search

peteWT / cec_apl

California Energy Commission has funded a project through the Electric Program Investment Charge (EPIC) to evaluate the viability of the All Power Labs Powertainer.
MIT License
0 stars 0 forks source link

Conifer assumption #9

Open jd-lara opened 8 years ago

jd-lara commented 8 years ago

Hi @carmentubbesing, we have been looking at the data for the high yield pixels. We noted that in many cases the assumption related to all the dead trees been conifers makes sense and the results look great. For example look at this screenshots, the purple dots are pixels with 0 biomass, and the yellowish ones are the pixels with biomass > 60,000 kg.

SC1

and

SC2.

This pictures show that actually the LEMMA database is properly telling where the forest is and where there is no forest.

But in areas where the conifers are not prevalent, the code is pushing all the dead biomass into single pixels and hence creating these high yield pixels. For example.

SC3

SC4

SC5

SC6

SC7

In some cases these polygons are located in the transition areas between Oaklands to Conifers, this might require a revision of that assumption in the code. This is why there are may pixels with high yield located next to others where there is a prevalence of zeros biomass pixels.

I am trying to make a query to detect where is this happening. But preliminarily this is an overview of how it looks like. The red stars are the polygons where the average biomass per pixel (for the pixels with biomass > 0) is above 60,000 kg.

SC8

carmentubbesing commented 8 years ago

@jdlara-berkeley That makes sense, but I can't see the screenshots you tried to upload. Could you try again or send them to me another way?

jd-lara commented 8 years ago

@carmentubbesing I hit post to early, I changed the post take a look.

carmentubbesing commented 8 years ago

@jdlara-berkeley Great work with this, these figures are super helpful in visualizing the problem.

I think this gets down to an operations question: do you want to model gasification of all trees, or just conifers? I'm not sure whether APL is interested in hardwoods or whether they could get permission to cut them down.

Another thing to consider that I didn't do in my R code is filtering for land ownership. That might cut out some of the non-conifer forest.

jd-lara commented 8 years ago

@carmentubbesing the work has been great and I think is not a major issue that can disqualify the results, is just a refinement.

I think that we should think about including more species, and add a new column that classifies that pixel by the wood type, and from there see what can be used or not. I believe that an analysis of the available biomass from the CalFire polygons, types of wood, distances to road, etc. considering the complexities is an analysis on itself besides the APL gasification problem which we can address later.

We should continue analyzing the data with the same approach, and later on based on other maps make a spatial query to see if the pixel is inside public land or not, that won't be that hard.

The main thing to resolve now is this overallocation in the case of polygons where the main species is not a conifer.

peteWT commented 8 years ago

@jdlara-berkeley @carmentubbesing we either need to restrict the calfire polygons to FOR_TYPE1 that are conifer dominant or expand the LEMMA pixels that can be allocated biomass to include all types that match FOR_TYPE1 from Calfire. Otherwise we are attributing the non-confier volumes to confer dominant pixels whihc doesn't make sense. This is great work.

jd-lara commented 7 years ago

@peteWT do you think we can work on making this fix to run @carmentubbesing code again before the end of the end of the month?