5 ms·

Show HN: A Twitter bot I made that posts real-time deforestation data

101 points by imnotanerd 6y ago

imnotanerd 6y ago
Heads up: [1] This is a project I've been working in about a year now and is an honorary successor to a similar project a friend started. You can read in detail how it works at https://imnotanerd.herokuapp.com/3 https://imnotanerd.herokuapp.com/3. Basically it takes data from the API at globalforestwatch.org that uses satellite imagery to gather area of potential forest loss and paints it in a map to make it easy to imagine. There is no consideration about forest regrowth. This bot does not track forestal area changes, it tracks deforestation. [2] I'd like to get some community feedback. The bot has now been running for quite some time and it looks like now it works with consistency (The GFW API data is not reliable when recently published). Mainly I'd like to see how this bot can be improved regarding the data treatment and the display of it, as I still think the maps, despite being accurate, are a little "underperformant" and could be improved. [3] Yes. I posted this twice, I forgot to put "Show HN" in the title the first time and I realised too late. Now I can't edit nor remove the previous submission.
- abpavel 6y ago
  I see a number of issues: 1. no deforestation before July 9th. 2. In 10 days between Aug4 and Aug14 2% of the area was deforested, signifying all forrest will be gone in about 1.5 years. However forrest area in 2015 was almost the same size as reported total. Can't be right. Are you mixing ha and km2 by any chance? 3. No tracking between what is part of Forest Management Act, what are is part of designated production forest, and what is illegal deforestation. 4. No data on loss due to natural causes such as wind, fire, and no data on reforestation numbers.
  - dna_polymerase 6y ago
    Point 4 might be irrelevant to the cause and more importantly part of a negative (hard to track) feedback loop. A patch of forest destroyed by human activity might be more susceptible to stronger winds or extreme weather in general.
  - imnotanerd 6y ago
    1. That's the date the bot started tracking. Actually, deforestation data starts in 2015 but I thought it'd be dishonest of the bot to count in deforestation that happened before the bot release. First drafts of the bot worked this way but that behavior was abandoned. 2. What do you mean by 2%? I guarantee I'm not mixing units, the remaining area shown is the remaining forestal area for the country in display, meaning that if the deforestation rhythm keeps that way, Suriname will be completely deforestated in 1.5 years. 3. The GLAD alerts dataset, which the bot takes as data source, don't make any distinction between forestal area of any type, they just track deforestation based on changes comparison in satellite imagery, Landsat 7 & 8 being the main imagery sources. 4. Then again, GLAD alerts does not provide any info on the causes of the deforestation. I think globalforestswatch.org have an API based around industrial logging that let's you track industry-based deforestation only, but it's based on in-site reports and it's not as reliable as satellite-based GLAD alerts.
    - abpavel 6y ago
      Thanks for the reply. For point 2, the source states that from 2001 to 2019 1.3% of the tree cover was lost [0]. However in the twitter feed the loss is 2% per 10 days. [0] https://gfw.global/3gm1aRN https://gfw.global/3gm1aRN
      - imnotanerd 6y ago
        You are looking at the forest cover lost in Suriname only for that time period. The bot gathers global forest cover loss for a given day and compares the aggregated of every country against the selected country from the list. I thought I made it already pretty clear on the way the statuses are written, how would you write them to make it clear that its global area compared to that country rather as that country area only?
quenix 6y ago
This seems cool, but somewhat unconstructive. Feels like receiving daily tweets about the continuous destruction of wildlife (but being able to do nothing about it personally) is a very negative overall experience, and doesn’t seem healthy. It is in effect receiving daily bad news which you can do nothing about.
- Taek 6y ago
  No you can do things about it. If more people are talking about deforestation, political parties realize it becomes a way to swing voters.
  - abpavel 6y ago
    To talk about the topic, one must first understand about a topic. Citing a twitter number does not help when it's not even known how much was destroyed by winds or fire and how much is reforested. Remember, wood is a renewable source, and chopping down trees is part of healthy forest management.
    - imnotanerd 6y ago
      > it's not even known how much was destroyed by winds or fire and how much is reforested I know that's an important design flaw: it can't differentiate between man-made deforestation and natural forest change. I'd like for this bot to not exist until there is a way to reliably measure the former. But if I waited for that to happen probably I'd never have the chance to understand myself and make others see the problem being discussed here. Moreover, reforestation is not taken in consideration because there is no evidence that reforestation undoes the damage causes by deforestation. Yes, the tree-covered area is there again, but is the biomass, diversity and ecosystems that the disappeared trees fostered there again? That's the real issue here. Let alone the fact that in most reforestation it takes decades on average for the trees to simply grow. Wood is a renewable resource as long as the rhythm at which we exploit it it's sustainable.
- gonzo41 6y ago
  It's more negative for the flora and fauna I would think.
eucryphia 6y ago
So if you found a security bug in some networks software you'd just make it public? zero validation? just Tweet it out publicly as soon as you found it? What due diligence did you do on globalforestwatch.org?
- imnotanerd 6y ago
  I emailed them about each and every flaw of their API that I detailed, and some other details I didn't reveal about it. Their developers and public relationships people are good professionals but I can't stop feeling they are overworked. Moreover, I detailed those flaws not as empty criticism but rather as instances of software being problematic not because of undocumented behavior but actually because of "undocumented design".
Reimersholme 6y ago
That it doesn’t keep track of regrowth makes the data point sound very negative, when it’s far from clear how it should be interpreted. What is it I’m supposed to take away from the tweet? “Taken all together, the greening of the planet over the last two decades represents an increase in leaf area on plants and trees equivalent to the area covered by all the Amazon rainforests. There are now more than two million square miles of extra green leaf area per year, compared to the early 2000s – a 5% increase.” https://www.nasa.gov/feature/ames/human-activity-in-china-and-india-dominates-the-greening-of-earth-nasa-study-shows https://www.nasa.gov/feature/ames/human-activity-in-china-an...
bosswipe 6y ago
Why does it only tweet about Suriname?
- imnotanerd 6y ago
  It will make map comparisons with Suriname until this country is deforestated, then it will move to the next country in the list.
zby 6y ago
How can it be real-time?
- imnotanerd 6y ago
  A little bit of marketing wording I must confess, but it is as real time as you can get with this kind of datasets. The API that I used for this project along the other APIs by GFW power an app called Forest Watcher that it's pretty real-time.
seapunk 6y ago
Hi Daniel, it's really nice project. Any plan to build a website for more visualization in the future or maybe a newsletter to receive the updates?
- imnotanerd 6y ago
  Kind of. Not exactly a project inside this project, but maybe a separate visualization tool using the GLAD alerts API to allow people to gather the same information the bot does, but in an easy to use website instead of having to use the package itself inside their machines. Unfortunately I'm not much of a designer so this could take a long time. The newsletter idea sounds cool. Maybe even send detailed reports? There is more data the bot knows on each request but gets ignored for simplicity of the statuses. However, as of now I'm more concerned about the continuity of the bot, as mLab (where the bot database is) will be terminating their service by November. Currently looking for a free or affordable place to deploy this app.
  - seapunk 6y ago
    > The newsletter idea sounds cool. Maybe even send detailed reports? There is more data the bot knows on each request but gets ignored for simplicity of the statuses. Indeed a detailed report would be nice to receive. About mLab, I think you could give a shot to the MongoDB Atlas Free Tier (M0).
blondie9x 6y ago
What’s the Suriname for?
- imnotanerd 6y ago
  Just happened to be the first country on the list.
dj_mc_merlin 6y ago
"Here we analyse 35 years’ worth of satellite data and provide a comprehensive record of global land-change dynamics during the period 1982–2016. We show that—contrary to the prevailing view that forest area has declined globally—tree cover has increased by 2.24 million km2 (+7.1% relative to the 1982 level). This overall net gain is the result of a net loss in the tropics being outweighed by a net gain in the extratropics." - Song, X., Hansen, M.C., Stehman, S.V. et al. Global land change from 1982 to 2016. Nature 560, 639–643 (2018). https://doi.org/10.1038/s41586-018-0411-9 https://doi.org/10.1038/s41586-018-0411-9 This does not take into account loss of biodiversity or effects on climate, but the viewpoint that we are "running out of trees" is a common one. There are actually more trees on earth than stars in the Milky Way by an order of magnitude. I would love to have my opinion changed on this however, so if further data contradicts the above study (not that one single study ever proves anything), please make me aware of it.
- imnotanerd 6y ago
  When I started this project I also had that feeling that we were running out of trees. But my motivation never was to alarm people that we were running out of them, my intention was to make actual deforestation data accessible and approachable to the public so they could decide how big of a problem this was. I have myself learned that I should probably worry less about it than I used to thanks to this bot, and that's fine. I also think you are not seeing the big picture here. 30 years is pretty much just the minimum for the average tree to grow to a size where it can foster other biomass. On the bigger picture: "Forests cover 31 percent of the world’s land surface, just over 4 billion hectares. […] This is down from the pre-industrial area of 5.9 billion hectares." -Earth Policy Institute. http://www.earth-policy.org/indicators/C56/ http://www.earth-policy.org/indicators/C56/ 1.9 billion hectares down in about a century and a half. My viewpoint now is that there is a deforestation problem in the sense that we destroy forest dependant biomass faster than what it takes for it to recover, but we are still in the time window where the already made harm and the future impact of our deforestation can be reversed, from the viewpoint of Hansen et al we seem to have reversed the negative tendency in tree cover.
  - dj_mc_merlin 6y ago
effnorwood 6y ago
Now do tree planting - aka reforestation

101 points by imnotanerd 6y ago

Heads up: [1] This is a project I've been working in about a year now and is an honorary successor to a similar project a friend started. You can read in detail how it works at https://imnotanerd.herokuapp.com/3 https://imnotanerd.herokuapp.com/3. Basically it takes data from the API at globalforestwatch.org that uses satellite imagery to gather area of potential forest loss and paints it in a map to make it easy to imagine. There is no consideration about forest regrowth. This bot does not track forestal area changes, it tracks deforestation. [2] I'd like to get some community feedback. The bot has now been running for quite some time and it looks like now it works with consistency (The GFW API data is not reliable when recently published). Mainly I'd like to see how this bot can be improved regarding the data treatment and the display of it, as I still think the maps, despite being accurate, are a little "underperformant" and could be improved. [3] Yes. I posted this twice, I forgot to put "Show HN" in the title the first time and I realised too late. Now I can't edit nor remove the previous submission.

abpavel 6y ago
I see a number of issues: 1. no deforestation before July 9th. 2. In 10 days between Aug4 and Aug14 2% of the area was deforested, signifying all forrest will be gone in about 1.5 years. However forrest area in 2015 was almost the same size as reported total. Can't be right. Are you mixing ha and km2 by any chance? 3. No tracking between what is part of Forest Management Act, what are is part of designated production forest, and what is illegal deforestation. 4. No data on loss due to natural causes such as wind, fire, and no data on reforestation numbers.
- dna_polymerase 6y ago
  Point 4 might be irrelevant to the cause and more importantly part of a negative (hard to track) feedback loop. A patch of forest destroyed by human activity might be more susceptible to stronger winds or extreme weather in general.
- imnotanerd 6y ago
  1. That's the date the bot started tracking. Actually, deforestation data starts in 2015 but I thought it'd be dishonest of the bot to count in deforestation that happened before the bot release. First drafts of the bot worked this way but that behavior was abandoned. 2. What do you mean by 2%? I guarantee I'm not mixing units, the remaining area shown is the remaining forestal area for the country in display, meaning that if the deforestation rhythm keeps that way, Suriname will be completely deforestated in 1.5 years. 3. The GLAD alerts dataset, which the bot takes as data source, don't make any distinction between forestal area of any type, they just track deforestation based on changes comparison in satellite imagery, Landsat 7 & 8 being the main imagery sources. 4. Then again, GLAD alerts does not provide any info on the causes of the deforestation. I think globalforestswatch.org have an API based around industrial logging that let's you track industry-based deforestation only, but it's based on in-site reports and it's not as reliable as satellite-based GLAD alerts.
  - abpavel 6y ago
    Thanks for the reply. For point 2, the source states that from 2001 to 2019 1.3% of the tree cover was lost [0]. However in the twitter feed the loss is 2% per 10 days. [0] https://gfw.global/3gm1aRN https://gfw.global/3gm1aRN
    - imnotanerd 6y ago
      You are looking at the forest cover lost in Suriname only for that time period. The bot gathers global forest cover loss for a given day and compares the aggregated of every country against the selected country from the list. I thought I made it already pretty clear on the way the statuses are written, how would you write them to make it clear that its global area compared to that country rather as that country area only?

quenix 6y ago

This seems cool, but somewhat unconstructive. Feels like receiving daily tweets about the continuous destruction of wildlife (but being able to do nothing about it personally) is a very negative overall experience, and doesn’t seem healthy. It is in effect receiving daily bad news which you can do nothing about.

Taek 6y ago
No you can do things about it. If more people are talking about deforestation, political parties realize it becomes a way to swing voters.
- abpavel 6y ago
  To talk about the topic, one must first understand about a topic. Citing a twitter number does not help when it's not even known how much was destroyed by winds or fire and how much is reforested. Remember, wood is a renewable source, and chopping down trees is part of healthy forest management.
  - imnotanerd 6y ago
    > it's not even known how much was destroyed by winds or fire and how much is reforested I know that's an important design flaw: it can't differentiate between man-made deforestation and natural forest change. I'd like for this bot to not exist until there is a way to reliably measure the former. But if I waited for that to happen probably I'd never have the chance to understand myself and make others see the problem being discussed here. Moreover, reforestation is not taken in consideration because there is no evidence that reforestation undoes the damage causes by deforestation. Yes, the tree-covered area is there again, but is the biomass, diversity and ecosystems that the disappeared trees fostered there again? That's the real issue here. Let alone the fact that in most reforestation it takes decades on average for the trees to simply grow. Wood is a renewable resource as long as the rhythm at which we exploit it it's sustainable.
gonzo41 6y ago
It's more negative for the flora and fauna I would think.

eucryphia 6y ago

So if you found a security bug in some networks software you'd just make it public? zero validation? just Tweet it out publicly as soon as you found it? What due diligence did you do on globalforestwatch.org?

imnotanerd 6y ago
I emailed them about each and every flaw of their API that I detailed, and some other details I didn't reveal about it. Their developers and public relationships people are good professionals but I can't stop feeling they are overworked. Moreover, I detailed those flaws not as empty criticism but rather as instances of software being problematic not because of undocumented behavior but actually because of "undocumented design".

Reimersholme 6y ago

That it doesn’t keep track of regrowth makes the data point sound very negative, when it’s far from clear how it should be interpreted. What is it I’m supposed to take away from the tweet? “Taken all together, the greening of the planet over the last two decades represents an increase in leaf area on plants and trees equivalent to the area covered by all the Amazon rainforests. There are now more than two million square miles of extra green leaf area per year, compared to the early 2000s – a 5% increase.” https://www.nasa.gov/feature/ames/human-activity-in-china-and-india-dominates-the-greening-of-earth-nasa-study-shows https://www.nasa.gov/feature/ames/human-activity-in-china-an...

bosswipe 6y ago

Why does it only tweet about Suriname?

imnotanerd 6y ago
It will make map comparisons with Suriname until this country is deforestated, then it will move to the next country in the list.

zby 6y ago

How can it be real-time?

imnotanerd 6y ago
A little bit of marketing wording I must confess, but it is as real time as you can get with this kind of datasets. The API that I used for this project along the other APIs by GFW power an app called Forest Watcher that it's pretty real-time.

seapunk 6y ago

Hi Daniel, it's really nice project. Any plan to build a website for more visualization in the future or maybe a newsletter to receive the updates?

imnotanerd 6y ago
Kind of. Not exactly a project inside this project, but maybe a separate visualization tool using the GLAD alerts API to allow people to gather the same information the bot does, but in an easy to use website instead of having to use the package itself inside their machines. Unfortunately I'm not much of a designer so this could take a long time. The newsletter idea sounds cool. Maybe even send detailed reports? There is more data the bot knows on each request but gets ignored for simplicity of the statuses. However, as of now I'm more concerned about the continuity of the bot, as mLab (where the bot database is) will be terminating their service by November. Currently looking for a free or affordable place to deploy this app.
- seapunk 6y ago
  > The newsletter idea sounds cool. Maybe even send detailed reports? There is more data the bot knows on each request but gets ignored for simplicity of the statuses. Indeed a detailed report would be nice to receive. About mLab, I think you could give a shot to the MongoDB Atlas Free Tier (M0).

blondie9x 6y ago

What’s the Suriname for?

imnotanerd 6y ago
Just happened to be the first country on the list.

dj_mc_merlin 6y ago

"Here we analyse 35 years’ worth of satellite data and provide a comprehensive record of global land-change dynamics during the period 1982–2016. We show that—contrary to the prevailing view that forest area has declined globally—tree cover has increased by 2.24 million km2 (+7.1% relative to the 1982 level). This overall net gain is the result of a net loss in the tropics being outweighed by a net gain in the extratropics." - Song, X., Hansen, M.C., Stehman, S.V. et al. Global land change from 1982 to 2016. Nature 560, 639–643 (2018). https://doi.org/10.1038/s41586-018-0411-9 https://doi.org/10.1038/s41586-018-0411-9 This does not take into account loss of biodiversity or effects on climate, but the viewpoint that we are "running out of trees" is a common one. There are actually more trees on earth than stars in the Milky Way by an order of magnitude. I would love to have my opinion changed on this however, so if further data contradicts the above study (not that one single study ever proves anything), please make me aware of it.

imnotanerd 6y ago
When I started this project I also had that feeling that we were running out of trees. But my motivation never was to alarm people that we were running out of them, my intention was to make actual deforestation data accessible and approachable to the public so they could decide how big of a problem this was. I have myself learned that I should probably worry less about it than I used to thanks to this bot, and that's fine. I also think you are not seeing the big picture here. 30 years is pretty much just the minimum for the average tree to grow to a size where it can foster other biomass. On the bigger picture: "Forests cover 31 percent of the world’s land surface, just over 4 billion hectares. […] This is down from the pre-industrial area of 5.9 billion hectares." -Earth Policy Institute. http://www.earth-policy.org/indicators/C56/ http://www.earth-policy.org/indicators/C56/ 1.9 billion hectares down in about a century and a half. My viewpoint now is that there is a deforestation problem in the sense that we destroy forest dependant biomass faster than what it takes for it to recover, but we are still in the time window where the already made harm and the future impact of our deforestation can be reversed, from the viewpoint of Hansen et al we seem to have reversed the negative tendency in tree cover.
- dj_mc_merlin 6y ago

effnorwood 6y ago

Now do tree planting - aka reforestation