Remove gravity-fed wells from the well dataset

[chase-dwelle]

Right now, we have "wells" that are actually gravity fed, like a spring or a rainwater harvesting operation, e.g.,

WELL_ID	LAT_DD	LONG_DD	FUNC	STATUS	COD_FCN	COD_QTY	COD_RESRCE	ADM1	ADM2	ACTIVITY	COUNTRY	WATERSRC	WATERTECH	INSTALLED	MGMT	PAY	SOURCE	RPT_DATE
486554	-0.12982	34.456458	Yes	Functional ( in use)	1	1	0			ksm372	KE	Rainwater (harvesting)	Gravity	2005	Institutional Management	None	Virtual Kenya	6/6/12
485707	-0.109488	34.438404	Yes	Functional ( in use)	1	1	0			ksm367	KE	Rainwater (harvesting)	Gravity	2005	Institutional Management	None	Virtual Kenya	5/6/12
486748	-0.123888	34.548462	Yes	Functional ( in use)	1	1	0			ksm296	KE	Rainwater (harvesting)	Gravity	2008	Institutional Management	2 KES/20L Jerrican	Virtual Kenya	25/04/2012
486747	-0.123888	34.548462	Yes	Functional ( in use)	1	1	0			ksm296	KE	Rainwater (harvesting)	Gravity	2008	Institutional Management	2 KES/20L Jerrican	Virtual Kenya	25/04/2012

These have no relation to GRACE data (since GRACE is groundwater), and therefore should be removed.

[ehaacker]

We can probably keep the springs. We can think of springs as simply places where the groundwater level intersects with the land surface (so a well with a very small effective depth); they're going to be more vulnerable to depleted groundwater, so may be the first to respond to depletion recorded by GRACE. But rainwater harvesting won't be related to changes in groundwater storage - we could maybe still compare with precip, but shouldn't treat those the same as wells/springs.

[chase-dwelle]

Yep @ehaacker is right. I took the lat/lons of a few of the springs/lake sources and looked at them on Google Maps and it definitely seems like the SOURCE column is the source of the well, and not the type of water source.

We then just need to delete the Rainwater (harvesting) entries.