Elasticsearch aggregate by geo location - search

Question

I want to group data by their location with shapes and bounding boxes. For example group the results by hemisphere:
GET cdc/_search
{
"from": 0,
"size": 0,
"query": {
"match_all": {}
},
"aggs": {
"group_by_geo": {
"geo_bounding_box": {
"field": "location",
"boxes": [
{
"top_left": {
"lat": 90,
"lon": -180
},
"bottom_right": {
"lat": 0,
"lon": 0
}
},
{
"top_left": {
"lat": 90,
"lon": 0
},
"bottom_right": {
"lat": 0,
"lon": 180
}
},
{
"top_left": {
"lat": 0,
"lon": -180
},
"bottom_right": {
"lat": -90,
"lon": 0
}
},
{
"top_left": {
"lat": 0,
"lon": 0
},
"bottom_right": {
"lat": -90,
"lon": 180
}
}
]
},
"aggs": {
"over_time": {
"date_histogram": {
"field": "date",
"interval": "day",
"format": "yyyy-MM-dd"
},
"aggs": {
"AvgTemp": {
"avg": {
"field": "hits"
}
}
}
}
}
}
}
}
So the over_timepart works fine isolated but this query gives me "reason": "Could not find aggregator type [geo_bounding_box] in [group_by_geo]"
My syntax is inspired by range aggregations but apparently this is not working in this case.
Is this possible or do I have to make separate queries filtering by the 4 boxes?

Answer 1

There's no geo_bounding_box aggregation, (there's a geo_bounding_box filter, though).
You can achieve what you need using a filters aggregation containing one geo_bounding_box filter for each bounding box. Basically something like this:
{
"from": 0,
"size": 0,
"query": {
"match_all": {}
},
"aggs": {
"group_by_geo": {
"filters": {
"filters": {
"box1": {
"geo_bounding_box": {
"location": {
"top_left": {
"lat": 90,
"lon": -180
},
"bottom_right": {
"lat": 0,
"lon": 0
}
}
}
},
"box2": {
"geo_bounding_box": {
"location": {
"top_left": {
"lat": 90,
"lon": 0
},
"bottom_right": {
"lat": 0,
"lon": 180
}
}
}
},
"box3": {
"geo_bounding_box": {
"location": {
"top_left": {
"lat": 0,
"lon": -180
},
"bottom_right": {
"lat": -90,
"lon": 0
}
}
}
},
"box4": {
"geo_bounding_box": {
"location": {
"top_left": {
"lat": 0,
"lon": 0
},
"bottom_right": {
"lat": -90,
"lon": 180
}
}
}
}
}
},
"aggs": {
"over_time": {
"date_histogram": {
"field": "date",
"interval": "day",
"format": "yyyy-MM-dd"
},
"aggs": {
"AvgTemp": {
"avg": {
"field": "hits"
}
}
}
}
}
}
}
}