rain imputation in micro_global errors where months have no rain days

[goldingn]

I'm running some microclimate models for dry places, and some spots have no rainy days in some months in the micro_global() source data. Niamey, Niger is one example. This leads to divide-by-zeros putting NAs in the rainfall data, when used with daily predictions and rainfrac=0, and then the fortran bit errors:

library(NicheMapR)
micro <- micro_global(
  loc = c(2.0840132, 13.5127664),
  timeinterval = 365,
  rainfrac = 0
)
#> If program is crashing, try run.gads = 2.
#> extracting climate data
#> extracting soil moisture data
#> running microclimate model for 365 days by 1 years at site long 2.0840132 lat 13.5127664
#> Note: the output column `SOLR` in metout and shadmet is for unshaded horizontal plane solar radiation
#> Error in microclimate(micro): NA/NaN/Inf in foreign function call (arg 24)

^{Created on 2023-08-18 with reprex v2.0.2}

The offending lines are here (but possibly duplicated in other parts of the package?), and could be patched with some error handling

NicheMapR/R/micro_global.R

Lines 915 to 919 in 329eb11

	RAINFALL1[m]<-(RAINFALL[i]*(1-rainfrac)*rainmult)/RAINYDAYS[i] # make remaining rain fall evenly over the remaining number of rainy days for the month, starting at the beginning of the month
	}
	}else{
	if(rainfrac==0){
	RAINFALL1[m]<-(RAINFALL[i]*rainmult)/RAINYDAYS[i]

[mrke]

OK fixed it by just putting a check for NA after the rainfall vector has been constructed and setting NA to zero. From: Nick Golding ***@***.***> Sent: Friday, August 18, 2023 2:44 PM To: mrke/NicheMapR ***@***.***> Cc: Subscribed ***@***.***> Subject: [mrke/NicheMapR] rain imputation in micro_global errors where months have no rain days (Issue #22) I'm running some microclimate models for dry places, and some spots have no rainy days in some months in the micro_global() source data. Niamey, Niger is one example. This leads to divide-by-zeros putting NAs in the rainfall data, when used with daily predictions and rainfrac=0, and then the fortran bit errors: library(NicheMapR) micro <- micro_global( loc = c(2.0840132, 13.5127664), timeinterval = 365, rainfrac = 0 ) #> If program is crashing, try run.gads = 2. #> extracting climate data #> extracting soil moisture data #> running microclimate model for 365 days by 1 years at site long 2.0840132 lat 13.5127664 #> Note: the output column `SOLR` in metout and shadmet is for unshaded horizontal plane solar radiation #> Error in microclimate(micro): NA/NaN/Inf in foreign function call (arg 24) Created on 2023-08-18 with reprex v2.0.2<https://reprex.tidyverse.org> The offending lines are here (but possibly duplicated in other parts of the package?), and could be patched with some error handling https://github.com/mrke/NicheMapR/blob/329eb11495b6f89cee8cda4bab4eb56e147c0313/R/micro_global.R#L915-L919 — Reply to this email directly, view it on GitHub<#22>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB7IUPBJYPGUKAIZXZL3PCDXV3XKNANCNFSM6AAAAAA3U7P734>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.******@***.***>>

[goldingn]

Awesome! So speedy, thanks Mike!

[goldingn]

Actually, I'm getting another related error now, where it's infilling Inf for the divide by 0. Previously it was 0 rainfall over 0 rainy days (0/0 = NA), but now I have non-zero rainfall on zero rainy days (1/0 = Inf).
A quick fix would be to change this line:

NicheMapR/R/micro_global.R

Line 933 in f3ed05e

RAINFALL[is.na(RAINFALL)] <- 0

to:
RAINFALL[!is.finite(RAINFALL)] <- 0

That makes the asumption that RAINYDAYS is correctly zero and RAINFALL is erroneously nonzero for these months.

[goldingn]

reprex on new coords for completeness:

library(NicheMapR)
micro <- micro_global(
  loc = c(73.0,  22.9),
  timeinterval = 365,
  rainfrac = 0
)
#> If program is crashing, try run.gads = 2.
#> extracting climate data
#> extracting soil moisture data
#> running microclimate model for 365 days by 1 years at site long 73 lat 22.9
#> Note: the output column `SOLR` in metout and shadmet is for unshaded horizontal plane solar radiation
#> Error in microclimate(micro): NA/NaN/Inf in foreign function call (arg 24)

^{Created on 2023-08-18 with reprex v2.0.2}

[mrke]

Great and thanks for the new lingo – ‘reprex’! From: Nick Golding ***@***.***> Sent: Friday, August 18, 2023 4:04 PM To: mrke/NicheMapR ***@***.***> Cc: Michael Kearney ***@***.***>; Comment ***@***.***> Subject: Re: [mrke/NicheMapR] rain imputation in micro_global errors where months have no rain days (Issue #22) reprex on new coords for completeness: library(NicheMapR) micro <- micro_global( loc = c(73.0, 22.9), timeinterval = 365, rainfrac = 0 ) #> If program is crashing, try run.gads = 2. #> extracting climate data #> extracting soil moisture data #> running microclimate model for 365 days by 1 years at site long 73 lat 22.9 #> Note: the output column `SOLR` in metout and shadmet is for unshaded horizontal plane solar radiation #> Error in microclimate(micro): NA/NaN/Inf in foreign function call (arg 24) Created on 2023-08-18 with reprex v2.0.2<https://reprex.tidyverse.org> — Reply to this email directly, view it on GitHub<#22 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB7IUPEK3LYPVGTJMRITKQLXV4AVRANCNFSM6AAAAAA3U7P734>. You are receiving this because you commented.Message ID: ***@***.******@***.***>>