// author : thembululquaUwU
// 3.9.2024
#include <bits/stdc++.h>
#define pb push_back
#define fi first
#define se second
#define endl '\n'
using namespace std;
using ll = long long;
using ii = pair <int, int>;
using vi = vector <int>;
const int N = 1e5 + 5;
const int mod = 1e9 + 7;
void maxl(auto &a, auto b) {a = max(a, b);}
void minl(auto &a, auto b) {a = min(a, b);}
int n, m, q, l[N], r[N];
int a[N], b[N], s[N], t[N], x[N], lca[N];
ll y[N];
ii p[N];
vi adj[N], query[N];
int t_in[N], t_out[N], cnt = 0;
void dfs(int u = 1, int p = 0){
t_in[u] = ++ cnt;
for (int v : adj[u]) if (v ^ p) dfs(v, u);
t_out[u] = cnt;
}
template <class T>
struct RMQ { // 0-based
vector<vector<T>> rmq;
T kInf = numeric_limits<T>::max();
void build(const vector<T>& V) {
int n = V.size(), on = 1, dep = 1;
while (on < n) on *= 2, ++dep;
rmq.assign(dep, V);
for (int i = 0; i < dep - 1; ++i)
for (int j = 0; j < n; ++j) {
rmq[i + 1][j] = min(rmq[i][j], rmq[i][min(n - 1, j + (1 << i))]);
}
}
T query(int a, int b) { // [a, b)
if (b <= a) return kInf;
int dep = 31 - __builtin_clz(b - a); // log(b - a)
return min(rmq[dep][a], rmq[dep][b - (1 << dep)]);
}
};
struct LCA { // 0-based
vector<int> enter, depth, exxit;
vector<vector<int>> G;
vector<pair<int, int>> linear;
RMQ<pair<int, int>> rmq;
int timer = 0;
LCA() {}
LCA(int n) : enter(n, -1), exxit(n, -1), depth(n), G(n), linear(2 * n) {}
void dfs(int node, int dep) {
linear[timer] = {dep, node};
enter[node] = timer++;
depth[node] = dep;
for (auto vec : G[node])
if (enter[vec] == -1) {
dfs(vec, dep + 1);
linear[timer++] = {dep, node};
}
exxit[node] = timer;
}
void add_edge(int a, int b) {
G[a].push_back(b);
G[b].push_back(a);
}
void build(int root) {
dfs(root, 0);
rmq.build(linear);
}
int query(int a, int b) {
a = enter[a], b = enter[b];
return rmq.query(min(a, b), max(a, b) + 1).second;
}
int dist(int a, int b) {
return depth[a] + depth[b] - 2 * depth[query(a, b)];
}
};
pair <ll, int> ans[N], bit[N];
void update(int l, int r, int x){
for (; l <= n; l += l & -l) bit[l].fi += x, bit[l].se ++;
for (r = r + 1; r <= n; r += r & -r) bit[r].fi -= x, bit[r].se --;
}
pair <ll, int> gett(int k){
pair <ll, int> ans = {0, 0};
for (; k; k -= k & -k) ans.fi += bit[k].fi, ans.se += bit[k].se;
return ans;
}
pair <ll, int> qry(int id){
pair <ll, int> vs = gett(t_in[s[id]]);
pair <ll, int> vt = gett(t_in[t[id]]);
pair <ll, int> vlca = gett(t_in[lca[id]]);
return pair <ll, int> (vs.fi + vt.fi - vlca.fi * 2, vs.se + vt.se - vlca.se * 2);
}
void process(){
for (int i = 1; i <= n; ++ i) bit[i] = {0, 0};
for (int id = 0; id <= m; ++ id){
if (id){
int i = p[id].se, u = b[i], w = p[id].fi;
update(t_in[u], t_out[u], w);
}
for (int t : query[id]){
pair <ll, int> cur = qry(t);
if (cur.fi <= y[t]) ans[t] = cur, l[t] = id + 1;
else r[t] = id - 1;
}
}
}
void solve(){
cin >> n >> m >> q; LCA T(n + 1);
for (int i = 1; i < n; ++ i){
cin >> a[i] >> b[i];
adj[a[i]].pb(b[i]); adj[b[i]].pb(a[i]);
T.add_edge(a[i], b[i]);
}
dfs(); T.build(1);
for (int i = 1; i < n; ++ i){
if (t_in[a[i]] > t_in[b[i]]) swap(a[i], b[i]);
}
for (int i = 1; i <= m; ++ i){
cin >> p[i].se >> p[i].fi;
}
sort(p + 1, p + m + 1);
for (int i = 1; i <= q; ++ i){
cin >> s[i] >> t[i] >> x[i] >> y[i];
l[i] = 0, r[i] = m; lca[i] = T.query(s[i], t[i]);
}
while (true){
bool ck = false;
for (int i = 1; i <= q; ++ i){
if (l[i] > r[i]) continue;
int mid = l[i] + r[i] >> 1;
query[mid].pb(i); ck = true;
}
if (!ck) {
for (int i = 1; i <= q; ++ i){
pair <ll, int> cur = qry(i);
ans[i].se = cur.se - ans[i].se;
}
break;
}
process();
for (int i = 1; i <= m; ++ i) query[i].clear();
}
for (int i = 1; i <= q; ++ i){
if (ans[i].se <= x[i]) cout << x[i] - ans[i].se << endl;
else cout << -1 << endl;
}
}
int main(){
if (fopen("pqh.inp", "r")){
freopen("pqh.inp", "r", stdin);
freopen("pqh.out", "w", stdout);
}
ios_base::sync_with_stdio(0);
cin.tie(0); cout.tie(0);
int t = 1; // cin >> t;
while (t --) solve();
return 0;
}
/*
10 7 9
1 8
6 3
5 9
7 9
3 1
3 4
10 1
2 6
5 6
9 4
7 4
7 4
2 4
7 4
7 4
1 4
8 6 5 3
3 9 8 0
4 7 6 15
7 4 9 3
6 4 8 0
9 10 5 16
5 3 2 4
2 8 4 3
6 1 3 3
*/
