	  finite size
+	- Added solar_zenith_angle to driver namelist to specify in
+	  degrees as an alternative to cos_solar_zenith_angle
+	- Added RAMI-V test scenes in test/rami5
 
 version 0.7.1 (23 October 2020)
 	- Added Apache license statement to each source file

Makefile

@@ -83,7 +83,7 @@ libradsurf: libradtool
 driver: libradsurf
 	cd driver && $(MAKE)
 
-test: test_simple test_rami4pilps test_urban
+test: test_simple test_rami4pilps test_urban test_rami5 test_single_layer
 
 test_simple:
 	cd test/simple && $(MAKE) test
@@ -94,9 +94,17 @@ test_rami4pilps:
 test_urban:
 	cd test/urban && $(MAKE) test
 
+# Single profile
 test_urban_single:
 	cd test/urban && $(MAKE) test_single
 
+# Single layer
+test_single_layer:
+	cd test/single_layer && $(MAKE) test
+
+test_rami5:
+	cd test/rami5 && $(MAKE) test
+
 test_code:
 	cd driver && $(MAKE) test_code
 
@@ -108,6 +116,8 @@ clean-tests:
 	cd test/simple && $(MAKE) clean
 	cd test/rami4pilps && $(MAKE) clean
 	cd test/urban && $(MAKE) clean
+	cd test/rami5 && $(MAKE) clean
+	cd test/single_layer && $(MAKE) clean
 
 clean-toplevel:
 	cd radsurf && $(MAKE) clean

Makefile_include.gfortran

@@ -14,8 +14,11 @@ OMPFLAG = -fopenmp
 ifndef DEBUG
 # --NORMAL CONFIGURATION--
 
-# Optimization flags
-OPTFLAGS = -O3
+# Optimization flags: note that experience with gfortran is that the
+# code is faster with -O2 than with -O3. The "-march=native" optimizes
+# for the architecture on which the compilation takes place, but this
+# means the code might not run on older CPUs.
+OPTFLAGS = -O2 -march=native
 
 # Warning flags: all except those that warn about unused stuff
 WARNFLAGS = -Wall -Wno-unused-label -Wno-unused-dummy-argument -Wno-unused-variable -Wimplicit-interface
@@ -30,4 +33,9 @@ OPTFLAGS = -O0
 WARNFLAGS = -Wall
 DEBUGFLAGS = -g -ffpe-trap=invalid,zero,overflow -fcheck=bounds -finit-real=snan
 
-endif
+endif
+
+ifdef GPROF
+# Add gprof output
+DEBUGFLAGS += -pg
+endif

README.md

@@ -2,7 +2,7 @@
 
 Robin Hogan <[email protected]>
 
-This document was last updated 10 October 2020
+This document was last updated 6 October 2021
 
 
 ## INTRODUCTION
@@ -105,6 +105,8 @@ RAMI4PILPS test cases that were used by Hogan et al. (2018).
 The `test/urban` directory contains an urban profile from Fig. 1 of
 Hogan (2019).
 
+The `test/rami5` directory contains the files necessary to run five of
+the actual forest scenes from the RAMI-V intercomparison project.
 
 ## LICENCE
 

TODO

@@ -1,18 +1,16 @@
 FEATURES
 
-Compute leaf sunlit fraction (wavelength independent)
-Broadband or spectral fluxes each optional
 Gas optics
-Closed forest: optimization by looking for veg cover = 0 or 1
-Report radiation components into a horizontal and vertical surface for calculation MRT
-Road/roofs/wall direct fraction, and direct flux attenuated only by air
 Dump out inputs like in ecRad
 Combine lw_spectral_props%calc_monochromatic_emission and calc_simple_spectrum_lw into one function call for broadband Planck function
 Support implicit solvers, e.g. windows in TEB with very low heat capacity
-Way to calculate thermal comfort / mean radiant temperature considering diffuse angles
+Store norm_perims for verification
 
 BUG FIXES
 
+Fix use of veg_contact_fraction
+test_surface_out.nc fails with number of veg regions = 1
+Fix crash when vegetation region has zero vegetation extinction, esp. in single precision
 Fix non-zero residual in urban longwave when nstream>=2
 
 CHECK