R update values based on event based on multiple columns - r

Question

I recently asked a similar question here for having all "Activities" in one column. The solution provided worked very well. Now I would like to change something in the design to be able to record more detailed information. The table shows information recorded from different fields over several years. All activities on the fields are recored by date. Now I would like to add a "Season" column that groups all values belonging to a harvest season. As harvest season I define the time in between two harvest events. (see table at the bottom on how the result should look like). The problem here is that seeding is sometimes done in the previous year (e.g.2012) but fields are harvested in 2013. All events would need to be grouped as 2013.
What would I need to change if I start recording more information and give all "Activities" a separate column? I tried:
library(data.table)
DF <- read.table(text="ID|Field|Date |Tillage|Seeding|Fertilizer|Spraying|Harvest
1|A |2012/08/01|Plough |NA|NA|NA|NA
2|A |2012/08/24|NA |Wheat|NA|NA|NA
3|A |2013/03/05|NA |NA|NA|ProduktA|NA
4|A |2013/03/05|NA|NA|TypeB|NA|NA
5|A |2013/07/25|NA |NA|NA|NA|9t
6|B |2012/09/01|Plough |NA|NA|NA|NA
7|B |2012/09/05|NA |Barley|NA|NA|NA
8|B |2013/04/05|NA |NA|NA|ProductB|NA
9|B |2013/07/28|NA |NA|NA|NA|10t
10|B |2010/08/24|Cultivator |NA|NA|NA|NA
11|B |2010/09/29|NA |NA|NA|NA|NA
12|B |2011/05/01|NA|NA|TypeB|NA|NA
13|B |2011/07/12|NA |NA|NA|NA|6t
14|A |2011/09/01|NA |Barley|NA|NA|NA
15|A |2011/10/10|NA |NA|NA|ProductC|NA
16|A |2012/04/10|NA|NA|TypeA|NA|NA
17|A |2012/08/02|NA |NA|NA|NA|7t|",
sep="|", header=TRUE, stringsAsFactors=FALSE)
DT <- data.table(DF)
DT[, Harvest:=gsub(" ", "", Harvest, fixed=TRUE)]
DT[, Date:=as.POSIXct(Date)]
setkeyv(DT, c("Field", "Date"))
DT[, Season:=cumsum(c("", !is.na(head(Harvest, -1)))), by=Field]
DT[, Season:=max(year(Date)), by=list(Field, Season)]
However, that seems not to work. The result should look like this with a "season" column at the end that indicates the season:
ID|Field|Date |Tillage|Seeding|Fertilizer|Spraying|Harvest|Season
1|A |2012/08/01|Plough |NA|NA|NA|NA|2013
2|A |2012/08/24|NA |Wheat|NA|NA|NA|2013
3|A |2013/03/05|NA |NA|NA|ProduktA|NA|2013
4|A |2013/03/05|NA|NA|TypeB|NA|NA|2013
5|A |2013/07/25|NA |NA|NA|NA|9t|2013
6|B |2012/09/01|Plough |NA|NA|NA|NA|2013
7|B |2012/09/05|NA |Barley|NA|NA|NA|2013
8|B |2013/04/05|NA |NA|NA|ProductB|NA|2013
9|B |2013/07/28|NA |NA|NA|NA|10t|2013
10|B |2010/08/24|Cultivator |NA|NA|NA|NA|2011
11|B |2010/09/29|NA |NA|NA|NA|NA|2011
12|B |2011/05/01|NA|NA|TypeB|NA|NA|2011
13|B |2011/07/12|NA |NA|NA|NA|6t|2011
14|A |2011/09/01|NA |Barley|NA|NA|NA|2012
15|A |2011/10/10|NA |NA|NA|ProductC|NA|2012
16|A |2012/04/10|NA|NA|TypeA|NA|NA|2012
17|A |2012/08/02|NA |NA|NA|NA|7t||2012

Answer 1

The only difference to OP's other question is that there are some additional columns and that the condition for extracting the harvest dates for applying my rolling join answer needs to be amended:
library(data.table)
setDT(DF)[!is.na(Harvest), .(Field, Date, Season = year(Date))][
DF, on = .(Field, Date), roll = -Inf]
Field Date Season ID Tillage Seeding Fertilizer Spraying Harvest
1: A 2012/08/01 2012 1 Plough NA NA NA NA
2: A 2012/08/24 2013 2 NA Wheat NA NA NA
3: A 2013/03/05 2013 3 NA NA NA ProduktA NA
4: A 2013/03/05 2013 4 NA NA TypeB NA NA
5: A 2013/07/25 2013 5 NA NA NA NA 9t
6: B 2012/09/01 2013 6 Plough NA NA NA NA
7: B 2012/09/05 2013 7 NA Barley NA NA NA
8: B 2013/04/05 2013 8 NA NA NA ProductB NA
9: B 2013/07/28 2013 9 NA NA NA NA 10t
10: B 2010/08/24 2011 10 Cultivator NA NA NA NA
11: B 2010/09/29 2011 11 NA NA NA NA NA
12: B 2011/05/01 2011 12 NA NA TypeB NA NA
13: B 2011/07/12 2011 13 NA NA NA NA 6t
14: A 2011/09/01 2012 14 NA Barley NA NA NA
15: A 2011/10/10 2012 15 NA NA NA ProductC NA
16: A 2012/04/10 2012 16 NA NA TypeA NA NA
17: A 2012/08/02 2012 17 NA NA NA NA 7t
Note that the rolling join has exhibited a flaw in the sample dataset. Row 1 shows Season 2012 although the subsequent harvest (according to OP's ID) should be in 2013. Reason is that the dates for tillage and harvest are intermixed for field A. Tillage date of Field A in row 1 is 2012/08/01 while the harvest date of the same field in row 17 is 2012/08/02, one day after tillage.
In case the column order is important, the setcolorder() function can be used to order the columns in place, i.e., without copying:
result <- setDT(DF)[!is.na(Harvest), .(Field, Date, Season = year(Date))][
DF, on = .(Field, Date), roll = -Inf]
setcolorder(result, c(names(DF), "Season"))[]
ID Field Date Tillage Seeding Fertilizer Spraying Harvest Season
1: 1 A 2012/08/01 Plough NA NA NA NA 2012
2: 2 A 2012/08/24 NA Wheat NA NA NA 2013
3: 3 A 2013/03/05 NA NA NA ProduktA NA 2013
4: 4 A 2013/03/05 NA NA TypeB NA NA 2013
5: 5 A 2013/07/25 NA NA NA NA 9t 2013
6: 6 B 2012/09/01 Plough NA NA NA NA 2013
7: 7 B 2012/09/05 NA Barley NA NA NA 2013
8: 8 B 2013/04/05 NA NA NA ProductB NA 2013
9: 9 B 2013/07/28 NA NA NA NA 10t 2013
10: 10 B 2010/08/24 Cultivator NA NA NA NA 2011
11: 11 B 2010/09/29 NA NA NA NA NA 2011
12: 12 B 2011/05/01 NA NA TypeB NA NA 2011
13: 13 B 2011/07/12 NA NA NA NA 6t 2011
14: 14 A 2011/09/01 NA Barley NA NA NA 2012
15: 15 A 2011/10/10 NA NA NA ProductC NA 2012
16: 16 A 2012/04/10 NA NA TypeA NA NA 2012
17: 17 A 2012/08/02 NA NA NA NA 7t 2012
Data
library(data.table)
DF <- fread(
"ID|Field|Date |Tillage|Seeding|Fertilizer|Spraying|Harvest
1|A |2012/08/01|Plough |NA|NA|NA|NA
2|A |2012/08/24|NA |Wheat|NA|NA|NA
3|A |2013/03/05|NA |NA|NA|ProduktA|NA
4|A |2013/03/05|NA|NA|TypeB|NA|NA
5|A |2013/07/25|NA |NA|NA|NA|9t
6|B |2012/09/01|Plough |NA|NA|NA|NA
7|B |2012/09/05|NA |Barley|NA|NA|NA
8|B |2013/04/05|NA |NA|NA|ProductB|NA
9|B |2013/07/28|NA |NA|NA|NA|10t
10|B |2010/08/24|Cultivator |NA|NA|NA|NA
11|B |2010/09/29|NA |NA|NA|NA|NA
12|B |2011/05/01|NA|NA|TypeB|NA|NA
13|B |2011/07/12|NA |NA|NA|NA|6t
14|A |2011/09/01|NA |Barley|NA|NA|NA
15|A |2011/10/10|NA |NA|NA|ProductC|NA
16|A |2012/04/10|NA|NA|TypeA|NA|NA
17|A |2012/08/02|NA |NA|NA|NA|7t")